Dr.

Paul Helliwell
Advrs e r o n Warc r Dove
lo pmont
Highfreh house prnno 070345322g
2 ManchesterRoacj Tclor
4j7575
Natley Abbey
Soutbempton
SO3 SDB
f-tr
G!

ffiw

Flood Design Manual

ror

Java and Sumatra

 GOVERNUENT OF TEE REPUBLIC OF INDONESIA
ffiw
l{IlrISTRY OF PIIBLIC }IORXS
DIRECFORATE GENERAL OF
RESOURCES DEVELOPHEVT
'ATER

FLOOD DESIGN HAIiIUAL

-- FOR
JAVA AND SI]HATRA

SEPTEHBER 1983

INSTITI'TE OF EYDROIOGY
DPHA
IJALLIIreFORD
BANDUNG
oxoN
INDONES IA
I].K.
 CONTENTS
PAGES
LIST OF TABLES (i)
LIST OF FIGURES (ii)
ACKNOWLEDGEMENTS
(iv)
NOMENCLATI]RE
(vi)

PREFACE
( ix)
CHAPTER I INTRODUCTION TO FLOOD ESTIMATION 1

I. I Flood stafistlcs. return period and probahil i ty 1

I. 2 Flood estlmatlon meLhods 5
t. 3 Cholce of deslgn flood a

CHAPTER zl HOW TO USE THIS REPORT 10

CHAPTER ESTIMATION OF MEAN ANNUAL FLOOD FROM THE ANMJAL I'IAXTML'}f.

S ER IES 1?

?t Introduction 12
3.2 Descrlption of method 13
3.3 Accuracy of results IA
3.4 Exanple of application 15

CHAPTER 4. ESTIMATION OF MEAN ANNUAL FLOOD FROM THF PI]f\KS OvFR A

THRESHOLD SERIES L7

4.1 InEroduction
,^
.+. z Description of method
L7
r7
4.3 Accuraey of results 19
4.4 Exanple application 20
4.5 Using lncompJ ete years of da ta 20

CHAPTER 5: ESTIMATION OF MEAN ANNLIAI- FI,OOD FROI.l (--A lcHlfi:\r't'

C HARACTER IST ICS 23
q1
InEroduction 23
5.2 Description of nethod 23
5.3 Accuracy of method 25
5.4 Exanple of application 26

CHAPTER 6: FLOoD FREQUENCY GROWTH FACTORS FOR JA\.A ANr) SL,'I,IATRA 2a

o.l IntrodrrcEion 28
6.2 PloEtlng the recorded da [a 28
6.3 Exanple of applicarion 30

CHAPTER 7: PLOTTING A FLOOD FREQUENCY GROWTH CURVE 33

11
Int roduc t 1o n 33
7.2 Descrlption of method 33
7.3 Accuracy of nethod 35
7.4 Exanple of applicarion 36
 PAGES
CHAPTER 8: IMPROVING THE FLOOD ESTIMATE USING LOCAL DATA 37
A1 In t roduc t 1o n
8.2 Uslng different rnethods at the same site 37
8.3 Uslng the observed MAF from an adjacent long-te.rrn 38,
sta Elon +o
8.4 uslng flood records from elsewhere in the catr:hmenr 48
8.5 Uslng flood records from adjoining ca[chments 50
8.6 Uslng staff gauge data 52
8.7 Use of flood narks CD
8.8 ExEenslon of a flood frequency curve 56
C}IAPTER 9: COMPARISON WITH OTHER FLOOD ESTIMATION METHODS USED
IN
INDONESIA 60
9. i Introduction 60
9.2 Ratlonal Eethod (]l
9.3 Weduwerm nethod ,t2
9.4 Melchlor nethod 62
9 .5 llasper method
peterson method 63
9.6 64
9.7 Comparison of results 64
ANNEX A: STATION SELECTION
A.I Int roduc tion 73
A,2 Offlce procedure 73
A.3 Station vislE procedure 73
4.4 IniElal Screenlng of Flood Study SEarions 74
75
ANNEX B: RATINC EQUATION DEVETOPMENT
81
8.1 Introductlon s1
8.2 Logartthurlc rating curve extrapolation
B.3 81
Slope - area racing extrapolation
8.4 Reconmended inprovernents in hydrology 87
92
ANNEX C: THE MAF ESTIMATION EQUATION q?

c.l
(-) Introduction q?
Regresslon equation for the mean annual floorl 9.1

ANNEX D: CATCHI"IENT C}IARACTE R IST ICS

101
D. t Introduction
101
D.2 CatchmenE area (AREA) 101
D.3 Main stream length (MSL)
i.1
D.4 Annual average rainfal 1 (AAR) .l
o1
D.5 Mean annual maxlmun catchment I day rainfal.l (APBAR) 105
D.6 Rlver slope (SIMS)
108
D.7 Rlver sLope (S1085) 108
D.8 Rlver slope (S085) 108
D.9 Lake lndex (LAKE) 109
D. 10 Flood plain index (S010) 109
D.l1 Geology lndex (GEOL) 110
D. 12 Solt lndex (SOIL) 110
 PAGES
n 1? Forest index (FOREST) 111
D. l4 Paddy lndex (PADDY) 111
n 1< Plantation index (PLTN) IL2
D. I6 Swamp lndex (SWAMP) 112
n l7 Catchnent shape index (SHAPE) rL2
ANNEX E: THE PEAKS OVER A THRESHOLD MODEL 11 l

InEroduction 114
t) Theory alt
tr? Incomplete years of data \77
Hiscoric floods ).L7
A variat. lon of the POT method 118

ANNEX F: GROWTH CURVE ANALYSIS 119

Int roduc t lon 119

FO
Poollng of growth curves 'J,20
F? Sub grouping of growEh curves 722
_r.q Signiflcance Eests 723
t.) Discusslon of results' .
131
Reconroended growth curves 133

REFERENCES 136

DATA APPENDIX (Under separate cover)

 LIST OF TABLES

6.1 Ranked floods and Ehe corresponciing vaJues of y For the'

Citarun at Nanjung

7.1 Table of growth factors

9. t Comparison of flood estimation meEhods for Java

9.2 Comparison of flood estimation methods for Suuratra

9,3 RMS error of MAF for the flood estimation methods

9,4 RMS error as a percentage for Ehe l0 year flood estimates

A,l Catchment characLeristics of stations used in Lhe analysis -

Java

A,2 Catchment characteristics of stations trsed in the analvsis

Suma t ra

A.3 CaEchment characteristics of some stations not used in the

analysis

B.l Calculating l//f from discharge measriremr','i:-

C. I Correlation matrix

C ,2 Regressions to est innte t"tAF

D,1 Summary of catchment characteristics

F.1 Kolmogorov-Snr-irnov tesE resul ts

 LIST OF FIGIIRES

l.l Island of Java showing catchnent boundarles and mean annrral

maxirnum l day rainfall (PBAR)

1.2 Island of Sunatra showing catchment boundaries and mean

annual m'xlmum 1 day rainfall (PBAR)

1.3 A flood frequency curve

1.4 Identificatlon of floods and thelr assoclated return periods

1.5(a) The p.ob"Uftfty density functlon of annual maximum floods

1.5(b) The dlstribuElon funeElon of annual naximum Floods

I.6 Use of broken records: maximlsing use of data rhrorreh,,q('^f

POT meEhod

2.1 Guide to the use of this manual

4.I Test for independence of flood peaks

5.1 Plot of APBAR against AREA for regression analysis

catchments

6.1 Flood frequency curve tor the Citarum aL Nan.iung

1,l Average flood f reqtrenc,v growth curve ior Java ilrrd Srrmatra

8.1 Extrapolated f lood Irequency curve tor the Ci tit rum .] r

Na n jung

9.1 Maximum recorded f loods in Java and Sumatrii .rr',r r',grr-:-si,rrr

' equation estlnates

B.1 Shlft ln station datum: Krtreng Aceh at Kampung Darang

8.2, B.3 Development of a stage discharge curve: Krueng Aceh at

8.4 Graph of -'
I
against Log I 6R for Ba tang Ha rl a t Sunga j Da reh
/f

C.l 0bserved MAF plotted against MAF estlnated by regression

equatlon

C.2 Distributlon of factorial standard errors of the regression

equaEion

F.l Reglonalizatlon of growth curve

F.2 Ef fect of "MR" on grovtth curve shape

F.3 EffecE of "APBAR" on growth curve shape

F.4 Effect of "AREA" on growlh eurve shape

ACKNOWLEDGEMENTS

During the course of this pro jec t. Dr C S Green. lfr llt B Boornra n
and Mr F A-K Farquharson of the Tnstitute of HydroJogy received lrcip
and encouragenent frour a large number of organisatlons and
lndivlduals. Wlthout their cooperation this pro,iect could never
have succeeded and rde express our gratiEude to all of the following.
Ir l4ardjono Notodlhardjo Assistant Dlrector General for
River Development (DGWRD)
Ir Rachmat Tlrtotjondro Dlrector, DPMA

Drs Tata Sukarta Head, Foreign Ald Administrati()n

(DGWRD)

Ir AzIz Bocking Assistant, Foreign Aid

Adml nis Eratlon (DGWRD)

Ir Muhadi Dipl H Chief. Hydrology Division

DPMA

Mr Soenarso Soehardi Chief, Foreign Aid

Adninistration DPHA
Ir Soeharto Chief , Hydrometry Scction, DPITA

Ir Joesron Loebis Chief . Hyd l Lr 111 gt Sec I ion. llPl'l.\

Ir Ferry Putuhena MSc Head. Water Resources Data Unit

(P3sA)

Surin Sangsni t Team Leader. UNDP/WMO

River Forecasti.ng Project. DPMA.

Mr R Kingston Resident Engineer, Si r Al exander

Cibb and Partners. DPMA Bandung

Mr A Bonnier SCFT Jnternational, T,-,rm l.,rtl, p

Pro-jek Hidrologi . IIPMA

Mr A Muzet SCET Intel'ir,rl irt631, l ntj rrpr.r

Pro -jek Hid ro 1 o gi . DPMA

Mr J L Irish Regional Adviser in Hydrologr.'

UNESCO

Mr F J Kaul Binnie and Partners, London

Mr B N Gunter SMEC. Tean Leader, I^/est Java

Design Unit Assistance Project
Mr B C Ferguson Internatlonal Engineerl ng Co.
Inc., Sederhana Irrigatlon and
La nd Develo prnent Pro -jec t I f

MrEWJCrawley First Secretary (Aid)

British Embassy Jakarta
MrI Thow First Secrer,rry (Ajd)
British Embassy Jakarta
Mr D CoEtrell Second ry (Ai d )
Sec re ra
British Embassy Jaka rta

Mr J Wilkins Secretary (Aid)

Brl tish Embassy Jakarta

Da rsono Urna ryadi B.E Staff DPU Prop. DTI Jawa Barat

Endang Sukardja Assistant. Java Floods Study

Pro.-iec E, DPMA

Dr A N l4andeville Chlef, 0I)A Sectjon, Tnstitute of

Hyd rology, Wa J lingford'

Miss S Smethurst British Council, Jakarta

Dr D A Jones Hydrologlst, Institute of
Hydrology, Wallingford
Mr S B Jones Hydrologlst, InsEitrrte of
Hydrology, Wallingford.
Miss B J Ridler Assistant, Institute of
Hydrology, Wal 1i ngford

Mrs V Bronsdon Asslstant. Instltrrte of

Hydro l.gy. ',Ja I 1i ngf t-,rd.
 NOI.{ENCLATI]RE

A = river."etoss section area (r2)

a = rating equatlon paraneter
AAR = annual average ralnfall (mm)

APBAR = mean annual naximum catchnent 1 day rainfalr (mm)

AREA o catchment area (tm2)

ARF = areal reduction factor
b - raElng equatlon parameter
f, = Atezy roughness coefficlenL
c - rating equatlon parameter (n)
DGWRD = Dlrectorate Generar of water Resources DeveJopment

DPMA = DirekEorat penyelidikan Masalah Air

(DirecEorate of ltrydraulie Englneering)
DPU = Departemen pekerjaan Umum
(DepartmenC of publlc I"/orks)

EVI = extreme value distrlbuEion - Type l

F1 = non-exceedence probability (plotring positjo.)

FOREST = foresE index

f. = Darcy - Welsbach frlctton factor
GEOL = geology index
GF rowth fac ror
I = acceleraEion due to gravity (m s-2;
h = river srage (level) (rn)

h1 = series of stage peaks (m)

ho = stage'fhreshold (m)

JOG = Joint Operations Graphics

LAKE = lake index ,

M = nunber of floods over threshold

MAF = mean annrral f lood (r3 l)
"-
MSL = IDaln strean length (krn)
}it = number of years

= Manningrs roughness coefficient

= wetted perlmerer (m)

PADDY = paddy index

PBAR = mean annual maxlmun point 1 day ralnfall (nm)

PLTN = plantat ion index

PLN = Perusahaan Umun Li sErik Negara (Electrici ty Authori ty)

POT = Peaks over a threshold

P35A = Projek Perencanaan Pembangunan :rrnl-cr-Strml-,'r Air ("'v.'.

Basin DeveloPment Planning Unl t, Di rectorate of
Planning, DGWRD)
Q = river di scharge (t3 .- I)

Qr.* = maxlroum flood peak on record (t3 t-I)

= roedlan annuat f lood 1t3 l)

Qned "-
aT = J year return period flood (o,3 ,- l)

qi = series of flood Peaks 1t3 r-I)

Qo = discharge threshola 1t3 s- I;

;: = nean of annual maximum flood series 1,n3 .-I;

S = w6rer surface slope (* *- I)

SHAPE = catchoent shape index
SIMS = river slope index (m tn-l;
SOIL = soll index
SWAI'fP = swanp index

S01O flood plain index (m tm-i;

5085 river slope index (m tm-I;
S1085 = river slope index (rn tm- I;
sd = standard devlation
V = varlable exponent of AREA

v = velocity of river flow (m s-I)

y = Gumbel reduced variate
p = mean exceedence (t3
"-I)
t = average number of exceedences Per year
 PRNFACE

Thls manual contains a rational set of flood estimarion

The Lt:cirniques
techniques applicable throughout Java and Sumetra'
arrd various wrtvs
lnclude a ,no daEat neEhod for use at ungauged si t''s
of .uslng data f rom Ehe site of interest or nearbv
flow g:rtrging
statlons.Ifusedcautlously,thetechnlquescanalsobeusedon
flow dala
other rndonesian lslands, certainly where good quajity flood
ot
are available. The report is the product of a two year stud;r
Indonesianhydrologywhicht'heauthorsbelieve'tobethemost
conprehenslve assernbly and analysis of Indonesian flood
data
from the
undertaken to daEe. The'no daEa'method has been developed
project,s large data base following a revlew of existing methods
and

drawlngheavilyonexperiencegalnedfromnumeroussiml]arstudies
performedelsewteretntheworld.AcourparisonofthismeEhodwith
previouslyusednethodsshowsirtobeasuperiortechniqueovera
wide varlety of catchment tyPes'

be
Foldout maPs showing the location of study catchments can
found for Java' as Flgure 1.1 and for Surnatra as Fi gure '2'
l

The study 135-'-!een carried---9ut in Ewo phases; Phase I

in 1981
exarnlned f looding in Java and Phase II in the fol I owi
n8 -v{rar extended
i'ir;rse T ':
the study to include sumatra' The report product. ll dtrri i';'
superseded by this report' The study was a cn'rp ; r'-: ' rf rrr''
Kingdom (l'K)
beEween the Institute of Hydrology, Wallingford' LTnired
where nany sirnilar studies have been carried out for dj I ferenr
regions

of the world and Ehe Direktorat. Penyeli.dik'an l{asalah Aj r, Banctrng'

Indonesia, custodlans of Indonesian flow data and the primarv
hydrologicalandhydraulicresearclrinSticuteoflndonesill'

The following sEa ff worked on the pro-ject in both Tnd'rnesia and

the Unlted Kingdom.

Dr CS Green IH Phase I and IT

MT FAK Fa rquha rson IH Phase T and II

Mr DB Boo rma n
ttt
ln Phase II
Ir Sunad ji Josoad lwl jono DPMA Phase I
 Drs Trenggono DPMA Phase T

Ir Moh. Arlef Ilyas DPMA Phase I I

Drs Petrus Syarinan DPMA Phase II

Both organisations provlded support staff for the durarion of Ehe

project in their respectlve instltutes.

The project Ltas jofntly funded by Ehe fndonesi;rn C,r\'(rrnmr,nI and

the Brltlsh Overseas Development Adrninistratlon.
 l. I}ITRODT'CTION TO FLOOD ESTIHATION

ll Flood statistlcs IS_!lfl_-p_gfl_9!_a nd probabi I i t y

For any f lood._esrtnation pr-ob1em lt is necessary to sPeclf y the

return period, or probabiltty, of the desired flood. This will vary
accordlng to the nature of the project and the consequences oi the
design flood betng exceeded. i., Practlce it i s oi L.n trs. iul Lo
consErucf a curve relating the sLze of flood to its probabllity of
occurrence. SuCh a curve, call-ed a flood frequency curve' errables
flood magnltudes corresPondlng to various design criEerla to be
estinated and the implenenEation costs and lmpllcations of failure of
such criteria to be appraised. Figure 1.3 shows such a curve. The
probabllity scale glves an exceedence Probability (ie the probability
of a flood 1evel belng exceeded ln any one year); the sca.le benea[h
thts shows retum psriod, or average lnEerval in years between floods
exceedlng thls leve1. Return period, T, is Ehe reclprocal of the
exceedenee probability and can give a more tangible sPpreciation of
the severity of Ehe f1ood.

If a very long record exists for a point on a river it is

posslble to consEruct a flood frequency curve from an examinaLion oF
Ehe record. Flgure 1.4(a) lllusErates one aPproach to thi s; the
record i.s divided tnto hydrological years ([o ensure indept-ndence of
flood peaks) and the biggest flood in each year i,s noted. R,v ranl<ing
the tloOds and assuming a partictrlar {-orm f'lr Litt it 'l isLr ii'rrt i'
can be assigned an exceedence probabil ity and so a i'lood f reqtterlcv
curve can be construcEed. It is interesting to note some cI thc
properties of this annual maximum flood series. It might be expe.'red
that the nean of Ehe annual maxima is exceeded by approximatelv lrirr i
of rhe floods and so have an exceedence probability oI louglrly O'5:rrrC
a reEurn period of about t\to years. However, since it is possible'ttl
have floods very uuch blgger Ehan Ehe nean and because tlrere is a
limlt to how mrch soaller fhey can be, the dlstributiorr i'1- f loods is
Skewed. In fact the-'mean of the annual naxirur i s rrsual l_r' L.ri:r'n to
have an exceedence probability of 0.43 and a return i,,'ri rd o[ ].31
years. Figure 1.5(a) shows thg probability density function and Fig
1.5(b) the dlstribution function for the annual maximum fLoods; Liris
shows the skewed nature of the distribution and inEroduces ttre concept
 A flood frequencY curve

=tt,
(t

;E
a
=c
o)
o
tr
!to
o
ll.

Exceedence ProbabilitY

Return period lyclrsi

Figure 1 '3
 ldentification of floods and their associated return periods

HYDROLOGICAL YEAR
Flow
t

rw
I I

"L

1 it it l

["]
tsJ Annual maximum floods , I

+itiittr trl

'rrr^tl
-i--l Iot t I

IUl Occurrence of annual maxima over a threshold. Intervals between years with floods

lr
-t-
lcl Occurrence of all floods over a threshold. lntervals between floods

Ftgure 1.4
of non-exceedence probability (the probabltity of a flood noE being
exceeded in any one year) whlch is frequently used in preference to
Ehe exceedence probabillty as it leads to easy estlmation of the risk
of failure of anv scheme (see Section l'3 )'

When consldertng only the biggest flood in each year' the return
period is noE the average interval between floods of a given rnagnitude
bug the average interval between years containlng floods of thaI size
or greater. In Flgure 1.4(a) it can be seen thaE the largest flood in
some years is exceeded by Ehe second or third largest flood in others;
a second approach to flood frequency analysis that avoids this probl em
conslders all the floods-over a cerEaln ltmtttng r;ize, noL -just th,
blggest 1n eaeh year. Such a flood sequence is cel lt'd .r partial
duraEion, or peaks over a threshold, serles. Tn Figure 1.4(b) al1
years contalning floods over a cerEain stze have been urarked and i n
Flgure I.4(c) all floods over that same size are indlcated'
Immediately it can be seen EhaE the average interval beEween floods is
less than Ehe average interval beEween years wlth floods. The reLLrrn
period of the mean annual flood frorn the annual maxLmum series is
about. half a lear greaEer than f r<im-'the parEiaJ. durat ion series , but
Ehe difference decreases aS return period lncreases since as the
Ehreshold is ralsed the Ewo serles become identical . A1 thotrgh the
partial duraEion serles approach is the more fundamental one, the
smal1 difference at--large return- periods and the easy application of
Ehe annual maximum method.makes it the more popular choice [or flood
frequency studies.
I.2 Flood estlxoatlon methods

When a long record is available estimation of the flood oi

specif led reEurn period is a stfaightforwarC task as outl i ncd abo'''e.
However, tt is usually the case that only a limi-rcd period of tl;tt,r i.'
available and it is eicher impossible to construcI a flood frequency
curve or Co extend iE to the requtred return period. 'l'lre ro itre tltrcc
broad classes of method tha! can be used in such ci.rcumst;l ncL's;
sEatistlcal methods, rainfall-runoff methods and internaIional
enpirlcal methods.

The lnternaElonal emplrical nethods are usually simpl.e formulae

relaEing flood nagnitudes to physiographical propetSil's rr'Ihe
drainage area. They are offen based on a straightforward conceptuali-
sation of the rainfall-runoff process and calibrated on a speciflc
data set. The growth of flood nagnltudes with return fcri^d is
achieved through using ralnfall frequeney relationshlps which are
generally more wldely avallable_ Ehan flood frequency crrrves. Because
of the nethods' generallty and ln the absence of anything betEer they
have been adopted for use all over the world and several variations
are currently used ln Indonesia. Examples include the Ratlonal method
and Ehe Creager and Franco-Rodier equalions.

Ralnfall-runoff methods also require rainfall frequency

infornation although often over a varlety of durations. The rainfall
lnput is routed Ehrough a rainfall-runoff model to give the design
flood. Ttre trethods have the advanEage of givi,ng a complete design
hydrograph but require a conslderable amount of good quality daLa for
c librllion befo19 they caq be applled to an ung4uged slte. This
requiremenE makes Ehem unsuitable at the present Ltme for use in Java
and Suoat.ra.

Statlstical rnethods are based on the reglonal generallsaEion of

statlstlcal propertles of flood distrlbutlons. Typlcally the methods
l.nvolve the estlmatio-tf-of an index*-f1ood and the scaling of this by a
factor dependent on return period co give the design flood r>r T year
f lood where T is the required return period. Thls meth.rd jras bce a
adopted for'use ln the current scudy as it makes best use of the
available data, provides for Ehe easy incorporation oi local dat.r i n
applicaEion and links in well wlth flood frequency concepts applicable
to long records.

The index flood chosen was Ehe mean annual flood (the mean of the
annual maxlmum flood serles) as this can be estimated at a large
number of sites ln Java and Sumatra frou exlsring records. As staced
earller this can be estlroated fron Ehe annual maxima for long
records. However, for short records it is better to use the partial
duratlon series (or peaks over a Lhreshold) method as this includes
lnfornation from nore floods and is Eherefore ulore accurate. Thls
nethod is also useful where a longer record contains breaks as the
start of year is not inportanE and incomplete years of data can be
included, see Figure I.6. For very short records even [his method is
not suitable. Here the no data uethod of an emplrtcal locally based
equaEion relating mean annual flood to catchment charactL'ristics is
the best. This equatlon is the only method availah.l e whtn rro rlata ,t
all exlst.

As descrlbed above, for a very long record a flood frequency

curve can be constructed from che record 1tself. However when a flood
esEinate is requlred for a return period much greater than the record
length, averaged ratlos of the magnltude of Ehe T year flood to mean
annual flood are required. These tgrowth factors' have been derived
in thls study using local daEa and depend not only on return period
but on drainage area as we1l.

1.3 Choice of desi.ga f lood

The decision of whaE. reEurn period is approprlate for the design

of a particular project is not solety a hydrologlcal problem. The
engineer is constrained by economic, political and envlronmenEal
factors ln his design and so cannot improve the safety or reliability
of the scheme without incurring cosEs elsewhere. It is however useful
to consider the probabillty, or risk, of tht destgn flood heing
exceeded during ,the expected 1if e of the pro jecr. ii tlre rl,:,sign f l,,,rd
has a return period of r years then [he risk, r, of' rlrt f lood hci ng
exceeded in tlre L year projt'ctt'cl Iife of tht' pro-jr,ct is r:iv.n bv

r - (1 -llt
T

Thus glven an expected design life of 50 years for a road bridge or

roajor irrigatlon offtake, there is a r.isk of 0.64 or 64 per cent, ot
Ehe structure experiencing Ehe 50 year flood durj_ng its riferime. The
risk of Ehe same strucEure experiencing a 1000 year flood is only
5 per cent or put anoLher way, only one such structure in 20 wor:l6 be
likely to experlence a 1000 year flood during a designed life of
50 yea rs .
Use of broken records: maximising use of data through use of POT method

HYDROLOG ICAL YEAR

___L_r

COMPLETE HYDROLOGICAL YEARS

COMPLETE YEARS FOR PO.T. METHOO

INCOMPLETE YEARS OF DATA ALSO

USABLE FOR PO.T. METHOO

Figure 1'6
 The rnethods presented in thls report are suitable for estimating
the nagniEude of floods up Eo the 500 vear event and for tentative
esEimates of fl-oods up to lO00 year return period. This is generally
adequate for the deslgn of brldges, sna11 irrlgatlon works and channel
inprovenent works and to assist in the planning of urban development
or the assessment of alternative large dam proposals. Many detailed
studles for larger projects will require a more exLrene fiood to bc
estimated; for exanple the 'probable maxlmum floodr nay be required to
design Lhe splllway of a maJor darn. The methods of estimation
presented ln this report cannot be used dlrectly in such cases.

Users of the manual w111 note that with each rnethod an estimace
of likely errors i" gi.r"r,; again it is the problem of the design
englneer to declde how best to incorporate thls uncertainty in esti-
Inatlon into hls deslgn. The aceuraey of flood estlmation <lepends
greatly on the quallEy and q.r"rrttty of availabl e da ta. Al f horr gh a
flood esEinate can be made at any siEe using the no data method, as
little as Elito yearrs data recorded at Ehe site will lead to a beEter
estioate of the design flood. At a site where data has been collectei
for several years but the rating is good only for low flows, a flood
esEimaEe will be greatly inproved by the development of a flood rating
following a perlod of frequent flow gaugtng. since rhe quaJjry of
rating equatlons ls of greaE imporE.ance in flood hydrology, rating
equations for all statlons used in this study were reviewed, and
f requently revised prlor to use. The raLlng accuracy stroul,l be
considered in the englneerts adoption of the design level.
 2. EGJ TO USE TEIS REPORT

The flood estination meEhods presented in this report use a

courmon approach; an index f lood, the tmean annual f lood' 1s estimated
and then scaled by the appropriate rgrowth factor' to give the flood
of required return perlod. The mean annual flood (MAF) at any site is
deflned as the mean of all annual mixlnum instantaneous flood peaks.
The f lood of return perlod T ye3.ps, wriEten throughout th'l s report
QT, ls the flood that on average wl1l be exceeded once 1n a period
of T years.
The flow chart-*t Figure 2.1 lllustrates whlch methods should be
used according the avallabl1ity of data.

Where a long flood record ls avallable rhe mean annual flood can
be estioaeed by the mean of the annual maxlma ln the sample of
available data as described in Chapter 3. For a long record the
sample nean should be a good estlnate of the true mean but for shorter
records such an estinate becomes less good. For this reason where
only a few yearrs data are avallable a better esLimate of Ehe },lAF is
obtained by consldering all flood peaks thaE exceed a rhreshold
level. Thls neEhod, called the peaks over a threshold (POT) method,
is detailed in Chapter 4. If no data at all are available then the
estinatlon equaEion given in Chapter 5 should be used. This equation
relates the size of the MAF to various physical and climatological
cha'racterisLlcs bf the catchnent that are indexed by parameters
obtained f ron unps. Wherever possible the l'lAF shorrld be ,:sr irurtr.d hv
more Ehan one method so, Ehat the estimates may be compa ro,-r. Chapter 8
gives various methods of using data f rorn anothcr st rtinn s! ': r;lf erl
eiEher on the same caEchmenE, or on a neighbouring onn. in r:on.i,ri.--''-
with either the esEimation equation or data from the site itself.

Havlng obtained the best possible estinate of the mean annual

flood it must be multlplled by a growth factor to give rhe flood oi
required return period. The multlplier is dependenE not only on T but
also on catchment area and is obtained frou the table of multipliers
given ln Chapter 7. In the unusual.case of a very long record being
available ttrei ttrts stage of the design procedure can be replaced by
the developnent of a flood frequency curve for the siEe of interest,
If the return perlod of Lhe required design flood is not
signlficantly greaEer than the length of the available record. The

developnenc of such a curve is described in Chapter 6

A conparlson of the 'no datarnethods of this report wlth a

nunber of alternative flood estimatlon technlques currently used in
Indonesla ls given in Chapter 9. If ls apparent that [he methods of
thls reporE glve consistently better results over a wide range of
catchment types.

In the annexes Ehat follout Ehe body of the repor[ some of rhe
methods described only brlefly in the relevant chaPters are explained
ln more deCail lncluding variations in the basic meEhoCs and
background fheory. These topics include rating curve development
(Annex B) development of the I{AF estimaEion equation (Annex D), the
POT nethod (Annex E) and growth curves (Annex F).

A data appendix (under separate cover) contains al I the basic

flood data collated during this study.
 Data availab'1e at site of te rt's
I

No flow data i n i I

at the site of -- ---l

interesE or less i

vea rs
than 1 year rs data l-3 years 4-10 years I n-?0 vca rs .r'nr 2O I

._ _)

I I

I I
I I
I
\/\ I
Estl IDA*,f,E MAF EsEimate |'tAF Estj mate MAF
from ca tc hnenE using the POT f rom i:h e annual
charac te rlst ics us ing rne thod max i series
regres slon equat lon
(ch ap te r 5) (Chapter 4) (Chapter 3)

I
I

Are longer records avai lahle

on ne i ghbour i ng ca tchmen t s ?
I

I
I

I
Yes No Plot flood I re q uency
te |,tAF
Es t lroa onp arE t imates arrr\ro

using data of MA r (Chapter 6)

from nelghbouring I I
catchments
(Chapter 8) I
I
I
I
I I

i
I
I
Y
Es tiu
t maa teE Qr f rom }{AF Exte nd to q ui red
re
OY rcowtt hF ctors retu rn perio d rrs i r'tg
((cCha
F -- r 7\ ar owth Fac fors
(Chapte r 8)
I
I
l I

If
i
Pr( Pri
PT o compa rL,
+
l:stimat e Q.r
^F{ i mat eof Qr f rom f lood trequency
curve

Flgure 2.1 Guide to the use_of Jltl_rgl'fgl

 3. ESTIUATION OIt UEAN ANNI}AL FLOOD FRo[{ ANNUAL IIAXIHIN{ SERIES

3. f Introduction

The present chapter is concerned with estimatlng the mean annual

f lood, MF, f rom Ehe annual maxinum f lood series
recorded at the si te
of interest. If Ehe slte of lnEerest is gauged and flood peaks have
been observed over a sufficiently long time, the mean of che annua]
maxima over the period of record may be used to provide an acceptabl e
estiroaEe of the MAF. It follows therefore that the MAF wl1l be
better
estltraEed where the annual maxlmum series is long and tts varlance
snall. From our experlence in Ehe uK and elsewhere' we su88e-st that
an acceptable esElurate of the MAF can be derived from a minimuur o[ 5

years' good qualitY data.

.In order Eo preserve Ehe independence of Ehe annual oaxina it is

advantageous to sEart the hydrological year in the dry season' The
chance of the same period of flooding contributlng to two successive
years is then least likely. Most caEchments in Java and sumatra
exhibit a distinct flood season between November and April' In this
study the hydrological year starts on the lst AugusE which, [or nnny
caEchuenEs sEudied, 1s the drlest month'

3.2 Descriptlon of rneEhod

The method involves abstracting the trighest F1 ood peak i n caclr

hydrological year of record. It is important that small floods from

years of incomplete record are not iricluded in the annual maximum
series and to ensure thac it nay be best to totally disregard such
years of daEa. However if a particularly ldrge ftood is noted in an
inconplete year its inclusion in the annual maxlmum series is
desirable; as a guider-. estlnate th-e I',[AF f rom conplete years of data
and Ehen include m-xima from lncomplete years greater than this and
recalculate the MAF. Broken records can be used' provided only
complet.e hydrological years are taken from within 1t'

If the annual naxinum series contains one or more extrerne floods

Ehe mean nay be too high an estinate of the MAF. The uK Flood Studies
Report (NERC , 1975) gives an approxinaEe test to determlne whether
chree tlnes Qued, the rnedian value of the series, the record
contalns an outlier. It is suggested that the sane test is trsed in
Java and Sumatra as the annual maxima series f or [he LIK a n,] the
Indonesian catchments studied exhibir a simllar variabilitv.

If the annual naximum series contalns no extreme floods. rhe MAF

is estioated as Ehe nean df the data:

1N
MAF qt
N i=1

where,

ql (i=1,2, .N) = flood peaks in the annual ElaXlmum Serles

N number of yearts dala

If the annual nreximum series conEains one or more extrene floods

Ehe MAF is estlnaced from Ehe medlan of the recorded series:

T'1AF 1.06 Qrs6

uulElplier, 1.06, in the above equaElon ls Ene avcrage raLIo

Ttre FL^
^ - -^Fi

of mean annual flood to medlan annual flood for aII ca [chments s tud i ed
in Java and Sunatra. (The uK lnulttplier ls 1.07 ).

3.3 Accuracy of resulE

The standard devlaElon (sd) ts used here Eo define the accuracy

of the estlmaEion of'Ehe MAF. There is a 68% chance that the MAF
estlnaEed 11es wlEhin one standard devlatlon of the true long berm
value.

The standard devlation of the annual maxlorrn Ll isclr.r rscs ls

calculated thus:

N ,
tn.
\Yl -
\.
d I
1/
sd (MAF ) I
i=l N-I
where

N and ql are as above and

q = nean of q1

Although Ehis estlmaEe of the standard devlaEion strictly appl ies

to.Ehe MAF estlnated fron Ehe.rnean of the series, it is suggested that
Ehe standard devlation be calculated in the same way when the MAF is
es tina ted from 1 . 06 x Qned .

3.4 Example of appllcation

For the Ci t,a rum a E Pa lumbon, 31 years of flow data are

available. This is suffic ient to provide a good esEinaLe of the i.,!AF
by taktng the nean of the annual maxinum floods.

The annual naxirna are ranked in order of descendtng magni tude

Yea r l'{ax. Flood -" Yea r Idax Flood

,o3 3
MS
_1
"-l
39/40 2733 - Qnax 77 /78 I 3 2R
78/79 23s6 29 /to r25r
68 /69 ?1L1 64 /65 ',24 I
30/3 L tgS/t 36 /tt r21l
3s /36 1994 62/63 I I58
67 /oa 1962 7 6 /7 7 ti58
66/67 r 841 70/7r 1121
80/8 I L7 23 72/73 I121
37 /38 1662 75/76 i 009
6e/70 1610 63/64 984
79/80 16 10 6s /66 984
73/74 I5 r3 JJ / J.+ 968
40/41 1439 7r172 q/'\
34 /3s l3 83 38 /39 !,;
32/33 l3 48 74175 790
3r/32 1338 - Qoed
 ThefirStstePistoteStcherecordforanyeXtremefloodwhich
could cause an overestimation of the MAF'

Qt"* = 2733 tt
"-t
Qmed = 1338 m3 s-I

' Qma*
= Z.O4
Qned

The ratlo of Qmax/Qmed is below the crltical level of 3

indicat'lngnoextrelDefloodispresent.Thearlthmetlcmeanofthe
3l year series is therefore usecl to estitrate Ehe MAF'

MAF = 1447 *3"-I

Themeasureoferrorassociatedwiththeestlmate,thestandarc]
devlation, is calculaced as 466 t3 '-I'

The accuracy of the method does not justify the

implied accuracy
of numbers quoted above and it' is therefore more reasonabl",o 1"''
1450 n"s-'
that the MAF for the cltarum at Palumbon is estimated as
wlrh a sEandard deviaEion of 470 rn3s-l'
4. ESTII{ATION OF HEAN ANNUAL FLooI) FROH PEAKS OVER A TERESEOLD
SERIES

4.r Introduc tlon

whenonly a llnlted perlod of record is available, finciing the

nean annual flood fron the annual nraxlmum serles 1s inappropriate as
the series will be too s-hort to estlmate the mean reliably. In such
clrcumstances the serie.s of peaks over a threshold (pOT serles) rnay be
used to estloate the nean annual flood using a larger number of flood
peaks' The POT mechod should not be used where less than two complete
years of data are available; the metho.d is valj.d for long records
although ln pracElce the annual rnaxiroun method is easier to apply and
equally accurate for records of over five years 1ong. A rnethod of
applying Ehe Eechnlque ls described in the next section followed by an
exanple- A.nore detalled deserlptlon of the por rneEhod is eiven i+
Annex E.

4.2 Descrlptlon of Method

A11 aval1able data shourd be assembled and the compJ.ete years of

data ldenEifled (the starting date of rhe year is not important).
Following a cursory examinaEion of the data a flow threshold is chosen
so Ehat on average between two and five peaks per year exceed the
threshold, the exact number noE betng c ri tica 1. From Ehc compl L-te
years of data (N years) all flow petrks exceedina rlrt- f]rri',rr\,)lrl . (l
are abstracEed; Ehese M flood peaks qi (i = r,2 v) form rhe neal.q
over a threshold serles.

Where the peaks are to be Eaken from a stage record a stage

threshold, !o, can be chosen as [he basis for peal< selection. The
absEracted stage peaks (hf) are converted co flows (qi) using an
appropriate rating equation.

In selectlng peaks care should be taken to ensure tllirt they are

lndependent. A slnp1e test for independence i-s illust.rated in
Flgure 4. l. To decidb Lf q2 1s independenE f rorn .1 . Litc
',t
separatlon of the peaks (T") must be greater [ha n Ih ree I i mes [ht
rlse tine (Tr) of the firsr peak and the trough (qt) betreen
 Test for independence of flood peaks
q2 indePendent of q1 if

Ts > 3Tr
and qt.tqr

THRESHOLD qo

3
o
tr

k€Tr- >l<.-..-Ts

Figure 4'1
the peaks must be less.than two thirds of the fjrst l)t':rl .l I. Thjs
arbitrary rule was used in the IJK Flood Studies Reporr fNFRC. 1975) es

it 1s objective and easy to apply. If the test suggests that the

peaks are not independent, only the flrst peak, 9-,, shorrld be
included.

Fron the M floods, Q1, over the threshold, go. the mean

exceedence I is calculated frour

IM
^\'
P = I(e1 -re)
M i=l
and Ehe average number of exceedences per year, from
^,.
}. -M/N

The rnean annual flood is then esttmated from

MAF = 9o + Fe.5772 + 1nL) ,3"-I

where ln is the natural logarithm or 1og".

4.3 Accglagy_of resufts

The standard deviatlon of the estimate is eiven by

B .l (0.577'+
-'--- 1ntr)r o.s
':'': '- r3"- t
sd (MAF) = ," t- +I ---- l'
/N -I I
For between lhree and five exceedences per year the term jnside the
brackets is approximaEely 1,1 thus

a
sd (uer) r l.l -: *3"-I
r'N

To assess the accuracy of the MAF it is helpful to remember that

on average, 68 tlnes out of 100 the estirnated value of the MAF wi 1l hr.
within one standard deviation of the 'true' va1ue.
4.4 Exanple of application

'
For Batang ltari at Muara Tenbesl jusE over four years oF data are
available. From this length of record the I'IAF is best estimated bv
the peaks over a threshold nethod.

From an tnltial scan of the record the threshold of 4000 r3r-l

seemed llkely to glve a sultable number of floods for the POT series.
Nine independent floods were abstracted as llsted below

Year Flood (r3"- I)

197 7 4365.6
4032.3
4026 . I
1978 4843 .4
4340. 1
4 r 13.3
1979 4596 .2
1 980 4232.6
446r.3

Floods in the incomplete year at the end oF the record were

ignored.

In the notation glven above

Threshold, 9o = 4000 m3s-l
Nunber o_f years of data, N = 4"
Number of floods over the threshold, M = 9

Therefore P = 334.5 t3s-I

I = 2.25 floods,/year
."i uar = 4464'.3 r3u-I
sd (MAF) = 190' B t3"-I

Thus, using Ehe POT method the mean annual- flood for Batang Hari
at Muara Tembesl is esEimated to be 4460 n3s-I rith a standard
deviaEion of 190 n3s-I.

4.5 Uslng incomplete years of data

It ls often Ehe case thaE fhe flow record fron a starion is

lnconplete and that nuch data would be wasted if onlr'comp1 (.tc veers
of data were used. One such sLation is Batang Hari at Muara Kilis for
which, at the tlne of this study data were avallable for []re period
March 1976 to October 1981, a period of over five I'cars. ,\lthorrgh
only two complete years of data are present in this record rnany large
f loods were observed ln the reurainder of the record ' In thi s
situation the threshold should be chosen so that beEween 2 and 5 peaks
are selected from complet.e years and then the entire record examined
for exceedences. These should be ltsted with a note of whether or not
the year is conplete. For Batang Hari aE Muara Kilis the threshold of
2300 rn3s-I r"" ehosen and the following peaks abstracted.

Yea r Flood (m"s-^)

March L976-1977 2329.5

(coroplete) l4Jq.D

2739.O
2562.2
March I977-1978 2308.6
(conplete) 266r .0
3230.8
2609 . t
2579 . )
2337.9
March I978-L979 ,c<7 q

(incomplete) 2400 .9
December 1980-1981 2596.5
(incomplete) ?3t)4 .4
2583 .6

The conplete years of data are used Eo estimate ttle,rVerage

number of exceedences per Year

nurober of floods in comPlete I:9_r:

number oi complete Years' N

'I O
\=
2
Theaverageexceedence'p,isestimatedfromallthefloods'thetotal
number in thls case being M = 15

F = 24g '04 t3"-I

To estinate the MAF Ehe same equatlon ls used

MAF = 9o + p(0 ,5772 + 1n\) t3.-l

= 2844.56 n3s-I

The standard deviation ls estimated by

g- + J Q.5772+1n\)
sd (MAF ) =
ul1 L. lly /tr

_ 249.04 + 249.04 e.5712 + ln5)

J (sxz) /r5

= 219.36 n3s-I

Usingdatafronbotheoropleteandincompii]L('\'earSLllt'mt.an
ima tcd [o be
annual f lood for Batang ltari at--Uuara Kilts t: t:t
t' Thus for sEations
2840 rn3s-I wlth a standard devlation o f 220 t"-
withinconpfeteyearsofdata,onlycompleteyearsshouldbeust'dto
\' brrt al l
calculate the average number o'f exceedences per year'
available daEa should be used to compute [he average exceedence, 3'
 FRON{ CATCHHENT CEARACTERISTICS
ESTIHATION OF HEAN ANIIUAT FLOOD

Introduc tion

wltt'ti no
Thls chapEer descrlbes a method of estlmaLing the }LAIr
h e me Ihod uses a
'f
flow daLa are avallable at the site of lnterest'
readl 1y obtai nabl e
regressLon equation relating Ehe MAF to four
caEchment characterisElcs' A detailed
descriptlon of how this
equat.lon was derlved nay be found in Annex C'

The regression equation I'las derived using data

representing a
be applled anywhere in
wide range of catchment characEerisElcs and nay
bt-l ow' Tt i s
Java and sum2t.ra subJecE to the constrarnts descrlbed
rrsed for flood
also recommended that the equation should noE be
estixoationinheavllyurbanlsedcatchmentsasthesewerenot
considered in this studY'

It is flrstly necessary Eo estlmate the four catchment

characterisEics,us6.<t_inE'hereg*esslonequationwhicharetabulated
obtained' As
below; Annex D glves guldance on how these should be
of
well as the maps prOvided in rh.is rePort' a topographic map
suitable scale coverlng Ehe catchnenL area ls required'

AREA Catchment a rea (tr2)

APBAR Mean annual rnaximum catchme'nt I day ra i nla L I (rnm )

S IMS Slope index (m tm-l;

LAKE Lake lndex (dlmensionless)

BeforeproceedlngwiththisrnethoditisnecessaryEocheckthar
within the
the catchment characEerlstics of Ehe basin under study ar"
ranges of characteristlcs of the gauged catchmt'rrl 'rsed
in Lirr:
lr't t t('teri-LLcs'
developmenL of the equatlon. of the four caEchmeni '
AREA and APBAR are the IDosE important in lndexing
the MAF' Figure 5'
shows the spread of,the AREA and APBAR data of the
regression data
be used if
set. It 1s recommended that the regression equatton only
lies
the AREA and APBAR conbination of the caEchment under study
within the lnner area shown in Figure 5'1'
 6
=.
!r(J
co5
Es
55.3 Iro
" ?
E 9
F.
E:
$ lr 5E 0)

1E i€
L
J
o)
8ir
;;c;; o_
o
!)

I Er
g!
ie
!E
./--z-\ \
= /\
@
g tl [-; .'. /o
/o
ao
o
o
3
I/ .' : l\ll
o
E
o
o
o
fl
.9 €al 'P I
U'
-a
(U
3. .' i'I
c o
o
o
/
a/
,
o I
o I
a -/t
E /.
o)
o -2 -.4 .O, N

?. ,. E

o
/.
./. o
o :
.ta o (lt
/. a c)
L

UJ f. 5
CE

o
'6
g)
o
E,
IIl
o-

o
o
o-

ooooo oooo 'oF oc)

(oro$elol
FFFF F. I o co (oto
n5 a final check SIMS and LAKE should in the range given below:

S IMS Ito 150 m km-]

LAKE 0to o.25

llaving obtalned AREA. APBAR. SIMS and LAKE for the site of
interest and checked that the values are wlthin Llrt accepr.tblc rtnr
the MAF ls estimated frorn the following equatlon:

MAF = 8.00 x 10-6x AREAVx APBAR2'44sx SIMSO'Il7x (r + LAKE)-o'85 m3s-l

The exponent of AREA, V, is itself a function of catchment area

and nay be calculated frour the formula:

L.OZ - OlO275 1og19 AREA

The table below glves V for various catchmcnL areas and rnay be
used to check Ehat the value of V calculated 1s in rhe rrlrrecl 12npe

AREA AREA V

(tr2 ) (tr2)

I r.020 500 0.946

5 1.001 l 000 0.938
10 0. 993 5000 0.9r B
50 0.973 I 0000 0.91C
r00 0. 965

5.3 Accuracy of method

The two previous meEhods of esEiurating the MAF (Chapters 3 and 4)

quoted formulae for estinating the standard deviation of the MAF.
WlEh Ehe regression equation, howerrer. the factorlal standard error of
the estimaEe is used (Annex C). In fact this factorial sIandard error
is analogous to Ehe standard deviation. There is a 682 chance that
any one flood estimaEe ljes wtthin the range MAF x I.59 (or v.AF + 5oZ)
to MAF/1.59 (or MAF - 367) ot the 'true' MAF.

This large standard error of the prediction eqrratton uny srlrlri.n

sone readers. However, the factorial standard error of the estimate
of the UK prediction equation (Flood Studles Report, NERC 1975) was
1.49 using daEa from 532 baslns. These standard errors are a measure
of the uncertainty of flood estimation on ungauged basins. However
ln many cases flood estimates may be iurproved by using local datal
this ls discussed'in Chaoter 8.

5.4 !1g_.p1u of applfcation

The MAF ls estlmaEed for . Ehe cinandtri aE Tegal Datar uslng the
method descrlbed ln this chapter. From Table A.l in Annex A the
relevanE catchment characEeristics for Tegal Datar are:

AREA = 495.1 kro2

APBAR = 94.O mm

S lt'{S = 2I .6 m km- I
LAKE = 0

Figure 5.1 the AREA/APBAR conbinatlon is within rhe

From
acceptable range. Furthermore srMS and LAKE are wlthin the t imi cs
deflned in section 5.2. The regresslon equation may Eherefore be
used.

Flrst. the exponent of AREA, V, in the regresslon equation ls

calculated:

V = 1.02 - 0.0275 logrr.r 495.I

V = 0.946 (This checks with tlrt table in sccrion 5.2 wh(rr€,\
for.an AREA of 500 km2 is 0.946)

MAF is then es t ima ted :

MAF = 8.00 x 10-6 x 495.10.9r-'6 x 942.445 x Zl.60. Il7 x (l + 0.0)-0,85

MAF = 27I *3
"-I

As the f actorlal standard error of estlma Lt. ,,1, tfrc ij..\i; j s I .

l

the }lAF nay be quoted as 271 m3s-I with a 682 chan,: e [irar rhe i'LA! l ics
berween 431 rn3s*L 127t * 1.59) and 170 m3"-i ( 27t/L 59).
 In fact Chere are 6 years of flow data available at Tega.l Datar
to provide a comparison within the regression method. The MAF from
the nean of the 6 annual rnaxlma is-361 r3s-I with a standard deviation
of 38 t3"-I. This estime te of l'lAF would be used in pref erence to the
regression equaEion estimaEe since lt 1s based on 1oca1 data and has a
signlficantly snaller uncertainty associated wlth it. AJ rhough the
estluation equatlon ,l_s seen to give a reasonabl e es tluuq [e o F Lhe MAF
in thls exarnple it should be remembe red Ehat this wi 'l 1 noc always be
the case. IE should be noted Ehat the sEandard deviation is unusually
snall for Tegal Datar, being orrly 10.5 per cent of the l,lAF, The
average value from the catchroenEs studled was 32 per cenr.
 6. ESTIHATION OF TSE T YEAR FLOd) USItrG A I,ONG RECORD

6. I Introductlon

If more than threnty years of data are available at the slte of

interest a flood fre<iueney curve-can be plotted whlch wl1I a1low
floods of reEurn perlod up Eo the length of record to be estlnated.
This procedure is described in detail 1n this chapter.

a more extrene flood has to be estlmated Ehe curve may be

When
extend.ed with reference to the flood frequency factors glven in
Chapter 7 by the method given in Sectlon 8.8. A flood frequency
curve can also be plotEed when a short.er record ls ava1lab1e but thls
curve should always be used in conjunctlon with the average flood
frequency factors even for low return periods.

6,2 Plottlng the recorded data

The annual rnaxlmum floods are absEracted f rom Lhe N )'r'ars oI c].rf a
and ordered so that the srnallest flood is glven rank l;rrrti the largcsL
rank N. For each flood a probabillcy of non-exceedence is assigned to
1t based on its posltton ln the ranked serles. Thls requires nraking
an assumption about the f-orm of rhe distributlon from which the
observed annual maxima are drawn. If the distrlbution is assumed lo
be a type I extreme value (EVl or Guurbel) distribution then a good
approxlnac.ion Eo the non-exceedence probabillty is given by the
Grl ngo rEen fo rnula : -

tr, _ i - 0.44
^r N + 0.i2
where F1 is the non-exceedence probabillty (or plott Lng position)
and i is the rank of the flood. In order to plot the frequency curve
on linear graph paper, the EVI reduced varlate, yI, must be
calculaEed frorn Ehe values of Ft; thls can be done using Ehe
approxima t lon
yt = -1n(-1nfr)
which 1s sufflciently accurate for plotting purposes.

The values of Q1 should then be,plotted agalnst Lhc' correspouciing

yl on linear graph paper. The iesultlng plot becomcs rarlier morL-
useful when Lhe reduced variate axis is rescaled in terms oF return
period, T. The y va-lues correspondlng to varlous retLrrn periods can

be calculated from

- ln (- ln tl-{>l
T

The following table gives values of reduced varlate for commonly

required reEurn periods.

ReEurn Perlod,T Reduced Vari,rtc.v

(yea rs )
2.O o.37
2.33 0. 58
5 r.50
l0 ))\
20 )47
25 ? ?n
40 3.68
50 3. 90
75 i{.li
100 4.60
200 5.30
500 6 .21

I 000 6.9)

A smooth line should be drawn throrrgh the llottcd poinfs hrrf n('{'(1
not be constrained to pass through the highest point wherr this lies a
considerable distance frOn the rest of the data. Ir tlre 'lIta plot ,s
a straight line, then the assumpLlon of a parenr i''. I rj ist ri btrLiort
appears valid. However, fhe plot is ljkely to show a slight curvature
suggestlng the parent distributj-on is something other Ehan an EV1
although 1n practlce the sanpllng error is usually too large to state
definitely thaE this is the case. Worldwide experience in plotting
'steeper
frequency curves suggests that curves come from 1ow rainfall
areas and from soaller catchments. This trend is by no means well
establlshed (see Annex F) but should be considered when rhe completed
curve is compared wlth one based on averaged f I ood f reorrenr'.,' f ac rore

For the range covered by che curve the flood corresponding to a

require the I'{AF to be estinafed. The upper limit of the range will
depend on Ehe varlabiliEy of the plotted data about the curve: even j f
the data plot on a straight line it should not be extended to return
periods greater than twice the length of record.

6.3 Exarnple of appllcatiol

An estlmate of the 5 year return period flood is reqrrired at the
Cltarum at Nanjung. This station has 2I years of data which 1s
sufflcient to draw a flood frequency curve. Table 6.1 gives an
ordered 11st of the recorded f lood peaks and the correspond i ng va l,.rt's.
of yi based on the rank. The flood nagnitudes have been plotted
against y,1* tn Flgure 6. l. Drawlng a line through these points is
difficult, the best solution possibly belng to draw the straight li ne
shown. The scatter about this line for higher return periods is largt:
and iE tray be best to linit flood estimqtion wlth thls curve to return
periods up Eo Len years.

From Ehe llne Ehe flve year flood can be estimated as 290 m3s

and the ten year flood as 312 *3"-I.

Table 6.1 Ranked floods. and the corresponding val ues_gl_: i Jlf !!S
Citaruro at Nanjun_g

Ra nk Floo d EVI reduced variate

1 Qr y{

2T 3 70.0 J.OZ

20 303. 0 2. 60
l9 297 .0 2. 07
l8 293 .0 1 .71
1-t ,ol n t .43
l6 288. 0 1. 20
l5 286.0 I .01
l4 284 .0 .83
IJ 284.0 .67
t2 27 4.0 ,)l
II 270.0 1e

l0 27 0.0
9 27 0.0 1)

8 268.O - .01
1
26t.0
t) 253.0
,q1 n - .41
226 .0
3 221.0 - .73
2 208. 0 - .96
t 205 .0 -I .29
 Flood frequency curve for the Citarum at Nanjung

;
(i'
E

o
400
o|
(tt
o
o
€
x
E
o-
3oo

RETURN P€RIOO - YEARS

2 5 10 20 50 100 26 soo
I I ltl

1..2"-34
''*{i.
Gumbel reduced variate IVI
Figure 6.1
 7. ESTII{ATION OT TEE T YEAR FLOOD USING BEGIONAL
GROITTE FACTORS

'7 1
Int roduc tlon

A flood frequency. curve gives a graphlcal

representa tion of th(]
relatlonship between rhe magnitude of a flood
and its return period.
If the graph 1s rescaled by dtvldl ng the
flood rngnitudes by the MAF,
a curve of growth factors agalnst return period
is the result. For
example, the all catchment average flood
frequency growth curve For
Java and Surnatra shown on Figure 7 .1.

However, analysis ln Annex F shows

that growth factors in
rndonesia vary not only with return perlod
but arso wlth the size ot.
catchroent under study. 1001 statlgn years
of data from the g2
catchments with flve or nore years record
were used in this analysi s.
Ttris al1ows grovth factors for events up
to 500 year return period L()
be estlmated. Growth factors for the r00o year
return perioci are
tentaEive and should be used with caution.

7.2 Description of nethod

Estlmation of T year return period flood, Qf, involvls

multlplying the MAF by rhe appropriare growrh [actor wlriclr is ;r
funetion of T and the caLchment area:

Qr cF(T,AREA) x illAI;

It is assumed here that Chapters 3, 4, 5 a nd 8 have been

consulted and an estitrate of MAF obtained from whatever data a
avai lab1 e.

The growth factor is obtalned frorn Table 7.1 interpolating for

both the required return period and catchmenE area.
 Average flood frequency growth
curve tor Java and Sumatra

ll.

=
a

RETURN PERIOO - YEARS

20 50
 Table 7. I Table of Growth Eactors GF(T,AREA)

eturn Reduc ed
Catchurent area
e riod Variate
(t*2)

r80
or less
900 r 200 r5 00 I
I

or morei
i

'z n" qBlo 7 o Q-r9tz O t9l

5 1.50 l.2B e r ..2.7
'.9r1-t (o., 92 5
L:24 p, l' )) I lo r. t/
l0 2. 25 r.56 | .54 r ..+6 L . q.+ | .41 | .37
20 1. > I i .88 r.84
50 3 .90
1.75
Lzq | .64 r.59
,L-. J ) 2-39 ? ra 2. ro ,n.l
r00 i.95
4.60 2.78 2.7 2 2. 47 2.37 ) )'l
200 5.30 3 .27 3.20 3 .01 2.3t)
500
2.78 2.6r. :l . '1
o.lL 4. 01 3 ..92 3.7 0 3 .56 l.4l 3 .2V --.'-."-_-l
1000 6.9r .+. o6 4 .58 4.32 4.16 4 .01 3.85

Alternatlvely a flood frequency curve

can be constructed for the
required area using the MAF ancl growth
factors for retrrrn periods
glven in the table; from this the
flood corresponding to any return
perlod can be read dlrectlv.

7.3 Accuracy of method

rn estimating Qt in this way errors

.ri se Irom f wo s(]rrrr.r,,l
error in the MAF estimate and error in
the growth Factor. chapters r
4 and 5 each give nethods of assessing
the standarcr aeviation
associated with the I'{AF estimate appropriate
to irs method rrt
calculatlon. The standard error of estimate
of the growth fac to r i s
hard to quanrify but the'IJK F100d st'dies
Reporr (NERC, 1q75)
suggests it to be of the order of 15% at
r = 10 years, 307. ar T = 100
years and 502 at T = 1000 years. This
approximaEe relaIiorship can be
summarised and used to estlmate the
standarcr deviaLion'i rirr, e.rr.rwLl,
factor, sd(GF):

sd(cF) 0.16 (logrsT) x GF

where GF ls the growth factor

 The standard deviation of
Qf is then found from

sd(Qrl = ar[,sd(GF))2 + r.dlIHF)r2iO.5

' GF --/ J MAi'

If the regression equation is

used co estln:lte the MAF, the
term
rsd(MAr).
\- I
z
MAF

may be approxioaa"O ,o O, (0.59) 2 or 0.348.

7.4 Exanple of agglicatlotr

The 50 year f100d is requlred

for Batang Tembesi at Marrra rnum.
Twelve years of data are
available at the site, sufficient
the I'IAF by the mean of the annual to estimarc
maxlna of the sampJ.e as described
Chapter 3. in

Thus, MAF r roq .4,?1m"s-.

sd ( t{Ar ) 34r.4 13"-I

The catchnent area ls 150_5 km2, and f rom Table 7.1 the
requi red
groirth factor is Eherefore 1.95.

Qso 1164.4 x 1.95 = 2271 m3s-l

To esEiuate the sEandard deviafion

of this estimate

sd (cF) U. )J

therefore sd(Q5O) 2271 /f( 0'53\2 + r\_--_---

341.4.21
I. y) 1164./+
,,, I

sd (Q50) 908 r3"- l

The 50 year flood for Batang, Tenbesi

at Maura Inum is estiurated.
co be 2300 13r-I, wlth
an associated sEandard deviation
of 900 m3s-i.
 8. HPROVING THE FLOOD ESTIHATE
USING LOCAL DATA

8.1 fnt roduc tlon

Thls chapter ls concerned wlth

naklng the best use of rlver ffow
da'a avallable near to the sl.e
for which a f100d esttmate 1s
requi red. As Ehere are a
large'-hunber of rlver gauging
sunaEra and Java it is likely
stations in
tha. some 10ca1 dara wil.r- be avairable.

rf data from a statron used in

this study are to be used for a
flood estlma.e then data avairable
aE the tlrne of the study can be
found in the Data Appendix. These
data should then be extended to
present and checked in the light the
of recent devel0proents of the raring
equat to n.

Having obtained a1l relevant

rocal daEa, the )lAF shorrr.l
preferably be esclurated by nore
than one method. The f echniqrrc.s wh
nay be used, that is the annual qaxlnum ich
serles, peaks over threshold
and regresslon equatlon, have been
described in chapter 3, 4 and 5
respectlvely' Figure 2.1 guides users
in Ehe cholce of suitabre
methods' A satlsfacEory agreement between
Lwo or more approaches
would lndlcate that Ehe MAF was reasonable,
whereas a disagreemenr
nighE indicate that sp.clal circumstances
need to be considered. This
point 1s consldered In Section g.2.
Sections g.3 to g.6 deal wirh rhe
problen of estirnating the I'{AF where
more abundant <.ra[a arLr avairabrr.
at gaugrng sEacions 1n the vicrnity or
Llre IIood esLimation si Le.
usually it is t.he case that Q1 can only
be estimated in one way _ bv
using the growth facEors glven in Chapter 7.
Howevsp, whenever
possible this should be conpared with
an estimate using a frortl
frequency curve protted from locar data.
A discussion orr how tlris
local data flood frequency curve may be
extended to enabre esrimatio.
of high return period floods is gtven in
sectlon B.g. Some genera.l
comnents are uade in Section g.7
on the use of 1oca11y availabre frood
level roarks' rncorporatlng such historical
data into tlrt,frood
estlmaElon process can only be done
in a subjecL i,,,,, wdy lrr;wi1,i, 1-. ,.,icl.
sectlon ls illustrag.gd with one or more
exarnples.
 Flood estimaEion is selclom .
"i.righrforwa rd Iirsk
unusual conditions require considered
Special or
.

judgemenr or even modi f ication

of the techniques described in this manual. For
these reasons it is
reconmended that the task of flood estirnation
should. be undertaken by
an experienced engi.neer or hydro-1oglst.

8.2 Uslng different methods at the sane slte

Tf possible rhe MAF shogld be est i ma ted hy moro

7114n onc mcthod .

If there are some da[a at ttle pol nL ol lrrLer:est Llrls tutv

lrt,;tr.lr[,.r,,.,1
uslng the pOT merhod (Chaprer 4) in con jtrnc tlon wJ
th el ther f ht.
regression eqrrht i on (Chapter 5 ) or the mean of
the annrral maxima
series (Chapter 3). In such cases the regression
equatlon estimate
will be a poor predictor of Ehe MAF conpared to
the annual rnaximum o r
POT methods based on real data.

Should there be a large dlsagreenent between the ciifferent

flood
estlmetes, the calculatlons should be checked carefull.y for
arithmetic
nistakes. rf the disagreement persists, the choice of an
appropriatc
estinate of the lrAF is a natter of engineering
,iudgement. Advice on
how such a choice may be nade is glven in general
terms in rater
sections of thls chapter.

If there is reasonable agreement between the different nethods

used, this may give added confidence in the estimate. How thotrgh is
this added conf idence ref lected in tlre quoted val ue Ior t]re l\fAIr ar.r,
its error? As explained rn the relevant chapters for e.rt rr int Ehod of
compuEing the I'{AF' an estimaEion error or sEancl,r rir deviat i,,n ;
s
associated with the calculated MAF and it has been slrown how tir is 11r.i.-
be found ln each case. The standard deviatlon or standard error
enables users to assess how the MAF obtained by any method mighr
relate to the "true" long term mean annual flood at any siEe. At
several points in the report it has been statecl EhaE using nornal
probabllity theory, the estirnated MAF would be expected ro 1ie wirhin
Ehe range of plus oi mlnus one standard deviatlon of the long term
IDean wlth a probability of 68 per cent. slmilarly rhere is a 95 per
cent probabillty of the long term mean annual flood betng within the
range of plus or nlnus 1.96 x the standard deviation of the estlnated
 MAF' The large estlnation errors associated wirh rhese merhods show
how lrnprecise estlna-tion of rare floods can be whcre o.ry
shorL
periods of flow data are available or from the regresslon equation
For
the MAF of ChapEer 5.

rn cases where the MAF is estimated by,ore than one method

each
estimate wl11 have a standard devlation associated with
it. Norrnarly
one of the methods is roost suitable for a particuJ-ar
application and
this is the estrmaEe chat should be used. rf other
estimates agree
with it then the various assumptrons made in obtaining
the estimares
appear validated; 1f not the methods and their
assumpt j ons shorr.r d he
revlewed to Ery to explain Ehe discrepancies. Example
2 shows how por
and regression equaEion estlmaEes may be reconcired
for one partlcurer
case' By eiEher gettlng good agreernent between methods or
findins
reasons for the dlfferences confidence in the
f100d estimate is
enhanced, albeit it in a rather intangible lday.

rt would nornally be inprudent of an engineer to assurDe

that because
the MAF was an imprecise estimate of the long term mean
annuar flood
one or nore standard aevtations should be added to
the MAI estimace Lo
allow for this imprecrsion. Such conservaEism courd increase
the
ref urn period of the f lood estlrnate dranatically wi
[hout t]re user
being aware of the fact. Given that the standarcr error of
[he
regression esEimace of the MAF is plus 59 per cent, acrditions of
th
error Eo Ehe MAF is equivalent to rhe difference between the 1O vea
r
flood and the 60 year flood for a catchment of 100 km2.

rt is normally appropriate to accept the MAF es t irur :,-cJ by the

nethods of this report for design purposes sirr.r. trris is rr;,, rr.s
t
available central estlmate of the crue mean annua.l f.l oorl . llowi,,.,t,r.
all avallable local flow data should be used Eo check and refine rhis
cenEral esEimate as shown in later sections of this current cirapter.

Exanple I

Problen description: An estlme Ee the lO0 year reLurn period tlood

is requi red on the Krueng Aceh at Kampung
Da ra ng.
 Ihree technlques are used to estimate the f{AF
and their. results compared.

DaEa abstraction : Catchnent characteristics were absIracted for

(regresslon equation) this catchment as described in Annex C:
AREA = 1068 km2
APBAR = 86 nn
SIMS = 2l n km-I
LAKE = 0

Data abstraction A threshold of 4. 5 m (to the new sfation datum)

(P0r) or 266 r3"-l ,." chosen For the pOT analysls
and fhe following peaks abstracted:
450 3s0 369 282
300 287 434 ar/, ul
rz.+ s - (in
-3--I r. 4 completc
309 3 13 427 1qo
JJJ years)
42A 3 u 337

Addltlonal flood peaks in i ncompl ete yea rs a re

279 748 13"-I

(Although rhe hydrological yea r for the annual

maxiurum series analysis starEs on 1st August,
the POT nethod can take data f rom any complete
12 nonth pe-_r_iod, regardless of the starting
month. This explalns why the largesI Flood on
record ab the stat.ion, 74g m3s -I , appears rn
Lhe incomplete yea rs ttere but in the r:r>mpl,.tr.
years below)

DaLa abstractlon There are only four compl t,f t,.hyrl roloeic,rl y1 rs
'
(Annual naximum of daEa .at this station an<l tlrt .r nnrra I 0u :< i rna
series) are:

1976/r977 450 m3s- I r97B/1979 748 m

3s- Io
L?77 /re78 434 m3s-I l 980/ I 981 337 .3r- I

*The origlnal sta[ion was desEroyed

after this
exceptional flood in August 1979. This year's
daEa may be considered complete as it is known
Ehat the August peak was not exceedcd durtnq
the same hydrologlcal yeirr.
MAF (catchnent : uslng data from above, the AREA exponent, v, is
characteristics) calculaEed thus;

V = l.O2 - O.O2l5 log10 AREA

V = 0.937

MAF ls estinated from the regression equation

glven ln Chapter 5;

MAF = 8.00 x 10-6 x AREAV x ApBAR2 q45 x

SrMSo II7 x (l + LAXE)-o 85

'I4I-:13?-'3"-I

The standard error of estimate of i,lAF i s 422

r
1.59 to 422/\.59 (6it ro 265 m3s-l).

MAF : Using the rnethod descrlbed in Chapter 4, the

(POT Analysis) mean exceedence, B, is calculated thus:

1M
P= x_(_q.r-qo)
M i=i-*'

9 = 104 ,3
"-1 1,,"ing al l l7 f
joocls)

fhe average number of exceedences per

_vear, ).
1s calculated

\ = M/i.I

Conplete years must be rrsecJ in this erlr_rrlation

and there were 15 floods above the Fhrnshntd r-
4 conplete years:

)\ = L5/4 = 3.75
 MAF may now be calculated

MAF = 9o + p(0.5772 +
Jn \)
MAF = 266 + 104 (0.5772 +
ln 3. Zs)
{4l--=--4€-d-l---t (s ta ncla rrr qf p1,i a I j o n
57 m3s-1)

MAF (Annual maxinum

The estlmatlon of l,iAF
f rom an annual nnxi
series) serles of less than 5 years
mrrm

in I enqth is not
reconmended. Tt i s ca lcul
ated here merel y as a
check on other methods.
From the 4 years whic
h
a-re avallable.

MAF = 492 n-?r s-^

(standard devlation
178 m3 s

Discussion : The three estima[es

of MAF obtai ned above are
in satisfactorily close
agreement. However th€,
preferred estimate of
MAF mtrst be that obtained
by the pOT method si nce rhis
i s most sui tab,l e
for the.
"t,o.t length of record at Ehe station.
This is conf i rmed b_y the
iow s f , n,r ,
or rhls merhod (57 ,3.-r, . ,::::, -r ^..:
, .

:: i.:;":,::'
annual **itut series methorr
(r79 -3.-l),.,,,J
the regression equation error
()lAF + 249 *1.-l
to MAF -. j57 rn3a-l). ilowcr,,.r
;rs i, ;, IrAi .r,rn
the annual maximum series
is sl ightlv higher
than tha t from the pOT merhocl.
i t j s wi se to
round up the pOT estimate
of MAF ro sa_v 1ZO
*3 /". The regression equa t i on
es t i ma te .i s
acceptabJy close but it
is alwavs hetter to rrso
flow data with the pOT or
annua I max i mum seri es
whenever possible.

So lu tlon
Qtoo for the locat;'.on is obrained
by
mulriplying MAF by rhe lO0
year return period
growth factor of 2.41 (AREA
t068 km2).
(Table 7.\)
 Q1OO = 470x2.41
= tt33 *3"-I

From Section 7.3 the standard deviation of the

100 yea'-r return period growth fac r.r i s 3oZ.
This nay be combined with thc stanCar,l
deviation of Ehe POT estiuuate o1. fhr-,yAF to
give the standard deviation of Qlgg

sd(Qroo) = Qt00 [("9-(GF]y2 + 1!dt$f,),zio s

GF MAF

= ll30 [f-'723t2 + (:7

'470')210.5
2.4t' J

= 366 .3 .-l

The estinate of Lhe 100 vear retLrrn period

flood is therefore ll30 m3s-I with a standard
deviation of 370 r3"-1.

Ele_'plL2

Problern Description: An estinate of the 5C year retrrrn nerio<i floocl

is required for the Ciliwrrnp nt Kcl-,on Barrr.

0n1y two complete years of flow Cata .?r.

available tor the site, This is too short t.)
attempt an estimat ion of Lhe MAF from tJre mean
of the annual series. The MAF can. however be
estimated by both the pOT method and rire
regreSsion equation.

DaEa : The catchments characLeristics for tiris river

(catchnent basin Lrere obtajned using Lhe procedures
characterlstics) described in Annex C:

AREA = 333 km2

APBAR= 103 nm
SIMS = 34 m km-I
LAKE = 0
 Data (POT anal!sis): 0nly two conplete years of data are available
and they contain the fol lowing peaks over a
threshold of 9O n3s- l.

173.4 106.8
97 ,7 131.9 ,3.- I
92.0 93 .3
105.1 9r.l

Four addi tional peaks were recorded in

incomplete years:

209 .4 166 .9
- .3"- I
97 .4 95 .0

MAF : Using data troun above, tlr., rggress ion equaI iL,n
( regression is used to estimate the MAF;ts described in
equaElon) Chap ter 5 ...

239 m3 (standard error of estimate

'fi'= "-I
380 to t50 n3s-I)

MAF (P'T) : Using rhe por nerhod described in chaprer 4,

wi th the data given above the MAF is
estimated
as:

MAF = r52 ml*- I (starrcrarcr d.rri;rr- irr. ?5 ,n3,,- i)

Discussion : Two years is arso really too short a record for

the pOT rnethod, however it is preferable ro
make use of even rhis shorc record as
a check
on MAI estimated from the regression equation.
The Ewo methods do give noticeahly different
results; 239 n3 .-l Irom the regress ion and
I52
n". s-' from pOT. Why shoutd rhis be?
 Oneposslble explanation can be
tound by
studylng the catchment shape
, of the CiJiwung
basin on Figure l.l (Catchrnent
number 5). The
cacchuent above Bogor nay
be describecr as
typical, but thereafter it
is very long and
thin. As the nountainous
catchment around
Bogor is the naln flood producing
region of the
catchnent, the 10ng river reach
ac.'oss the
coastal plaln towards Jakarta
wl l.l slgnficantlv
attenuate floods produced in
the upper
catchment. Furthermore the
lourer catchment mav
not produ.u-. l.rge flood
runoff because of its
narrowne.as. One mlght expect,
therefore, that
the MAF produced by the regression
equation to
be coo high - which is indeed .the
case.x
A tentativ€ estimatu ol
Uef, incorporating a
factor of safeEy for the shortness
oF the
record would be 200,3"_I

Fron Table 7.1 the 5O year

gro\rth factor for
catchment of 323 kn2 i s
2.29 giving rhe
following estimace of
Q5 o:

Qso 2.29 x 200

o-^
x)u 458 -t
m3 (standard dr.vi:rf ion
?l
rq) m-s-.)

*Incidentally a shape factor indexing

was conslder in the the na.rorn"r" of the ca tchmenI
regresslon
slgnlficant when apptled ro analysis (Annex C) but not Forrnd
the whoie d.r;";;;."
 8.3 leUe_tt'u_gbserved MAF from an__ug_-igg5!J_]plg rerm srarjon

The nethod descrlbed in this and the following two sections

uses
a siurl 1ar approach to adjust the MAF. rf more than one of these Ehree
approaches is possible at a site, ar1 should be tried.
Engineering
judgenent shourd then be used to assess the
inportance of each l4AF
estlnate in order Eo produce an overall balanced est imatc of
the l,tAF.

The technique described in this secElon is usefrrl whe,rr,

a flr>r.rd
estimate is required at a statron, A. where the period of record
is
shorE and Ehere is a^-*stat10n, B, with a longer record nearby.
MAF
-
when estimated from a short period of record uray
be higher or lower
Ehan the true long tern uean because the record may come from a
time
when floods are higher or lower'than normar. due
to short term rocar
cllmatic varl-atlors. This sampling error be reduced if it assunecr
thaE Ehe sane period in the history of the'.y
sta.ron h,l th Ehe longer
record was'slmilarly wet or dry. For thls to be true the
caLchmenrs
should be ln close proxinity and share similar climaEic
catchnenc
characterisEics (AAR and APBAR). This assumption Day be checked
bv
plotting the annual maxina fron the common years of operaIion
against
each oEher to establlsh the degree of correlation.
The acl iusrecl MAF
is c,alculated thus:

MAF
n
I,{AFA * g!
MAFs

where
MAFA Adjusted MAF for sEation A

llAFi MAF from the record at station A (Unadjusted)

MAFn MAF from the enti re period of operation of staIion B

MAFi I'{AF f ron statlon B during period statlon A was

operaElonal
Elegll-e-!

Description An estimate of the I'{AF is reqrrired on thr.

Batang Pasaman at Air Gadang. Six years oi
data are available at Air Gadan5l . However tlre
nearby s[,ation at SlJaping on the Batang
Batahan has a \2 year record.

Data --:' Yea r BaEung Pasanan Batung Batahan

Air Gadang (342) SiJaping (3,41)
13 I 3 _i
MS
"-
39-40 139.3
40-4t 247.)
4r-42 388.3
317 .2
73-74 303. P.

74-75 466 .3
75-76 898.5 17C.?
7 6-77 t147.9 !6(,..J
77-78 970.9 .,,1c l

78-79 694.4 399.5

79-8 0 1036.1 5C8.0
80-81 ll4l.0 430. o

CatchmenE characterist ics

AAR 344C mm 3 l rl(-r mm

APBAR 103 mm ll,q mm

AREA I267 km' It,* ;:m

Discussion AJthough there is a consirJerahlc dilrt re,r.lc(, in

caEchmenc area, the tvJo ca[chments are s jrrit;rr
cljmatically. jn close proximiry ancl drain in
the same direction. A1 though the correlation
between tire annual maxima of these two
statlons. over the common period o1'record
(75-76 to 80-81). is poor (correlar ion
 coefficient = 0.38), the example is contlnued
as an illustration. Tn tlre, notat irtn desc ri h, ,l
above:

MAF'342 = 981.5 ,n3"-l

MAF:a: = 359.6 ,3r-I
MAF'343 = 408.8 t3-I

The adju,sted MAF at statton 342 is calculated

thus:

MAF:az = 981.5 x 359'6

408. 8

MF:aZ = 863 ,3 I
"-

Using this technique the revised estlurate of

the MAF is 863 r3"-l

8.4 Using flood recg:lS_l:om elsewhere ln rhe catchmenr

Sometimes it will be necessary to nrake a flood esIimatc.rt a

poinE A on a river which ls some distance upstream or
downstream of rn
established gauging stacion B. provided chat the differences in
caEchment area are relatlvely smal1, these clata nray
be usecl to assist
in flood estimation at the point of interest. Tt is suggested
thar
this technique only be used if the'difIerence in are;r bctw{,en tlrt
iwir
catchments is less than 502.

The ,oF at the point of interest, IGFA, is

calculared [hus

p MAFn
MAF^
A = MAF^
A X ---3 MAFB
 where
R
MAFIi = regresslon esEimare of MAF at A

P
MAFii = regression estimat-e of MAF at gauglng station B

!{AFn = MAF ar sration B from dara

E14tg!_e_a

. DescrlpElon : The MAF is required on the Krueng -lambo Aye ar

Ranpah. For the purposes of this example no
flow data are assumed to exist there.
There are. however, g years of flow data
downstream at Lhoknibong. TF the di fference In
catchment areas is not too greaf, these data
may be used to refine the MAF.

Krueng Jambo Aye

Rampah (117) Lhoknlbong (lt8)
Anre (krn 2) 4o6t 4403
.APBAR(mm) 65 67
SIMS (n t<m-I) 10.8 s.35
LAKE-OO

The MAF at Lhoknihong. estimated from tlrc

annual series of 8 years. is 932 rnl .-l

Analysis : Be[orr. app']]'inll thr- tcchniqrrr.di.sr.ri]rrrr ir.rl,

section, we mLrst check that the two catchment
areas are wi.thin th e 5O"l of e6sh other. The
difference in area is in fact onlv g.a\. The
technique nay therefore be anpl i ert.

Using the catchment characteristics gi ven

above. th; regression equation of Chapter 5 i"
used to estlrnafe fhe MAF af both stations

MAFT,, = 598 t3/" (Regression esIimare

LL/
for statlon 1i7)
MFT, o =
116 572 t3/" (Regression estimare
for station tlg)
 From the'dnnual rnaxinum series at Lhoknibong

MAFttS .= 932 m3s-r

The regression es Elma te o[ ,yAF at Rampah i s

modlfied thus.

MAF,,.,
LL/
= ltlFf,, x YI-t1t
*tT, e

MAFll7=598"212
672

MFttZ- = 829 ,3
"-I
The revised estinate of the llAF is therefore
,
830 n3s- i

8. 5 !11ng_l1ood_ reco rdS*ffeg_egJ9_1 ni!€_egJg!s9!_r!

rf Ehere are no sui-tab1e gauglng stations within the catchmenr

of interest, records from a station on an adjoining river hasin
may be
used to assist with estirn*tlon of MAF

It is
suggested thac this technique only be trsed when tlre
two
catchments have broadly similar cha-racteristics
and i n parricular tht-
two cat.hr"nt areas differ by no more than 5OZ.

The MAF at Ehe point of interest, MFA, is calculated [hr-rs

MFa = MFx] g*
MAFR
B

where

*tl = Regression esrimate of MAF at A

*tfr = Regression estjmate of MAF at gauging station B

MAFS = MAF at staElon B fron data

 Exanple 5

Descrrptlon : Estrmare MAF on the Batang i\r r Dingi n at l-ubuk

Minturun for whlch r/e assume there i s no f .l ow
data (for rhe purpose of Lhis example).

The adjoining catchurent of Batang Kuranji at

Gunung Nago has 8 years of data which may, if
the catchments are slnilar, be used to assist
with the estlnatton of the MAF aL Trrbrrk
Mi nturun.

The followlng catchment charac teristics wer.r

abstracted for the two sltes usins fhe
procedures described 1n Annex C.

Batang Alr Dlngin Batang Kuvanji

Lubuk Minturun Gunung Nago
(321) (314)

AREA (tr2) I l4 r22

APBAR (mr) 147 147
S IMS (m tn-
I; 75.6 70.3
LAKE 0 0

The I'IAF at Gunung Nago, estimated From the

annual sdries of g years, is 4I5 ,3
"-i.
Analysis : The two catchmen.s have similar caL,..'ment
characLerlstics and are i r-, r. lose prr.x i mi tr,.
The difference i-n area is 1"/. Thc |lAt:
adjustrnent technique Day therefore be appli ed.

The catchuent characterlscics given above

enable MAF to be estimated at both staIions by
the regression equation of Chapter 5.
 vanR 266 n's-^ (Regression estimate
J l.l
for station 314)

*tzt 253 m3 s-1 (R'Ur'ssion tstirat€

for station 321)

From the annual maxlmum seri es at Gunung Nago'

MF:tq 4I) M
3 S _i

The regress i on es t ltoa te of MAF at Lubuk

Minlurun is adjusted thus: -

I{4tr
''^'314
MAF3
t *ttzr *
2
*.1.J lq ,

ql)
MAF:Z r z>J x
266

*t3zt 395 *3
"-
I

8.6 Using staff gaute data

rt is possible thar statf gauge readings have been taken L)\,.r d

period of time close ro the si te of interest and nonL- or
'ery f.u
current meter gauglngs exist. rf it is not possible to develop a
raEing as suggested in Annex B, two techiques described belcr^l enable
these data to be used in a flood anarysis. Both techniques require
five or more years data. The annual maximun series should be
abstracted and wil1, of course, be ln stages (m) not flows 1r3 r-I).
 The flrst
rechnique invorving the rrse of stage data
re.quires the
abstraction of the median stage and its conversion
Io discharge rrsing
a flow resistance formura such as the llanning or
chezy equation. The
median of the annual ,naximuo stage series
when converted to a
discharge ls directly equival-ent to the
medlan of the annual maximum
flow seri"r Qr"d ' Analysis of data from
s[ations used in Lrris
study showed that. on average

MAF 1.06 Q*"6

This relationship may be useci to derive the I,{AF after Qp€,d has
been determi ned.

The second nethod of uslng data from

an unrated section is useful
where Ehe station has a 10ng record
of staff gauge data. Here it is
possible to plot a flood stage frequency
curve rrnd rhis is done r,n lh.
same lray as for a f100d discharge
frequency curve which is described
in chapter 6. There is a direct correspondence
between these two
curves and the T year return period flood
stage in metres ls
equivalenE to the T year return perrod
flood f10w in cubic metres per
second' This stage can be converted to
f10w by either the }danning or
chezy formula' However in flood deslgn
it is ofEen Lhe maximun level
of the T year return period flood whi-ch
is important. rn this case
conversion to discharge is unnetessary.

-r"9eLls I Perempuan Cantik aL Banyak |4asalah

A1 thoughflood peaks have been recorded by an

obse rve r for 9 years the station
is unrated.
survey of the channel was underIaken to
estimate the flow at [he median annual
stage.

The followJ.ng peak stages have been

recorded by
the observer;
Da ta : Yea r peak s tage yea r pea k s ta ge
Dt

IZ_/J 243 77_78 4.01

73-74 3 .07 7 8-79 2. 44
/t+-/) 2.78 79_80 3.48
75-76 3 58 80,-81 2.8g
7 6-77 2 .88

From a survey of the river cross_section the

following inforuration was obtained tor
the
rnedian stage of 2.g9 n:

Cross sectional area of

f1ow,A =l.O3n2
Wetted perlmeter, p = 43 m
Water surface slope, S = O.0lO/ ur m-l
Estlnated Manning's n = 0.0.1

Fl rstly che hydratrl ic raciius, R i s calcurated

thus

R = A/p = IO3/43 = 2.4 m

The average ve}ocity of flow. v. nray

now be
estimated using ldannjngts ect rral,.o-.

v =_l
n
x2/3 sr/2

2.r+2/3 x o.Ol 07r/z

0. 04

v = 4.64 m s-I

Median flow discharge, Qmed. is of,rrincrt by

mtdtiplying velocity by cross scr_tion;r I trea,rf
flow at 2.89 n srage:
 Qmed VxA
Qmed 4.64 x 103

Qmed 478 rn3 s- I

Using the relatlonship given above between the

MAF and Qned, rhe MAF is then calculated:

MAF 1.06 x 478

MAF 507 n3 s- I

8.7 Using flood marks

Levels of hlstorlc floods Day someEines be found nnrked in the

vicinlty of a rlver. These typically occur as a Line inscribed on a
brldge pler or a ttldet mark on wal1s close to the river. In the
eourse of this study,-fsolated flood marks were found ln the vicinity
of several gauging stations. The exlsEence of such flood roarks is
often known to Ehe gauglng statl.on observer or local inhabitanEs.

occasionally in rndonesJa these excepEi_onal. floods ei ther

submerge the gauging station resulting in the float reaching irs poinr
of maximuu travel, or in the station being washed away completelv. rn
either case the peak flood stage is not recorded and rnay be
substltuted by levelllng in flood marks to the station datum. Tt rna_v
only be possible to geE a crude estimate of peak crischarge at this
high 1eve1 due to excessive rating curv€r extrapolation. llowevt,r r't i:;
better to include this albelt doubtful flood record in rht,annual
maxinum f lood series or POT series f or the sta t i.. rhan rrri t i
r.

entlrely.

If the extraordinary flood occurs outside the period of record of

the gauging statlon rhe. problem becomes more complex. Section 2.g of-
the united Kingdom Flood srudles Reporr (NERC, rg75) discusses rhe
problen Ln some detall and it is considered further in Annex E.
However to nake thls conplicated analysis worthwhlle a series of flood
uarks above some datum and Ltlth dates are requirerl , Srrch a si ttration
ls unlikely to occur in Indonesla and was not noted dtrrjng rhe cotrrse
of thls study. rf rhis exceptlonal flood is quoted bv loeal peonl(- 1c
'the blggest f lood in llvlng mernory' , a very crude estimate of rht-
 between 30 and 60 years. This is equivarent,
trsi ng Lrrr. cringorIen
fornula given in Chapter 7, to a return period of about
50 to IOO
years' At best this can be only a very crude check
on the f100d
estlmate obtalned by other means. what it does show
is that rhe river
in quesEion is capable of producing froods of rhat magnlEude.
Thus if
the peak flow obtained frour thls extraordinary fJ.ood was greater
than
say Q500 estirnaEed by rhe regression equatlon
and growth curve it
nlght indlcatE that the regresslon equatton esElmate
v,7as rather row.

8.8 Extension of a flood flsquency curvs

rn those cases where sufficient data exlst to p10t

a fr-ood
frequency curve it is often the case thar this
curve cannot be
extended to the required return.periods. For very
large return
periods of 500 years or more it.is best to use
a flood frequency curve
based on Ehe growth factors given in chapter
7. However for
lntermediaEe return perlods and to avoid a sudder-r junp
from the
plotted curve to Ehar based on growth factors the
following procedure
is recommended.

use the method of chapter 6 to p10t the f100d

frequency curve
based on the available data and decide on
the rirniting retrrrn pertod
up to whlch the curve may be used; call this L years
and the
corresponding flood QL. From the table of growth
tactors
(Tab1e 7'L), interpolate che values appropriate
t. tht'catc6me.t area:
deno.e these chosen growth tacEors by GF(T), where
GF(T) is the growth
factor corresponding to the return period T. If
GF(L) represents the
value of GF(T) for T=L, then the f lood
Qt i" gir., ,. r,.

Qr o, cFlrl
XL
" GF(L)
ror rerurn periods greater than L up to about 10
x L years. For
floods of return periods greater Ehan 10 x L (or 500 years,
whichever
ls smaller) estlnate Qr using Ehe mean annuar
flood derived from the
recorded flow daLa and growth fact.ors of rable
7.1. A smooth curve
should be drawn to link the three rrne segnents.
This curvc can rhen
be used to estirnaEe the nagnitude oi the desired frood.
Exanple 7

Problem description: For the cltarum at l.ianjung a frood frequency

growEh curve was constructed in Chapter
6.
This was con.sldered vaLid up to the 10 year
flood but how should lt be extended for use up
to 500 year return perlods.

Analysis .-- The nean,1_nnua1 flood can be esrimated e'ther

fron the mean of the annual maxlmum series or
by reading the value frorn the flood frequency
curve driwn fron the local data corresponding
to T of 2-33 years (Flgure 6.I).Frour the anntral
maximum serles
MAF = 27O.f ,3
"-I
Q1g was estiur,aEed to be 312,3 s-I in ah"
example in Chapter 5. The catchnent area
for
Mnjung is 1833 kn2.

The following table sets oLrL the c;rlctrrafion,

for Q1 . In the first colum;t are tite return
periods for which points wi I I be plrttted on the
flood frequency curve. The growth Factors
corresponding ,o these obtained from Tahle 7.1
are given i n the second co I umn. TJre thi rd
colunn has rhe ratios GF(f)/Cf,(tO) and rhe
fourth column gives Qt based on these ratios.
The final column gives Q1 basecl on fhc
zroi,,tlr
factors of column trro and fhe MAF.

Figure 8.1 shows the chree por,Eions oI i. loor]

Irequency curve from the avai ]able data, tlrc
GF(T)/GF(10) rarios and rhe growth facrors
from
Table 7. l. The smoothed tine drawn linking
these segments should be used to estimate the
requlred f.toods. Thus Q(100) ls taken to be
the upper limlt of the extraporated locar data
curve,5l5 r3"-1. Q(500) ana are react
 Extrapolated flood frequency curve for the citarum at Nanjung

t, Transition to gror,vth curve for

large return Feriods

t,
n 600
g
)
2
,
a
t
) 5oo

Based on growth
curves ol Table 7.1
4CO
Based on Qf[l
GF lrol

Qro

Flood frequency curve frorn available data

RETURN PERIOD . YEARS

23 4
Gumbel reduced variate tvl
 fron the upper curve based on the MAF and the
gror+/th factors of Tabl e 7. 1 .
Thesc a re given
in the f inal column below as gg3 m3s_ I
?! ,nd 1(l4C
n"s-'. Floods for intermediate ref urn per.iods
such as Q(200) are read from the smooth
transiti'on curve and is taken to be 655 mJs_ I
in thls case.

Return Growth Factor Itatio

Period Q1 fronqlo Q1 from
T cF (r) GF(T)/Cr(r0) and cF(T)/cr(ro) cr,(T) and
MAF

2 .33 1.0
r0 I ?7 1.0 3t2.o
20 i.59 1.16 362. O
50 1 .95 t.ta
443.0
r00 , aa
l. 65 515.0 (411 I\
200 2.66 (1.e4) (605.0) r7r,q 5\
500 ', 11
(2 .37 ) .4)
(7 3e qe? )
r 000 3 .85 (2.81) (876.7) I04 0
 9. CGIPARISON rITH OTffiR TLOOD ESTIHATION HETHODS USED IN INDONESIA

9.r Introductlon

This chapEer describes a conpari son of di fferent mt-thods

of flood
estlmaElon for ungauged sltes currently used i n indones ia
wi rh tr,{- ,. r cla ta
nethod using the regression equatlon described in this
rnnual. UnfortunaIe1r,
no lndependent statlons were available for a tesE of methods;
of necessity
all acceptable data were used in this study. For the first phase
of the
projecE on Java 1n 1981, eleven trial caEchments
rdere chosen at random before
the analyses colllnenced to give a representative sanple
of catchment a reas fo r
conparison' For the second phase on sumaEra in 1982,
ten carchmenrs were
slrnllarly selected at random before startlng che analyses.

the Rational method and rafional type methods of }felchior,

wedrrwen and
llasper used in the comparison require dal1y rainfalls
of specified rerrrrn
periods. For the first phase of the study on Java,
the only data readiry
available were the maximum, the second highest and the mean
annrAl :n;r;<imrrm of
I day ralnfalrs for al1 rarngauges on Java (r.M.G. Met
nore, 1969). Data From
arl raingauges within each catchment were protted ro form
an average rainfalr
growth curve uslng a Gumbel reduced varlate and
Grlngorten plotrj ng posi tion
(chapter 6)' The average of the highest and
second highesr daiJy rainfall was
plotEed aE the posltion approprlate [o che ]ength
of recorcl . .l']re mcan annuaJ
uaximum daily rainfalr was ploEted at a return
period ot 2.33 years. A
regression line through all these data rras used to
estimate r.rinl..,rrs ,:rf the
required rdturn periods. This proeedure was not that
specified by trie va rious
meEhods but served as the hest approsimation
wi tlr thc dat.r .ve il.rbl,. .

For the sunatra Study in 1982, rainfall yearbooks were

obLained givi,ng
details of annual maxinuru one day rainfalls for each
of years lg5l-1g77
(r'M'G' Yearbooks)' Raingauges on or close
to each of the ten serecteci
catchnents were ltsted from the yearbooks, and for
those with sufficienr
yearly data, annual rnaximun one day rainfalls were
absLracIed. Rainfall
f requency growEh curves were.plotted for
eacll r;r ingarrgt rrsi ng th s;r ,re Gumbel
reduced variate and Gringorten plocting posltion rrcsc
ri bed abovt .rrrl ,,,.,.1-;1 g6-
curves drawn in subjectively for each catchrnent. This
approach shorrlcl vield
better estirnates of the required rainfalls for flood
estlm:Eion Ilran rhe
stnplified method used for Java. rnsufflcient time was
avai-lable bo enable
the earller ralnfall estimates for Java to be re-computed
using tlie rainfall
yearbook data. rt 1s berieved that any inaccuracies
inherent in the
simplif ied
 rt should be noted that most of the nethods descrlbed below arc
varlations on the Rational formula method (section 9.2) and use
the same
ralnfall data as input' Thus they assune that the 500 year flood
is causc-d by
Ehe 500 year ralnfall and only the ralnfall
areal reduction fac[ors and runoff
coefficlents are changed in each method. rt is interesting
to note that
wh1lst the range of growth factors for
Q(500) glven in secrion 7.2 is from
3.27 to 4.01 wlth a rnedlan value of 3.7, t.he equlvalent
rainfalr growth
factors are rather lower. For Java the range over the
eleven catchments
consldered was frorn 2'06 to 3.01 and the medlan is
2.4g. sumatra has a range
of fron l'95 to 4'2, the latter flgure belng something
of an outlier ancl the
nedlan for SumaEra is 2.23.

' Flood frequency grolrth facLors would normally

be higher than the rainfall
growEh factors that effectively produce
the floods because nany of the factors
controlllng the conversion of ralnfall to flood runoff
on a catchment are
relatlvely conscant' rnterception losses of ralnfalr
on vegetation and on t6e
soil surface are largely constant and the Fatt, rrt
inf iltraLiorr of rainfall
lnto the soll also varies only srlghtly f rorn sto rm
L() s torm. Thus Llrt,
Proportlon of storm rainfall remalning for flood generatlon atter
these
relatively constanE losses have been taken off lncreases
as storm uragniEude
increases for higher return periods and flood growth
factors increase more
rapldly than the ralnfall growth factors. ThaE seemlngJ.y
sma11 rainfalls
night produce more extreue f loods is not entirely
suprising si rrce rnany ractors
control the conversion of rainfall to flood runoff.
The effect of antecedent
catchnent c-ondrtlons are of great_,-inportance in
this respec t ; the f rood
produced by a storm will be greatly reduced
if i t f ol lows a .l ong ct ry period or
enhanced if it cones after a period of unusually
wet vreaLher. Suc:h
considera!ions irlustraEe the weakness of mebhods
that ass.rrrr tllat rrrrr'f l
return period equals rainfall return period and
hence the advantage oained by
estlnation meEhods based on f lcjod stat istics. The
f ollowing sec t ions gr.ve a
brief descripEion of each flood estimaEion rnethod.
For a more cletai red
explanaEion the reader is referred Eo the
sources quoEed.

9,2 Ratlonal nethod (MuhadI, lg76)

This version of the standard rational method

which is us,.cl i rr fr,.lnrr,.si,r
is adapted from Japanese practice. The principle.
of tlre rnethod i s to
deterni ne the flood peak Qa (in m3 S_ t. ot
,) return period T years Irom t]tr.
formula

Qr
. r(r)
 wne re
c is a coefficient varying with the nature of
the terrain which was taken
from a table in Muhadi's paper. r(T) is
a ralnfall intensity corresponding to
the T-year re.urn period rainfall for a duration
equal to the time
of concenEration of the catchment. Empirical
formulae are available which
relate r(T) to the strean length, srope and the 1-day
rainfall of r year
return period, R(T); these formulae are derived from
Japanese daIa.

The Ratlonal method 1s usually applied

only to smal l ca tchments , al tlrough
no informeEion is available on the range of
application for the version used
here. An arbitrary upper,.Ijmit of 2000 km2 was
used in this study, although
Ehis rnay be too large for sensible applicat ion
of the Ratlonal method.

9.3 Weduwen Methlrd (M"h.!!. 1976 and r.E.C._rL2f7)

This nethod is essentially a modification oF tht'Rational

n,lLltc,J, and was
developed for condl !lons near Jakarca.
The flood peak Q1 (in m3s -l; ,,F
return period T-yea rs is determined from;

Qr = sFq AREA R (J]

240
where

R(T) is rhe I-day rainfall of return period

T years (mm)
and apq ls a combined areal reduction and runoff coeffici
ent, determined
graphically as a 'f-unction of eatchment area
and sJope f rom a f igure in I.E.C.
r977.

The method is considered applicabre

to catchme.ts with an area of tess
than 100 kn2.

9.4 Melchlor method (Muhadi L976 and I.E ._c. . r2ll)

this nethod ls also developed from the RaEiona.l method,
ancl is suitahl
for use on relatively 1a rge catchments. The i;
f lood pr.air (in Q1 m3"- of
i
reEurn period T years is determlned from:

QT opq AREA R(T)

200
 whe re

R(T) is the l-day rainfall of rerurn period T years (mn)

ls an areal reducElon factor dependenE on

the tlne of concen[ration
of the catchment, tc, (the relatlon is avairable
in taburar form
and tr is deternined fron catchment length
and srope). In fact F
is usually deternlned graphlcally frorn catchme.t
area Lrsin;,a Fisrrre
glven in Muhadi I976.

is a coefficient deEerrnlned froro t.

and the area of an equivalent
ellipse" for the catchmenE by a t rial-and-error
gra phica 1 method.

is a runoff coefficient whlch is arbltrarlly

selected Irom the range
0.42 < c < 0.75. Melchlor suggested an
averagg value of 0.52 but
current practice is to use higher values
in the ranpe 0.6 ro 0.75.
The arbiErary nature of a means that
the estimates for Q1 determinecl bv this
nethod musE be regarded as app.roxlmate.
The meEhod is considerecl applicable
to caEchments wiEh areas greaEer than lO0
kn2, and the graphs requirerj are
avallab1e only for equivalent ellipses
smaller than I0,000 km2, catchmenE
areas smaller than-f200 kn2, and times of
concenEration less than 20 hours.
A noajor eonstralnt on the nethod is that
iE can only be applied to catchments
having a mainstream length of ress than
about 150 km. Because of this many
long narrow catchuenEs having'areas of only
4000 km2 o. ,o may weLr have
equivalent ellipse areas outside the range
of Ihe graphs, eg. catchrents ir5.
I18, 20L and 707.

e.5 rt3lpgle$lgglMuhadi. 1e76)

This is another modified Rational method,

which Iq very simj lar- in
concept to the Melchior meEhod. The
flood peak Qr (i nm ll s ) or retrrrn
perlod T years is deternined fron:

Qr cFq AREA R(r)

where

R(T)is the l-day rainfal I of return period T years (rnm

)
ais a runoff coefficlent deEermined as a funcIion
of AREA
D

)
rs an areal reduction factor
deternrned as a funcIion
) tine of concentratlon of the of tlre
catchment, tc, (which 1n
determined fron catchment turn is
length and slope) and the
) "equl valent area of an
elllpse ,,
is a discharge coefflcient
determined as a funcEion
) of concentration' Different of the time
functlonal forms are used
ranges of tc. for different
)
applicable ro carchments whose
) I:.r;;T:":: rlnes of concenrrarion are less

9.6 Peterson merhod (I.E.C.

l9D)
Thls method was developed
as part of the Sederhana Irrigation
and Land Developnent project Rec lamation
for application throughout
regression equations rdere fndonesia. Mut ti p1e
obtained fron whlch the
for return periods T = Z,5rlO,ZS flood peak Q1 (in m3"-l)
yu.." can be estinated.
These eqtrations are:
Qz= 0.00000143 anea0 .e64AAR r -5e
Qs= 0. ooooot74 AREAo . e5oAAR r. 72
Qlo = 0.00ooo189 AREA0
.,e42AAR r .73
Q25 = 0. 00000159 AREA0. 9,r,+MR L 77

where

is the
AAR r

rhe comparison
"::i;:T;',::";::..:::: ::::::: i:,];" :::",::":;i,:;":.:,
these equatlons were deri
ved from ca[chments with
to 414 kn2 and with mean annual aroas i rr tlre rii,gr. O.43 km2
average precipltation in
5226 mn. They should [he r,rngt, Iap? 6rll [,r
nor be used for catchments
outside these ranges. in which tlrt' values are

9 .7 Conparlson_gl rg!_g.l rs

For each of the eleven catchuents

on Java and ten catclrmencs
flood estimates were derived on Sumatra,
uslng the methods just described
and also fron Ehe regression in thls chapter
equatron and flood frequency
developed durlng thls growIh curve
current study. ?ables 9. I
Ehls comparlson for Lrre and 9.2 show the resul[s of
the uean
nean annual
ann f lood' MAF' and for
and 500 years reLurn periods of l0
 For each catchment, ari- estlmrte of [he true MAF is available from the
observed flow records aE rhe gauging statton. Fo r all catchmrrrts L..r.r.PL
nunber 433 In Sumatra an estimate of Qi6 is also available from the observed
flow records although the estinate may often lack prec i si on due io tht shorE
flow records available- A measure of the success wlth whlch rhe observed MAF
or QIO is predicted by each uethod is provlded by the root mean square error
(RMS error), where:

1 n predlcted MAFl- observed MAFl

RMS error (7") = x x 100%
n i=1 ohserved MAFi

Irlhere n is the number of observa_tions. A low value of RMS error indicates

good agreement between the predlction method and the observed values and vice
versa. It should be noted that the RMS error places greater emphasis on
overpredicEion comPared wlth underpredlctlon slnce an underprediction can only
be up to 100% less Ehan Ehe observed value, whereas an overprediction uay be
several hundred pE-r cent great€r.

Two RMS errors are given on Tables 9.1 to 9.4, tht first being for a1l
catchments to which each nethod was applied. Ttrus for J:rva in T,ible 9.1, the
Weduwen method was only applicable on Ehree catchmcnts, numbcrs 25, 17 ancl
29 spaces. The RMS error has been conputed for
-just [hese three catchmenls
given as Ehe first RMS error in column (a) of the tab1e. The Indonesj.an Flood
Studies ReporE (FSR) method of this reporE was applicable ro all eleven
catchrnents used in the comparison. Hence the RMS error given in c6lumn (a),rF
Ehe Eable for this method is for eleven catchments and is not direcrlrv
comparable wlth thaE for Ehe Weduwen meEhod for example. fn order Eo compare
Ehe neEhod of thls report wich oEher uethods directly, a second RMS error nas
coDpuEed for the Indonesian FSR neEhocl using onlv those catchments common to
each method in turn.Thus as a comparison with the Weduwen method, the RMS
error has been computed for the Indonesian FSR rnethod for the three common
caEchnents,25,27 and 29 and this is given 1n column (b) of Table 9.L These
second RMS error estlmaEes should provide the best comparlson of the flood
estimqtion methods pregented in this report and others commonly used in
Indonesia. Tables 9.3 and 9.4 summarise Ehe cornparisons for the MAF and Q16
resPectively by glving rhe RMS errors for each tnethod for Java and Sumatra
lndependently and also conblned.
 rE ls apparent that the methods of this current report
glve consistently
better estluates of 'l'{AF and Qro than any other nethod, wlth
the excep[ion of
the Peterson nethod whlch is a sirollar regression noder.
However, the
Peterson nodel ls only applicable for catchnent
areas of up to 414 km- and for
return periods up to 25 yea rs ' This report provides
a more compre6ens i ,,,e set
of neEhods applicable for catchment areas up
to 20,000 km2 and for rerurn
perlods up to 1000 years. 0f Ehe other methods,
the Metchror approach for
caEchments Sreater than lo0 km2 produces
reasonable results on the whole as
does the llasper roeEhod despiEe Ehe latEer,s very
high RMS error for Java.
Thts RMS error is dominated by the Hasper urethod's
gross overpredic tion of l,{AF
and Q19 on the tl{o very sua1l cd'Echments,
27 and 29 . rf these are exc l uded,
the RMS error drops to only 77.2"1 for Java and 76.6%
overall. rt seems that
lhe meEhod should perhaps not be appried to verv
snal1 catchme.nrs.

Ttre Weduwen,mathod doesnot perform partlcularly well and appears to

consisEently overestitrate floods while the Ratlonar-
nethod grossly
overestlmates floods. Both methods perform poorly
for the very sma I I
catchrnents, 27 and 29. Neither of these nethods
seems to provide a viable
alternaEive Eo Ehe approach of thrs currenE reDorr.

0vera11, the authors berieve thaE the range

of frood estimation
technlques described in thls currenE report provlde
the most reliable flood
estirnaEes for Java and Sunatra. The methods
may well be applicable elsewhere
in rndonesla if used with care and results checked
against local da[a as
described in Chapter g.

A second and indepenclent assessment of the methods

of t.his r.pi)rt ls
shown in Figure 9'l which shows the rnaximun
recorded floods for Java and
sumatra pl0cted against catchment area (Binnle
and partners, iggo). Also
shown on rhis figure are Ehree estimators
of Q5o0 against catchnent area
derived from the regression equation of section
5.2 and rlre appropriate
nultiplier for Q500 on area given in Table 7.1.
The upper r_ine was cj t,rived
assumlng the highesc ciiobinatton of ApBAR,
srMS and LAKE encountered on a
caEchnent 1n this study and the lowermost line
was derived assuming tht, lowest
encountered corobination of the same paraneters.
The centrar rine was
calculated using average values of APBAR, SIMS
and LAKE.

The 500 year flgure adopted here is perhaps 1 ndicative of rhe likely
rerurn perlod of che hlghest recorded f100ds
shown on Figure f. i. Many of
these nay in fact be nuch nore commonplace events,
having return periods of
 UJ
Y

LIJ ;Y
yI J o
o
\9q z o
\ $cr
'u, \ o* a
$
o \
-t',.o\
it
C)
a
=
.E
o
.\lJ.o
.\..2 ;s
o .. \ __69 t,|-to
E \r--
.l0i I i Ei 6
.9 co
X ut,
- ''\
'.\v=
(5
.. t '\.. 8 o==
06f
V,

'
ct .\..*
o \.\.n
1\\o *+ \.
F

I
oa
xs =
c r-o-o = o
o
.9 \' . '.
\ J b o
(t, \' lJ.Hs
@
o r\'\
\\ t'-S
o)
o
o9
o
|f)lcc
A
1l tlotl
>; /\
o v/
(Ji ol
($ /t E
UJ\
o 9o-:
deo
l|
(l
E
F
E P9 g
(
cn =s
d1 ao
ro +. c
o
6 6E
G' @

$ I C

f o
tc
c
uJ
z
F
U'
E
o € ?Y
o Lr,l ,O J- \
z ,/\
=3
gt@
t!a
//\
o :F (r
E
o
o 8X
a') < ;F
x=att
o 9
lr(69
Htr Eo.
o 9P a)E
E X@
3 ut
et
e Pr o)o .G
5:
.E
x
o
8
ut(\lF
c
t6
93 Iz
o
= = ?8 otr
o
N(,
r =
=oll
)<FF
ll)Z
ogo
o)
o = ;fi o
J F
(r,
ct) t!

gI
TL

saur poot{
only 50 to 100 years buE some at least will be very rare events. For the
average combination of APBAR, SIMS and LAKE encountered in this study, the
estinated Qlgg and Qtooo floods are arso shown on Figure 9.r. rt is
apParent that the variations ln floods of various return periods for any
partlcular-tyPe of catchrnent are".significantly lower than variations of say
Q5gg between different cypes of catchmenE. The 100 year flood for a
catchnent havlng Ehe average combinaEion of APBAR, sIMS and LAKE may be over
twice as large as the 500 or 10oo year flood for catchments with low
conbinatibns of the sane characEeristics.

It should be noted that there is no return period attached to the rnaximum

floods shown in Figure 9.1 an'd sorne of the f loods plotted are of cloubtf ul
accuracy. Hence some of che large scatler of points on the graph will
undoubtedly be due to errors in the estlmation of the magnltudes of these
floods' However, as has been emphasised in this report, tht-re ar'.,arso
stgnificant errors and uncertainties in the estimarion of the Q500 ii.cs
drawn on Flgure 9.1 usirrg Ehe methods presented in this repor!. for any
catchment there wilr be an error associated wiEh the esEimags ef rhe MAF.
wheEher this estimnte comes from recorded flood data using the poT or annrral
maxlnum flood series, or from the regression equation as
is the case for
Figure 9'l' The flood frequency growth factors of Table 7.1 used ro conver[
the MAF estLmate to Qr also have errors of estimatlon associated wjth them
and consequently the plotEed lines are only a best estirnate of
Q56g in each
case shown' However Figure 9.1 demonstrares that the Flood studv meIhod doc.s
produce reasonable ans!/ers over a wide range of catchment areas and r*Des.
 I6 E!t
E€l
e.:cn":n
Es$sGR
F6NO9{
6tsO60d
66€tsts@

Eo?f_
E
*T
3 ll ? @066 N
siil
tsis909
*$tx$gs -g^Fs€
s .938t €N{6Q6

E-
-l€': qa egqa 69 q q.:
'ttE qF'RhnE ey SxF:
-
s{ rxg6
€6.

"fi8-
Is HE, , tl 8S, , , , '

"lls !g' ,
$R r r r r r

nlle
B.n
:;F'EI; ;:;'a:; F€
b
o3;:
b<-
33:';;3 53:';;: i:':;
!

.ii* i$'iF3'EsFxi'
o €9
E I xSS

$ €9- n\-: .:9a tn n:n g: q

n.!i.r
d6
ilF;'ES! R:3'SnP 56 g8 9F I NN

E
i pq
Np+g ':': -: n': q ,'r

H FF'iga' b9'pF!' n, *dl

,=
td-
g

B
-g;s E8'!e3' Ri'El$' gB 1-;rc9
56o NN dN6

. F- NN eRq- I lq
^itr$ sF'Eida
i

E
Sih€

h vn
*d n*diri ;'sst
q E9^
FI-g3H ;3'il;',
€
:q'gE
sN
" 3'bg'
I on i€ 6 o-

a
v1

r l"t!
!eu

I
;
i rn ?il ;
I
0,
F{
!ifi sriiilrt eliltrll girirr
 a6€ooe Soooo-
;i!r
u?o
gi$tieg 6Fiiriid
I
* 8g
-:9l1gj
I; tie
Eeil.
!9qq.:i
36jR33 ;Eg3Sn

fiIn g: ":9:
sJqt rn 8P ' €5; I
ER'*rF' $'ilR:
ro
-t
iE-
u g$t
idniiN€
nO1Oa66 eA-o6{
gdiadig: ssdiri

:9 | qn 6 GO N-
?= sbP I
$d'$du' :? ile
'lg' {g N-6
^6 --i t

9g.
e!$9 ig' ' ' ' ' 6i orll

dl
il a -:a q o{ o6
*l!B It *:E
' =$g' I j,>rjRN:
r, 6 snc
=n E6 I'C:g
t9
?l

sl aila
gB
gi'dge'
OC)

i.iv $x5 .X'l3o'

6-N
-s 6xR *."8*
i,R:H
O

P
F
Hl
nt <ri
o 4to :::i11tq 99q1lq
t{l
HI
:fPd
r{= sEFRX:E 9E3FSnp ;
EI J I
I
Yg- :1 I q9n1 Q €6O6

sp'ol:l
6 €N9

Sqisi s; i:tF R IgSR E

t{- rr{
P
B

,fIfr I
-: ^ r
$i{ r f, r r RR r

r
trtl
$r,!,
T
E

5p E

3l qll
-:t|g €€
'Jgir r r r r 3
{si' ' ' ', a I

T: N:

-{,
I
!
i:"""
g
I I &

6l
iH- -s tE Yn xE
-1515*tr8 gllad o
Sal:9:lXFd
n
I
Or
l{
F
!irr Sllisiti 3i;isEi HEll *-J pt
^lft{rr
vFt
 z c\] ^
.u.)
rr) @^
.@
\l
\Ov
\O
ca
F\ -.a
.i.v
o \I) v l-\
co U1 LA
tr

F]

z f\ G-
ca^ ca
H \O v
n ica6
ts
ral -J ta -f Ol-
-Jv
d

fa vl O^
z
g
cO^ \? r\@
. c!
.o\ o\
\O
ca i Ir\
-Ot f\ v lfl C{v rn
H

E
&
r
L|,lA \O
F . \o. C-..1
.-j ^
Clv -t ch- n
zH ,J
r
E
r

t<z
E3 \o.^ -$
H
FJ
@^
|')ca \t ic!
o\
\t
.^
ra ul.
r^
c\
r \ov \o Lnv cn O\v vl
= 3 \t
z
F
E
H
F
z
l:]ci
z<^ @^ ?.
\lN
coi cO+ !nv
f z $v

r
Fr
:
9
al -
a.)
tJU)
,.o ! E o
E 'rrtrq.) AF
TJ
+J !,
h .-i E $ q
CN U)E O(J
!roEc.l tr
tr!u &u TJ
trro c < 6rfi
cJaJtr(d oqJ O d Ei nr
& E-rOcEro
E A
E.'{ &e F -aMA
ld -CFr
OA-C flc -cFi a E r)4 Cn E
!r oo. r!O E !o.tlro
(h 4 Ao o >J lrct(J5 rlAc)
p.. )rl o (UE o O cucul a) - c
q^ o .'-{ o osL a0!
(4$q Fli 1J
F{ O o Tt .-ro a Fi ! .'-l O
v 4l o o -l O Or! 2 ! o tt'L
Ol -l-C,r U d F{ -C qJ < $! o ot
1 4 i ouo crr F rdUtr!
o\ <ea
t{O>l
<i cr
rrErr.otqt oo < IEt oo
.O H < !cJ oo=
z&& <l o a) Htr ! p l-r ObE!L)
El tu.c4 Otrtr< fttrltE
-l o E
u) fi"E H O F) -H
FJ irrl o
 z -s^
.lr) s \a !^ O\
.@
ln v Cf) cn^ @ Lrl v o,
ca ca \on
F

Fl C\
\T
z @^
l^@ Ol (:t \o^ $
c.) CC C! \.o
\O v t-\ -$^ ca
F c\ {'

\t@ -ir
(7)
.\o
C.) O'
rn
t-
-t^ \.o ca+r\
\C F- !r\ trtv \c
O\v

r^
H .i'^
.\O
c! r\@ \c^
.6j
a1 .if v O\ \o^ -) O- u'l
F1 F\ \O O\o .a O\v
frJ
H
4
=

Fr
a
EjZ
El lrJ
>3 cnl/)
. ^
Nca F\ --F^ O \a!n o
Fl e{v @ t\ C! a.l \:lv \.c
3 \o

z.
H

!r1 g @^
Fr ZLh .O
O frr @* ich \ov
\ov
z
H
v)
o
F q)
zrll 9 a
t\ (! qJ
!o
e
6F A.\
lJtrO !u
..1 F o o
o
rJOErr
ahE (nO
trcru &,i C\, tr < E
OOCj6 Oqiod E nr
E-{0J&O
sA E rJ7
E*
.c'1
e
q A F e.aeC
O A.C rl G E oFi a E
'LJ Ao
oc\hoE !Q.F.O
cEo F-l !JO..r- $ao
E d(0! z $Atr /)
Fl o $ (! E o O$li
(J
Fll
4l F{o'9 '.r
(',oo 'O.FiOA
r-t O Oq{ Z
F.lrO
F+ O
.- C
Oqi
al rrtqr O d -C A 1.
.if (0!ocl]F '-{ 4-

Ot
<l.l
>l
o
fr F l. o t+{
"O
o <E dtJq
lrEE!
tr L
oo < lrq
!Eli!
OOC
<l
-)l oocS
t4.c'o H! (A
otrL< H F)
LHE
f&-C O
F1 IJE.)
I
I ANNEX A. STATION ASSESSTTENT

) A. I Introduction

) This annex considers the factors affectlng the

useful ness oi clata
collected at gauging stations and shour d be considered
an essenrial
) first step in starting a design flood estlmatlon,
sec[ion A.2
descrlbes offlce procedures and section A.3, fierd procedures.
) sectlon A.4 glves details of how these were inplemented
in the currenr
s tudy
)
A.2 Office procedure
)--_
The first
stages of the flood estimation procedure
shoul6 i n'ol.r-,
the conpllation of a list of gauging stations at,
and crose to. the
site of interest. The llst shourd incrude not
onJ.y stations near ro
the site wlthin the sane caEchnent but arso those
in neighbourlng
catchBents' The primary source of lnforuration for
this list should be
DPMA in Bandung who are responsible
for the nationar hydrometric
network. other bodies do operaEe gauglng stations.
Some oI these are
given at the foot of Table A.3.

The reason for collecting daEa from as manv s()rrrcc'S

as pOsSibl
1s to ensure Ehat a1l relevant inforuraEion i s
considered i n produc i nc
a balanced flood estimate.

Before site visits ir is advisabl e to check

the data whit:ir arc
vallable for each gauging station. In particuJar:

(1) Proxinity to Ehe sire of interesr.

(2) Type of statlon (contlnuously recording
or sraff -arrse o-l v)
(3) How many years of data available.
(4) The quality of the rating curve for
flood f1ows.

Thls lnforuratlon uray be used in assessing the

relatlve usefulness
of each station in" the flood estirnation procedure.
A preliminary 1ook
at the data avallable may raise quesLions which
can be answered duri ng
vislEs

Field visits nay now be updertaken as outllned

in secrion A.3
 After fleld visits and the identification
of useful staIions
peak stages should be absEracted
for por or annual nrax imtrm sc.ries
analysis fron the orlgrnal charts
1f the staIion is arrL()m,ari(-. .,:{:.rm
the observerts flerd sheet if there
is only a st.3ff g,ruge. Data frorn
a secondary source (eg year books),
t useful lnfornation such as annotation
a large flood (benjtr besar) is
is prone to error. Furthermore
indicaEing a sticking froar or
often only avairabre trom the oriein:r
source
D

t Ratlng curves should be developed

addltlonal flood dlscharge neasurements
preferably wlth a programme oF
to reduce the degree of rating
) extrapolation. Guidellnes for ratlng
curve development are gi ven in
Annex B.

t A.3 Station visit procedure

)
Prelimlnary site visits shourd
involve inspection of gauging
) stations and discussion of gauging
pracEice and historic floods
Ehe observer and 10ca1 peopre wi.th
respectlvely. This will lead
understanding of the relaEive to an
accuracy of stati ons and any
loca1 factors such as the depth special
and extent of flooding and
locatlon of any historic f1ood. the
round Eo be . """;;i-;;; ;;".";;,ii"j"j"r;:i:'" :,=:j:::".;,::::.::,,
reEurn to the offlce.

The condition of the equipmenI

at rhe statirrn sire-ul,r be rrr)f1r,.].
rn particular that the staff gauge
is tirmly fixed, its rnarkings
legible (includlng netre narks), and
that the currenr reading :ttret,s
Ir'ziEh the ehart readlng
if Ehe station is auf 6mn;is. For
statlons a check of the observer,s non_arrtomat ic
notebook with tho crrr.r..,rrL sL;rif
gauge reading helps to
assess the observer,s cli I i r,,,nce.
worthl.Jhile looking around for Ir r s
other sraFf gauges i n tlrn ,,ir: iniL-,,,
Ehe statlon and if found r;i
take readings on both old and new.
staf f gauges are replaced Of[en old
by new ones or an autorna t ic s t.r
installed nearby r4rlth a dlf f erent tion
datum. Noting this i n the f it:Id
save problems laCer in the may
office.

An assessnent of the like1y behaviour of the station

flooding should be nade. drrrIng
tf rhe level 0f rhe maximum flood
has been
 abstracted from Ehe data in
time for the field visit
visuallsed by reference to the this may be
staff gauge. If overbank flow
rhe deprh and wldrh of rhe occurs
f rood pJaln should
::j::"::u"t, be

.n" o";";.""" T; ;j:;::lJ";;:,::,;;'" :":;:.;:":l::.

bridge abutments, which nay
confine flood f1ows, imnrrliately
:: ::j::":" :i
downstram
often local people will
:: i::ji,itri;". rerl of exrenr and rrequency

Special note should be made

of flood mark:s near the station
Ehese should be levelled and
into the station datum and
occurrence deternlned. when the year of
devel0plng the rating curve
statlon lt is useful to know for the
the type and shape of the
contror of the statlon. stable hvrjraullc
bed material, perhaps large
or a rock bar dot/nstream indlcate boulder"
a good stable control. Bridgc
downstream usually have pi(.r:i
Ehe same eFfecc. poor
found in rivers with unstable control is usually
bed materlal such as sand
This may forn shoals in the and graveJ..
rlver which reallgn after
station with a good contro-r t100ding. A
should have .".Jonrb1y
" stable ratins.
A.4

The purpose of the staElon

selection procedure described
rras to ensure that onl.y bel0w
data f rom the mosr rer iable
entered into the analysis gauging sEac rons
for thls pro-iect. Thr-s and
visit procedure described the stat ion
in Section A.3 was cn(
parrs or rhe.srudy. From
sourcc,s ar Dpn^ ,.0r";::
stations on Java and Sumatra ii: _:::"t;':,;::;,
rder
staEions were idenriried,
bur ;;r"t"::;::'"""^"::::]":jr:;.::.j:::
eliminating all stations
with shorc records, rhose r,
doninated by an upstrean affecred by ,;".:
lake or reservoir, stage
those of obscure origin, onry starions and
and those wich very poor
ratInss.
The procedure adopted
for fleld visits has been
in SecEion A.3 with the discrrsseci
necessary of flce work describecl
4.2. in Section

The last stage of Ehe

station selection procedure
studylng al1 the available invol ved
inforuation for each station;
the rati ns
 ZU\11
td,.n!!! [ (rr a1rL' (!
'al O I'J f.J N tJ lJ lJ rJ lJ N tJ c 4-
*l .r*::ruu)-l qu oro to rtr (I (rl tJ - l! (o -l D UI g'r,rJe
@ Co \ Itr (Jl ! ral l',J - O @@ \l c| U|l'il t..) F ?u@-lrnut
oa-,
E E7 rrz ooo oxo o .)ooo c)oor)ooo oo a)oc)oooo('roc)o
c) o o o o
r '_'_:-::: OQ
tl i, .- :"- :
a//'
'_'_y'_ tf; U€
U t 3 --Fc
o o r
oabUl
c ! I O F e(. I C o G
d h - 3 3
! ! c o !
x o - -- - ! d
3 3 tio
g g
o DtC0!vloEo o ; d d

t rFfq
'rrntart I,Fd
qJ €E tU' €rN'
tri tr
oo
\laalld3aCl)oollIJfl
U' 11.It(l)-o C E tu D
!q<"!r)-lrlAooo
d tr s E I - - -- ooo F C
gO-XFX
;
:r D -.3 l1 €.
:r
xra 6lJl @
0 4
C aCCadgt0gcC
a 115
r t r
- I
.t
-@t t ll lu 1 .rOlO tr-'I C trlDo<!! 06 0 @ 'O
o 'o Ho
l OO N. o@ p n oo!38-qo:ld dd 'l o!s 0! g o d
--l (|{n tn< (n €€ l o o<,o doq !C!ddd d t o O oO c o -n d <
olrl'rrn UC ot oo .o EO -b.o-Itdo-ord-d or
o d d d Er). l d 6ddd-
6
:' tl oo
a-EX)-r)
@ 1t{o f-
r- o'ou, Z
OOO < FrOc o P 3
OP3OO74@@s
c) rri q, , |ro, ! 3 oo '.o _o
{r
'Ir -.
d,r
do
dQ
dtr
T,
I tr
'J
dl

txld
hhq'D qOCOsoOr-F!6431
oso-€''orDtrut'ldcooc
muo14d@-l!g!-OoC! €X
; c i ! - - o
€
3
F.3F
r Q
u
!D rU'rC
fr 0 O 1f
ft
Cr
- C
l

l l
a-c - 33-t
f zt
o€!d
) - . . rT.r!"
d - F O -o 3 C I f, O- ! - oouuctlro.-l
U
:. o Il p
' r O'<aItrl;
oi,rgo- a MT fol I Oo o - 0:t,o O - I I . e I D I n o l,o
o ,o o o I .-
a,j
ry'iC oh u
o
o ory
.C tl
wl
Ot
FC)
l0
@ a<
l @
,o d.
cd
-g
.-..utfo-0-.ooc--o
d- .
I -o
trXO.
; 6c 'o l
IUd6'0ryi
ooo
U
uI'o
l(rd tO o o3 0'O
f 9r to'O l-
3c
co >d. . -,o. t. :.
C

: . : : : -: :: : - : :. - lF3: 3 ::::::.:.::
€ tr
d
() l 0 j
-a
\- ol 'o- l 'ul ,o D. . . ' _ 16 A. ! o
c., o- :: : : : : -'.'. -: - : : :: : ::': !
o- =
@
:::;
F: l

o
o

d'ror4qr!voo,@o!! doDcooooooooqouoqooooog!!ooddro!Dnr D

rar FF.IFI! {jrr, 1..! d-j

a I u,.r ti it --e!+
-.(. r.Jto oD.) lri,:'ilr0o- g'(rt!- lJociufroo6&\lcc|buottlrooNlilufi- ''
,--
ln ll@*
m, - * tDr.l -- U I.J ! hJ l- tlUl-rTr { DO'al (rrN- (, (!.l.tr ('|- N(a'N ! tJ -A'-
r,J
* 0 ! op . . Ot N-
@N -!(I€"1 - OUI5' @ tOOrfD oO r^OF (n Ur! UItjS
01 l.}
'IOc,o@fij:1 r!o !1lurLrLimd.l-!{Ii(tl
o Ul- D\D - O \-- tJUUn.
1. - frrltlulto!-uor(illJoor@-ootI|N!..'.1 !(!r-!uo
U'-l '.|,
=
,. -. .
--ro(our- @. fJ\I. - o- - u- -*'l'\1'.1- . . F\.1.. -.'.1- -
lh ,t l o- - (,tJ. rOl.)o\-{!- @}Jp 6!Ul(,D-tJ@N!N@o\b'{soor.)CoOhJ- o,*
a+
qr @co*Lrl- @F FN -F3\-{e-lutrF}lJ-
-1@-*6\iJ0! (to F ! @m0-lOJ-orol\(r|!\Ul(r@r!-hJNu(r)[o
F. t,l 'l o U O \l U ! o ! U \ m€ --!\(trN
I O - ! t F N 3 o rrr o O'-l -o (d UN O N\ A| iJ@ - (o ! \. grN"''- S t{ N O (rl O L'
r'.rl,ll.J1l|.rol.JoNol@toOFF\l$NN' . - - a-.tcDo! UIl.o@!!oNo()m}.N'. . - (o.A(n'' t\r\lhl
. . .
o . - ur- - (olooNm(J|ol
aQ l r.toa ! @ -. o o ! o -o 0||uu!! ! arrtJ- c'larts uron.Jroo ourol'{ Fo(! ooool'1
, ,,ij rr,,, lj l:j [i 3r 3 !j [-,1 - q.'1o,^Tr,'r ! rJN-Ioo.u(! u @rarI6^r- - urur o-uT,o'r' 3
a
o. tr ror.r',f,r9p:::l.r'j'PP-.r?:3P?::,'9,1,91?y:'::P93::lu:9:r9:311.r91.'l.'3:' -
i'tLs-r,Jou!*(!lut oDN*qo!uiraroulu!(rruom{rr-}fi(o-\N'|!0!ulotruulul },lm!
tJ lJ D Ul- ti-lO o p tcro .. F FF F @- NO*F \l FtO- (n \F \l ol F F! \lh)Ult.J 't:
F' : (n.
. (t.
!\Jl
. @frr(ai!'tr.
! mUlr!FcD
. Ln. oO'Or!O'UINFFO\loF'!"
(!. -. (r-Fo- - - \l-
Or-'{-'lo''6tt'Lr(nf.l"
u. \6- |'lr!
Ol f.l-
- tJot-
tJ"'
-{hlt!
!
c
!.. N. @o- cn
o- Ln.ttJa!:lrrrmoulfj!idfJhJh)u.!r.J- - - - o!o!ulrnougi(Borto'nulJq0N\tl!!Lnrrr.!raro* u
c+
!*.rto@6rJ!*-!tJr(oUn@i!-tJFeFetluNN-flr,jN-rrr--tJF>(,tJ(o6ulu'Jr(o'JFF
o {iFd- -lo-. LnOf.JtJ'{@Ul&UlU(noFi(ar-. OOr{O.}t(DO*l'Ar-Otl|l,o. .l'rt\OO'rr
-.. r.,li!'l'.1 t\)q*,) . TOOCDOO- -. !'. "'
i,/- ! lgr :tlJ \lo. _t _i. nJo(rotJ@. o3'{- 'arur
!-itlorfi Ln('cooorra)(Do,E rrt.jorr,,.d!d 3
rnT =
U,- f.Jr.j 'r' .. tJ N'11 lJ'l iJ !' -.1 * m rj Ol -J
? rJ! FurlForTrF. J: \l1l F F O\N\@&ljl-llglioc4'',t..
,.i(o!fJcoulfJ6tsF!'.lo- o1.''(o'rrt.:l! 'TrF
{rto-l'J(n-lJ\@(lLt-\lF(,,loa.u)-. ' - tJolm
Y o('lrtlJtul toooul!l .. (r . - N{nfJA(Dr!(aiNo!!@- - - - (o'.looooo. 6
(n .t-tu(rrto-ulN- oo
'' lJ !'{'otJ''' 'OOoOOOOOOa-
o o o,J o o o o o (o o o Ln 0 o ol r! o LI ul hJ ul o .! o (r \l N o o o o D o o o o o o o o o o o o o o .-' o
o i-
I
-o
d'l I
> () (J <) c- o a o o.o o g o o o o o oe o o o o o o-b-o o o o o o oo o oo o o o o o o o e o o a o o l
t.
* .
ID c- o c) oo oo o o o o oo ooooo oo oo ogo o oooo ooo o ooo o o oo o oo oooo a
fj
rrrrJ!tJrJ f$\l ts N* rals-{---@F1r'.tlJForu A!
-ljo '!F rI'&UlO\@GO.:l!dU-(rl
!&r.J*fJooo'l
,) (o lJ qt o
o_1 -r.DUJOOF. u N a O a o Q !ts {O -fiOO.rr
! } O \O alo o O Oi Or Ul' Ul -''' ! 3'
.u- - @ fr '{ o to O 0 } o &
. \t*!t.l- - .
}-'
. \1 . - . o-
N'
NNaoooD.
[, Ul
-' o_l OOOO!hILTAOOaJOOOOAOLn!(OOOOOIOOoOOOO !noOOO'Oo

* f.l t.J
:- :t :-' :) : ?9 I: 9993 I9II 999 I 3Cr Io Ia 3o 9o :',o Io lJm 9o Io IooI ?9? I 9I 9? I :' I 9 :- :: l :t!
o l.r - o o o o Q o o O o o o a o o o o o o o o O a o o O o o c c' o c' o'-' o D

.T
,r,'tlI **-
-\t|n! \lm N
.'.Jo'.tJU'tDL- -ooooooooooQooQoooooooo6ooooooooo'-_ _ ' ooo
' .Jt - O) qr- . .
o. O o. . O o o cl c' o c, o c, O c'o o o oo o o o o o Ooo ooo o UIO o o o oo o o o c'o o.).
. :l f.-) r1rr.r rJ fJ r.r rJ r1r r.l Nf, ri rJ . lJ! ! Gt (DOl (ltqlhJ 5 0Jct(r)Nt (])(r)tl N(!Nq NNNN NQAttat.!illJf.l
@ trj r Nj - - - r! - ! Nl o o ! N N (o o o ! (t M @(n@ |o o N (r! - t! 9l NltJ ! Iqqu
N cl \ q o \ D ! 5 \q ! N qr
€urn u-r rr., ! rJ N s ocrr i:ri-ltou'qo eoNld\o\€o* o)'or-l o rarb clu !!4o 3l>
:: y, 9 ? :, 9 a i; 9 : II sy : y: : : -@
:' 1' : lJ 9 : y ? :i
PP II :)II:
9P-3 I 9P : 99y9
t
a
(o .r @ co @ -.r 6o I I i
o ., o * -,jlr. o 3 i' I : : : : . ; [ ; e e ; ; ii ; e [ ]! re @ co e i ; F 3 i : . .y
g l. ': . : r r::r
c)
99 '9:'': yy i" 9"19'a l'r 39| 9 PllJ:O Fl'9
o o O O oO O O Oo O O O O O OO O O OO OOO
9::!"
O O OO O'
y I )'iJlJ 9I
O O O O O'1
=nr
T
o,:, o o o o o o o o o
 i.,f. :o (o t! (o.. ", t..t- i.) -J -I'I ! D U Ul (, ul u (n lrl dl ! ! rftalNjtO-
! ! J. ! ! ]r l! ur.r, tlr [l rai (! (al [r (! fJ t! fJ fJ fJ tJ f,J l,J r,J t.) t'J tJ .- zLtn
rrl,rlt,JfJr..r r..t - * * OO C - O O O- tJtJtJ - !! ! aart\l - i F F F D ! - F - oOO -OO OlJe e e Oo
U *}tJOtJarlJ-eeD}rritJ-Orll
rI,L l,nl-,,,rr,-tr r,1,,-r,J1{,.1.-orr,t\t*rrl0!ht*i-lrtrqr -.! 6CorI(.0&lllf LnlJ-tr(!\l')U,Ll
o<2
U) Lt a > 6 rll |I| |xl F-i rI!'n trt c| lIl |Il I:d lI| EI uJtotrlmmrntrr rn |I! tr| |I| F > rrt mF 6(Jn rn qi U](n cr| ul - F I : --
v' dGdadd' oo'''' ll'l'! la
:rl .Itctrit
'3 r >rtrl!> (n
:
-l-l3f-JICr-Xri
oE-oooood-occc>t3
f u. o 3 ? 1'1 ', I t I ! q 3'o 3 D
'oinorr-tii
oEcdOd-
):3drn:lh1 o
: c s o - -D*>O-6-
1< -:c c-OOoo
rrjUro - O o ('|LL -E > J
-3C o4 G n!
.-flotO:Y -1C
O
l ':
r 'r?t. r
u!o!ad -.oo ttD f, F-tuc.OO:--eaPDU33CO3
T- -CoA:ro € COUsln:iP
!oot ! a.13ry9 f c f I ry I c,
l ' 'r C U U'o ! ! d o d o'- I d 0@lO- oc,odo ,OOti lf,OlO IOOt o
t-f d dltr o *-'l :f f :tc o ! o d I a I ! l
:
--/a@
I '' - DdTw I O 0 P€ o - d ! -o O d d, d.d{D @ c d o D >'O d'O
d<<O
t)rru 3 3Cltryq- Pd0odd, xd4il p n (n o o f I!
drr3 CuCdd -J'1 l o d a @ ot d c)-
-rb-ocricld--tn il rIl r-(I' I 3 atrId-ECPZTXI{to I'd
I !, I F,o -Of aa d a tnlC oCrC-l r0 OlUroOO>!OUqP=
}\a c ry€ d gl h Ead i 4occ t
-l a O'1 f,3 - - D
l: o c ! c ru I t c o c _ * t o .r F P do d d c o q xo.-
tu d I C o C t --lP - O'U a O ltatro3o-r:aoc, ,ory
.' ,r.olrC-l -ry ult- D E N <UT 'oCdL ?o,rao.o!a l'3ddf X'blo
t' - l.o D c o r-.* o 5:| - r rnr q Fo c\ d E :| CE.- o :| l-c),o. -:rPO@'C 4 rl
l 1t u. -JO-CoD O't< c o - ! -Jl o lno >-: C C I T'o -l d -' I O3
D t - -. q1- h rD >C _ O _ : l lr o oo r f -o:r ,O.. Pcczd., -J.c-,o-l-c
(- . r). I 6 o 3 I 0 d,o ry 1 OZa OJooryq-.3. Cl!!t,U
{ I ao
- f, 11 b -
? r" I .
l _ oo'-'
' u tul-,e l- 3
No
do - Q 3 " Cd
d- .-,()I l -. 'olf
l'O'-
r, a o' d -
-lll-
u !: :
]-
..-: i: ::::l:.;: e-. "'f
o- uo-
--.1-.-*'Oi'f_'.Da
!.o. l- o" lD
. - -
t-, F- ry:\- - .'d - -,c.. l
. . ,o - l ot:::.::..:: :: ::- "
l
- _: (
-- - - C
::;:: :::: .:.::::::::.
^

ry O q qil ,ryD
\t I b q u I ! u ry d or ! r ! dr ! cr a t ! r' o ry ! ! o o d o q d o D d d o d d - ti \) ) D N u L c
:tr
d 1Z
'l' r,,r ' r)o
i,lr
o u-- -6on L;r',r-',,-n,r,ilm6'i'rlo,-oG.
*, rJDq,,a,r,J D
o o Do
(r !n(o(olo;[orn:fo5u'ror
! '''r r'
DoD
4ln l

Jj ,rrr.tr,r !r.t!., lutJ tJ-* (rrFfJurcol.Jr.l'_-(rrllror{r6(Irtl}* 3

* q,q lJ * !.o ! * t.,l ! (DcD oo:1 (, - rri ul(, o s (o rrr - o m o '!\ ?.
to I * q,e -t u, i(r Ln I.J et r.J a ! i, --c, or - - ! o (I| s rrl o o t.r 6 to,!
to I.o.,,.1,r, o ! r. tr, q, ,,,ri u,.-o'1odt' !@ ! (r (, t _1 tlt.t('o0 ! u tr(lrryrto ! (I,'. D o -{ mtJ hJl:n lar'{c-Cnl,oU ! dE (, l{i o F
- Lr,- . ' @r'l 'at Cor--'oJ*o.-oflro(omtl'r*(ar'_rcr_ UIUUIUI ' _'rq
. ljt i !Jr. - LO' _'' r!!irrou_!trrc-oootrLaooroo'
utt.rgrr,rr.tLr'r.

tJ * !-\l .! t! rr' !!lJ!_I

I,t! *l'J rar'r \'f f- q!!!oomLtto\nlJ'l- !o t'] o ! :-!
,r(,u+dUlD ! *-.rr ! ua! l'1(!'o!I-D Cood*oi lr.',iOi
.:,:,t.J-or.tD*o(or!rr,q,!6!!ra,of.Jrr,qLo!ii *oAirloroo.,lou--ldo.arutuNr+r!- - UlN(otrJt.Jr!N'1(ll'"1 \r \r -cDq] aD*co@
*.n g g
.,, : ? 1 : , , o, -, .j li i i; F . ; g g J [r n ']' rr; 1, .r
,.r,
lir p l" l :-, ] 1, : 9 :mrlt*f\l'1l.r l
y :r 3 y g 1 ]J 6 P y !" ., 1J 9 i- 9 : r,J )

l r Dln,!,q,.lDa,oo,r*ooD.ttD- lri- . -:tD- GJ-!trl*'lrJ,oo{.('o Cb'looudDDooOo'J_{tlDrr'

! ! F F - l,l F rrre,rr+ fr U ! O l.Jr4rCotJ mur Ftr' * F -'- tolJ lJ ! * qr I

.. lq, F F t.J* - O * o -'l- : @_ lar(D't{rti'o:'-'" '
Uo- . Urai- *- !--- !' !l.lfJ-!nrJrrrUL'(lD't "
_l
- - Uq' -,j'. -o. - rr,- oj rJ'l'1_ -O.' F_.ar" _-'; " U'!(D'S !'to- -*
-t(o,at,arf.J!!rr'utoqr!6jjaicotJ_toDGroqreqtJ'{-l!od66(nro\luror-n11'DlJlrlr-t'(o[oiar':o-'r

' (O l.l l.l _ t.r rd r.Li '- L

- D rtr' , )' L !' ' ql u
. - - - {rr. !
'I,
l*1r',r'[email protected]

Z-lo ur F F iJlraruF fJ urrrr qr lJ

Fo-rrr rrr
: I :?7
- -! -tJ-5 tJ- o.1 tJl-rO o I
'rr ,r, .-a,n-c'*!\l*Ul! ,r g y -, !m\l(rdl-i
NoA- Co-lJ!.1 ':f,J -l'l
9 lJ y ? I ?
r..r
:, r, ., .,. t.' :i :
:
, r
- :. :a F l; y $ :" s y f: t y ,." g ':' y y 9' 9 II ? :: I I:::: III lJ y 3 ,s
.: q,rl -rc,qa- uru,_!to td!(oNl-g'- -_-tr'li tlo-"lt'-lrlu(,u@*ooa-lJooohootnoro'itllF(o-oo-oc'
t
,rrut*'.r.: oi.i ii .tu:-otu l4rlrl UlhJ'l) lJ (r' toul F[r l\) LiltJ'-o r'-r l]riu*
lo,r,01
:r(o(ooorJtJoD(rolrG,oqirroOcrrr!fJCooolcloaool'Joo.}-Folmrr'ldo-lJo'-oou!o-'l
o
:: :' :: : ?:: ir lr
fl lrr o o o ! n U1 outouD
- €!.
o * * (o - i o-tu ul| !4orar! (trca - o- ul o (l -'!lJ0(1|ou'| \1 ! m (Jloclo 6.) *
'f('aaooOO

o o; ;; :J
LJr r'1

:':'::) 9:::)::-:',:-:'?a:::: y',y':::: P 9 4 9: ?3::: ?::3::::

o 6 r.!r!r ! !.i\t.J
? 3:::9::9 ? I 9:':',::
y
q'O<
:
a - -'ta J L .-e a L-,r,.,rC,[,aiit)rjr.-ir]ra)rrrF*OOOoOOOOOOOOOOOOOOOOOoooOOO[OOQC
-

I
iorcr6cn(o(!-l.Jr-r@
:_' :r_,,,e e:'
I.1o€co0@cDcoq:19!!o!o@@-\l@@coq:o:-(o!-r}[email protected](t',O(orrrr'lrJt
: !.,t',1 r r | !":: F F:: FeeF )lII gg e l:: ?F r g': l:': j v a a P
o OO O O O OoO OO
e lOoe lJoooov e lJ sO eO l.'-'e
-J o o
el
o
lc!{
'r
o
3
=
'!
r
ocr O -'.-r a (--'-, cr O OO c' o oo Oo O o o O o O oO O OO O OOO ''
 8L

. sTs{TPu8 aql uo.rJ

pa^omaJ se^1 uo-Jtels aql 'anlssaaxa aq rqEltu s.ro.rf,a asaql leql pa:eadde
tT araqll .luecTJ1uE1s aq dau snoTJ >1uad ;o uoTlEIIIT?sa.raPun alqlssod
,rrlt stuauqr.tPr lalleus JoJ lnq's-lnoq duw ro;1se1 deor:1ead poo1l
rql .)snuraq alqutd.rcce a-re dlTeP satrTl aalr{l sSulpea: aSneE JJBls
stuauq"lec a3:eT .roJ 'ase3 aql sfenTe fou sT sTql fnq 'lueuaJlnba:
aqt sr se 'slanal PooTJ lead proca: saop f,an.Iasqo aql SA?uef SUT alllos
uI pass tur aq ^,(eu lead poolJ aql acuaq pue .(ep e saIIlTf aa:ql ,(1uo
ua1e1 d 1 1 eurou aJ e a.raqt suoTteA:asqo asnecaq slsdleue aql uT suoTf Bls
n l Utr .rSrreB Jlets Jo uorsnI"ul aql lnoqe lqnoP auos sBll a'l aql

.s1sd1Pue aqt uT paPnTauT

.rr,if suoltrr:t s ,)n I.i I Ie dPnls slql ol ssaulnJasn IeTluclod Jraql Pue
:--rrlt Jo r1.]l^ uI 'uollelode:lxJ aAJnJ 3u11e: aql u] acuaPJJuor arou
aneB s: -,". .'nl1 Jo af,uasa-rd aql suoTJBsoT asaql 1B ua{ef SfuaItraJnSaaB
(

a8reqcsIp ttoIJ no1 dluo alan a-raql q8noqlIv 'd1re1fGTs PalcnJfsuoc

se|l qfJlJ aqt uoTteBf,oJuT TeJoT 01 SuTP.IoccV 'salTs a^TJ aql Jo rnoJ
Je s.r.!ao, paJsa.rr d:eqs Jo acuafsTxa aql Pa.Ia^oosTP fcafo:d slql tu1:np
stTST^ pTaTJ tnq (Sr6I-6e61) refl PTrofl Puooas aql uT PauoPueqP a'rafl
suolf ef s aqJ . Ba.le u I zu)i oE AoTaq sf uaulqclec 01 uoTlenba uoTssarSaJ
.Jqt jo uoTlerlldde 3o a3ue.l ar{t SuTPualxa uT TnJasn.(.r.an {lleTlualod
a.l o3a.rorlt ate puB e^er uT sluaurqcleo TTsIus d:arr uo elep
r{1uo aqf
d^eL{ {aqt se rsaJalul lelcads Jo are asaqr '(t€ pue o€ '62 '82 'Lz
srequnN) Sunpueg Jo qtnos aql ol saJB {aPTATC qgll€^t aql uT sfuatrqofec
sa:n8T.{ uT unot{s Pue Z'V
ITsrus a^ll 3o dnorS e saPnlauT Z'I Pue 1'1
pue 1.y sa1qtsI u{ ua^T3 suoTlels 3u18neE 911 lo lsTI TSUTJ aqJ

p^q aqt Jo a?ual.radxa aql SuTSn slsfleue aqf uofJ Palea[aJ

.s1s1Bo1o.1

fo PaPnIcuT arall suoTfels Pue anllcafqns ]uatrxa a3]el B of sBrrl ssaCOrd

eqJ .etEp ,s:ead irral e lsn[ PBtl ]T JT ueql -P*local Euol a1qen1 en
B pBq 1T JT paPnlsuT aq or.{1a>111 a.rolu sB|r 3u1le: f,ood B qlTlr uoTlEls
e snqJ 's1s.4lPue aql uT PaPnfJuT ag PInoLls uoTl€fs aql raqfaqA
aurrl]rtap ol LrolleJS e le PJoDar' jo qftual aql qlTll uoTlJunEuo"
uT paJaplsuo? seA 3uTle.l aqtr uo.uoTfeBroJuT sTtll 'uoTf-ef s..EuTEne3
aqt lo tua6ssasse PIaTJ s,lsTSoTorp,{q aqf qlTrq raqla3o1 alqeTTB^B
sQog pue
uaqlyt uoTlcas ssorc aqf (aEe1 s PaBneB unlllTxan aql
t
s q,oH

a3ets pap.rora.r unurlxsttr aql sT xroH a:aqn) 'anJnc

(
xglnH wt-lvr e.{l
 t
I
a

I
I
l
I
(

I
I
(

I
I
I
(
.suoTf€nba 3u1le: Pue elep cTsBq aql
3u11caqc tsrTJ tnoqtTll Pasn aq lou PTnoqs g'V aIq-sI-u1 uaa13 sJVI^l aql
uosaal slqt faC 'CYH Jlaql lnoqe slqnoP Jo asnecaq s1s.(1eue aql tuofJ
I
pal I Iuro ar.a^1 suoJ tet s asar.ll ' SuoTlef s asaql Jo auo Jo dlTuTf,TA aql uT
(
pa:1nbar sT atBuJfsa pooIJ e JT uollenba uolssa:8al aql uT asn Jo aq
{eu etep asaqJ . srs,(1gue uorssaJ8a.I aq1 uT Pasn lou suoTlels aql Jo
(
auos roJ prt)eltsqP sc]lslfalre.I"q" lun*qclec aql sanlE g'i "fqtf

I
 I
(

*l
lUj @ CoCo (o Co \ -..| (n m ! !
!(^lN 0JN [lN N NN F - F (r).al * zgr\
Ea-
(
(]ll 5 (It (^l N O O O O U.l AN
-(, !(D (,@ N NO uN
(rr.L N
m od
rrt o U||UIN - @'.' r.O(Jn Q|l tJAl ! @ N- \l | ! 3CO
+ ai t + + a + + + aa t I l+ l aa i a+ a I t oao
cP o< o
uo €€t€€(r)>g!>trrtrtttl Erllttd(,U}(J)xx xx ')
of
o !x(/){tlt
o o o o G3
eita
't 't ' ' '
3 0(,)x
<t!d01...14
. X- J
> c''ItE
'1 0
: (
3d3txEltltr'3-D f-!I,c,.{ 6 O ? { f
D'U3!UtffultJ3 0'.oo-'lg't!
PnO)Fn3F3
=1t It
o
g't (rl
lufv:l-.-l-rco ID
c6 1] 30!r-30 E r-O Cc D g 'to -l
- o 1 !3tt'llUn-. ljQ oo o
c) c I
a6ddllddorcfl
'i!
C!
'O ddn x
do
f !'1
c d
90!o
u
il
(
cf iO! o { 3 o d D Tddofddd o! g
C/ o
UddoOOC,t(lxddtr
oJ
o l
g 3 f d3CoC N u
Cu-il<ooE-{{o.
0
D mrc ru tr'r -l-lf
J--trLFCAOcO Fd!-o3llO -1
3
=
o qr Cn ! t,.l -
U}
d>
o€lCllo
:COdof6-)-Jcc-
03 0 f-
llc o
OId-l-trgll
- C
d oc tr c
o O o:l )o@
3c
6l
n
r-
(
/ Dol-C)
'r!
.l' I C F>x u 3 a f 3 6o o f6
|rlCr-UTrl o-o6 - loco- !c@u- .- go
a a<uc - - o c f - 'i! f 6 f, D - - 30 . rnf
-
d - o -(r-t!
IrnZ0D3
o
oool- -.ol
o - ' o,o' .o.o
do-F
C -U B' - lo'
-. Of - o ;l
1 0
o.Ii
]' T
Cd
6P
. - - 'o'F
:'l
. -- c
lC'Jl
6l-
-. o
- - €- -
fH

a' (
d-
a ;aTl 'c'o -t :::o:: o
-l
o
a (
oo
af
tl (o @b r, t ! - !- rJN rJ; ! !F P F - ta, *
to!- - .J
P' ,o iltl
U

I
r O qq vgil d d qil ! @ o do q 6 q oddoo o ! oo d D
UO
o-
.\ ?
,O ! r'r o.r
-o tJ ! -
o't ru or(! NrJ @ r u oN co'{co lDo @ oo "r9
,1 O oloou
o ofu*cnl!* nJ* -tlJtD-
!o
OF
cn--uloNto!o
(roOllOOO-Ol',)rtt
. (rto. . ooorJl.
\-\l(rcro.Lb
{ar(oO(ar{no}!
. o. .
.>lrr 3
6tr
(
o ::-^- -l
o
*6 tt rJtr- - oO. - . . @.rJO-o|O!' NUl J ')
a 6 Co
.l- Fr! tJ lJ r-l
c+
[|- No !\l -o F Tr0
Fa-oot'
^r
d
F.,Lff
dt:13
iJ* |JU o@o(, Ol-N
o tJ nJ O' @ 0
!o'iotn\
(al N N tO |.J \
- Ul'{'{ ! f.J Ul
A € f'+
O
\l -cn D rTr (rt - (n N (t) rD @ ol ('' \ @ t - (al (D U 5 F \l t\.) 1l
3m
ro>
(
F. f d !o o . D. . -
o dtlcc l ! o ! u 5 - o ou. @u- tJoou or!q\ ouN(! 5
i rca:
l
F

5
T-OTIO
Ir-<
f
o
r.
\l t,)
l!- N
u rI (Jl @- D N- (ntJ ! ! - n O ln(Jl (, r!(Jl'a' -i r-
o O - U 6Cn - - S \lar].J - A Al O - o'{ UrhJ ! - @
:0
(
oooc'1 \t - D (o ! {n co o o ln l.o frr (n (r1 .'l cD - tJ rI, o fJ \ o \ l'r
^'ic.,Oln l
G tutaod 0
o S G F F {.- }(nO-- - r!- |'|-rtr\-..'{NLr-- - 4D\IDFF (!'t,-1 D
C
nnq-
-' -6@o l -. (r. !- !mu.
rUl@' -'!-- N' ' \N\' F(g
ra
\v o n
OCo. Co'|.r'
4' ! l.'.1 t.l'.1 o "
(O !
! ol U \ (, (l \ tO @ A ! lrl F tJ {n ! (
o- ld o-'.iJtJarN-DF(oq) * 2A
o . -lOGt ND* tJtu@-
o--tJtJ
! mlJeOo| !'{- (o OtO- -@tJ(rr tJ
F3
:j - - . -
-
!tu 'Z UI
'l ! * ! (, ! - O t1t -{ - r'l lal tJ t3 tJ toUI u ! } !n (r| (o 3A
.+€
.U -

tJ-@
(
+O *!(rl!Ocn-O!(^lJrar(rh) F!
o= ! U o m O CI o iar F Cn (o l'| ! -tJ!A
B tD |! UrUl \'.1 CO m (J] Ln rf
hlT 9 : P : I : 9' )' i"1 99 : : : !'r : )Y
|J \.1 ! O r!
: 9:
oO O F l!@ o-
9?
(,
; r.l q
Pa
o
C' (oO (r € fJm - - O o
T (
PO *ulJu [lF u
ot,.j*0ou orSN l!o\Fo-|'i ol
FUl Arrrr'.1 --U1 o!(Il-ON'TlD!' OCOo6OOD
-'{- - O'.' \l
-.. 5 O\l o \ a O 0l (Jl(r| o mo Oo l
.I
2
T
@ tJ a m A

tJ
*
r!r!-
Uo
qqr
om(DtJ
(iNf
NUr -Nol
qh (r-'49cD
TI I
d o a o - rlro ruo \t (o@ (I,N ! Ul o !ul o -\n(D co ! r,t o
F o (, o o (!' (Jn fi o N co N o rrl 0 0 hJ \ral m Ln (ll o o o
Ln
(
-E
O ooO o I o ooOlJ\loO OO O OOOO-'lOO O 34
f.J 3
o ooo oh)o ooo \looo C'lo o oo oo(oo o o

'; \ (
o ooo o oooo(Doo oo oo o ooo oooo o TJF
OoOO OOOOOFOOOO Oa oOOooOOo o
N N I.J N
fJ ; u t
u uo I
N U O @N N NAI O N tJ N
t o,r o65 NN (J| - Ul \
trln m arcoo NtJl o o N \l
N NN TJ TJ N TJ
Ul o(I o ! \ fJ
0 om o uN \
(
o ootlt ooo ooo ooo9_oo oooo ooo J r T
-tl
* cD 6'.1 co o o - 0 N \ O t! t! oo rtt oQ @ F
@
ra .l ! o u '. -'{ qt !'.1"1@
(^1 N ro o N (o uUl t! @ Crl'.!

rJ J
T (
c,OOOOoOoOOOoaaoOOOOOoOO

(
 T8

.a3ue.r aBets aqt

tlanoc pue (alqels lou s
du Jo suoTtrod luaJaJJTp
I3uTter aqr uaqo) auT.f uT spo;rad
3u1:anoc suoTtenba qcns go tuaraJJTp
Jaqunu B aq uPc aJaql alTS
auo due :oJ

. sJalarrrBJed an:na
tuTle: ct q' P
t^
,T-o tt) q a;Pts 1e j3:pqcslp x
U
(E) eBets Y

'aJaq.+1

q(e + q)e b

e spq u"ru^".tr:r;H:
BuIt", lo uroJ lereuaB y .leaurf_uou.{11ensn
sT atls -re1nc11:ed
e le o3:eqf,srp pue (a8els)
Ia^aI -ratp.,l uaantaq dlqsuortefar aqtr I
uo rr eioaDl r xa -Jn: n c lEu
rr er-Tuq-r ptol z's (

'arnrnJ rqr ur ap"'n aq or sfuauaJnseau (

a3:eq3sTp poorJ aJoo
o'0rre pInoA qcTqi't'drlauorp,{q
uT sruauaAoJdul a1duls auos
?'g uo]lcas
r{lIo' pauJA3uof, sT
pue uo}reEJoJuT uoJrcas
e.g uoTrcas
spuautroDaJ
ssoJc rauupqJ JO asn aql
!alqelrpAp aJe sruauaJrr"""r-aE:eqcslp
I
diuo uaqn sanrnr 3u11e:
Jo fuaudora^ap aqr sJaprsuoJ C
z.g uoTrDas

ssuJc ieuueqc se qcns uoTrern:oJuT

'sa11Jo:d aceJ:ns .rale61 puE
suoTfoas c
TnJasn Jaqro due snld sruauraJnsBatr
t"Ilslxa Jo asn tsaq aqr Bullern uo
"t':i::t:
du l-rur uo >uo'.' aqJ
parra-r a.roJeraqr saArnc C
'aTqBTTeAe e.oIr aqr uT SruauaJnsBetr
reuo!llPPe d{er-rapun or Tec}rce:dur sBf,t at:eqcslp
uT pasn 3u1aq suoTlBls
fJ (oot ueqr e-roB) lcafo:d aql C
Jo J.aq'nu ag:e1 ar{l Jo asnpDag .suoTlBfode:lxa
a^'rnr 3u11er a3:e1 u1 3u11rnsal
oqt se,'t ualqo:d re1nc11-red y .
sfuauaJnsean aSreqcsTp poorJ
Je C
sasz(1euB aqt uT Blep alo13:
d1 11enb 'JBT
poo8
3u1neq Jo acuptJodrul eql
Jo asnef,aq sen-rnJ 3u1ger Jo luaudoTa^ap
uo dpnls slql 3u1:np ruads serir aql C
rJoJJa Jo JunouB arqpraplsuoc
v
C
uoTtrnPolluI I.S

IN sRdOT?
C
^:,rq
NOIJVObA eNIIVf, 's xeNNV
Lr.\rT crlt s€ atrs uo patsur}tsa aq PTnoc 1T dTqPraJa.rd 'auTT lqETeJls
E a^T3 t f lf] teqt anlB^ aqf ot arqTssod sP asoID sB ., e,, qlTlt {TTeTlTuT
tJpts ot a1qerTSaP sT 1l Pu?tl dq auop aq of s1 luaudole^aP 3u1le:
aqt lI 'ado1s 3u1sea:cuT Jo al-rnc e f nocle 1o1d elep aqr (aldurexa
aqf u] t'0 - = c) qSTq ool sT ,,f,.. JI 'ado1s SulsearcaP Jo qled pa'r:nc
e AoIIoJ PtreP aqf (7'g a:n8Ti uT 91.'0- = J) TTeus ool sT ,, c,. Jo anf PA
aqt l1 '(ITetaP aJoln uT ralel paraPTSuoc s1 alduexa aql) Z'S arn31g
ii ! pate_rtsnIIJ s] ar.nteA.rnc srqJ .paleurmITa JO patnpa: d11e1lu3lsqns
aq deu (c+q) 0I3oI lsuTe8e.b 0I3oI 3u1rlo1d ueqA B1EP aqf
(aJoEJaI{ffnJ
Jo allilBnJ.nJ'AIlJa.rJo? uasoqJ sT,,D,, .ra1aore:ed aql JT
.qde-rb dr{t -ldn o dluana a.roru sJuaua.rnsBaID a3:eqcslp aql lno Sulaeds 3o
tcaJJa aqt suq r{cTqn a8ue: ao13 q31q.9q1 slcerfuoe- pue a8ue: AofJ AoT
.rLit spuedxa aues aql 3o SuTllold cTtrq1tr-reEo1 'aJaqirrasla oaJ PUB
"tep
s,"rtrl i A()l trr i- tulod lo Jalsnlr a3:eT B LllTtt ualJo aJnlen JnJ pacunouo:d

Joau}|UulloJ:adedJPauTIuoPa]1o-IdSluaua.InsEeIDa3.reqcslq

'aT qsu os EaJ

sr uasoqc anle^ aql lelil 1uel:oduT s1 1T qcns sP pue ralaue:ed

Jupu rruop aqt sT ,,q,, a3:e 1 s1 3u11e: aql Jo uoTlPf ode:1xa a.raqlvl
.g .a r),t f ' I aBu r'-r aqt uI aTf 01 palcadxa aq dlqeuosea-r 1qtlu ., q..

tnq ,sa:n31_1 osaql 3u1n1.rap u1 suolldunssB Jo.Iaqunu e aJe a.Iaql

'sIauueqf, :e1n8ue1:1 :toJ L'Z 9'Z q

slauu?qc c11oqe:ed Jo lePTozade:1 roJ Z'Z 9'r q

slauueqc :e1nBue1ca.r -roJ 9'I Y

:ap1n8 e sy

.Iorfuor aqt te uoTff,as ssoJc:a,r1: 5q1 ;o adeqs aqf uo luapuadap

s1 -ratr:me:ed s1q1 'd1t1suo11e1 a: a3:eq"sTP a8els aql uI {11:eau11
-uou sasnpoJlul PuP aAJnc 3u11e: aql Jo luauodxa aql sI ,,q,,

.( luaur8as f sorrrJanol aqq f o anrl {1uo sT sTql uaqf aAJnc 8u11e:

rqt ot stuaurEas a1dlt1nrn aJB aJ.aql JT) noTJ o.Iaz oJ Sulpuodsar'roc
a3ne3 JJets aql uo TaAaI aql sT 3- a.IoJa.IaW 'oJaz aSnBA JJEIs aql
pue Io-1 Juoc uoTtpts 3u13ne3 aql Jo uoTlBAaTa aqf uaaolaq a"ua-IaJJTP
3r{t foJ r10IIe otr Peaq aqf 01 palldde uoTfcaJJo" aql sT..c,,

suoltela:dr.'luI TecTs.(qd a,"eq,,q,, PUP,,?,, s'rafatneJeo aqI

 .Eue:eq Bunduag
lB qeDV Buan:; aql
roJ uortDa_r-roc p tlrns sAoqs
1g a.rn31g (r) dars o1 0E (aulI
.
auo s? lold slulod) f:olrpJsTSes
g1 .urntep arues eqt ol
s1u lod 11e 3uTf,q ot sanTel
a8els aqt ot uoTtca-rloa e dlddy
.Bue:eq Bundue"21 e)
le qacv tuan:;
aqt roJ 1u10d slql salprlsnllT .(l)
I.S a:n31g dats ot o3
lJ lqs ou JI .unlep uT tJTqs uappns due
au lut-ratap ol saTBas
rpau r1 uo d11ea13010uoJqc sJueueJnseau
ag:sq"slp a3e1s roTd (i)

.3ue: eq Bundue";
qacy Suaril) aqr .roJ ruaurdolanap a^Jnc 1e
3u 13e: 30 alduexa aql sasn
pue
/1olaq paulltno sT pesn aJnpaco:d
aqg .ssacord sTql lsTssB of ualfT:a
se'q ue:3o:d:alnduoc
NvuJuoJ e'rr"Eo-r.o srqr roJ u.oeJp-al pu'
parcaqa
oq or pepoau qJIq6 san'Inc 3u11e:
Jo raqtrnu aE:e1 aql Jo asne?ag

,TdurBxA

'PJ oDa.r aqt Jo spol:ad

tua.raJJrp
JOJ por\!lop aq '(eui san:nc gurre: ale:edas :o (sanlen aEelS a1e1:dO:dde
dqt ot uoltcaJJoc e Bu1d.$de dq unlBp arnes
aql bl p"lr_rr.ro"
aq dern ruaud'r.s'au a3'reqaslp'sJncDo
lJTqs p qcns Jr .aEneE JJBrs aql
3o Buluo]trsoda: pap.tocarun.{q:o poolJ:ofan p.ralJe
fauueqJ aql Jo
ruauuSlre-er dq pesnec aq pTnor rJlqs y ..onrep
uJ rJTqs z(ue sleaaar
' 1u rod dt'1 1u10d'd1iea130JouoJqJ
sruaEaJnseau a3.reqcslp tuTlrord

.ro1d cFtrrlf 1.reBo1 (

)Lit uo adois u r aBuBi{J p se luapT^a
aq TIT^1 sTrll .d1qsuo11e1a:1red
oat e uT ?{n5.;-1 ..,uru Ia^aI _relpl*r JgIncTl:ed
b 1e IoJluoc lupaJlsurrop c
r l!qS u JO SrxJJq J)r)()i, sp qCnS UoTlCaS
SSOJJ U.I saSueqa 1dn:qy
C
(a+q) 3oT q a e 8o1 b 3ot
c
: ,, q., :a1aue: ed
sl Etep aq1 q8no:ql uAeJp
"Truqll:eBo1 auTT aql gio adols aqg €
'(086T 'or^ll1) 3ul8neg ru'arrs uo
poLltau aqr dq qde:B ts-r r aqt uo.rJ
rBnuE{ aqr ur peqTlcsap C
J d11ea1qde:B palmTlsa :o passan8 aq
tr BJ i r, Jo anf eA rreu e pe^Jnr
dl8uo-rls sT rold Bullfnsa.r
aql JI oJaz. C
Jo,J, qlTA sT lu1od Bu11:els alqTsuas B ufioulun
sT r.roJJar umlBp
sit{l uaW\ '(u 9.0 - Jo Bulpea:. a8ne8
o-raz rPqr palradxa aq prno. 7r
JJEts e punoJp Jncco plnon noIJ C
z'g a:n31g u1) ao1; oJaz rp agneS JJBrs
t.g arn6rJ

I r-s eu 1 abteqcs;q

ooo-lj sl lsn9nv' or uol8d

NOIVIS oto l-v sl.N3n3UnSV3n 3gUVH9slo

(n
ot
(o
o
ru 9t'0 . v
a

I 8Z6t rsngnv rsodl

Nollvrs l1lns38
lv srN3H3UnSV3n 39UVH3S|O

6ue,req Ounduey 1e qacv 6uanry ! utnlep uollels u! lrlt{S

 t
)I I cas-ssoJ J (
rsLrroutr.rvor JBau STr{1 :O; aIqzuOSeAJ SI ?'1 = 9lUaUOdXa
- ---4,,-- ^a -

aql . slupq daals qt TA f a^TJ' apT& ,(1an1le1 a.r e sT Eue:eq (

Sundurey re qacy Suan:;i aq1 '8'Z 01 €'1 a8ue: aql uT aq
ot pa1llurad dluo sT..q,. (uoseal lelcads e sT afaqf sseTun (
:uoTt3as-sso.rJ uoJlels 3u1Ene3 aql ,roJ alqsuoseal sT
.,q,, luauodxa aqt teqt uTetJaJSe o1 apeu uaql sT lcaqc V (0I)
(
'..q., JalauP.rPd
Jol anIBn alqeuosea.l e qsllqPlsa ol dressacau (
s.L uo !lcas-sso.1 c lan J.r au.f 3o aEpallr!oul 9 UOTIBnlTS STql
ul ' 1eu13:er! s i 1u1od uTel.ra? e .ralle 1Tl ul luauano:du1 I
rq..l ldndl,loH .,,?,, uI asEoJJul ue qllo d1nO1S aseaJcaP
ot sdiluituoc rdlJed.raql Pue'd1p1de.r sTTeJ KTIeTITUT (
uo]t cunJ ro-tr.r .)qt a-raH '(S) dals u] .Iof,Je lllnuJuTIn e uJelqo
ot alqrssodurr aq detl ll 'sluaua.lnseau a3:eq"sTP aql ul (
Jallecs aTqBJaPTsuoc sT a.laqf PuP sfloTJ;o atue: PalTuTI e
-ra^o paf e.r sl uoTrets e alaqlr dTTelcadda 'sacuelsu1--auos uI (O) I
.r'ro1aq (g1) dals ol oB paulelqo sT trnuTuTIu
e lI '?'I = q Pue l'9t = e LIITlr '99'- s] ,,c., unulldo (
aql 3ue:eq Sundue) fP leql sAoqs C'g af,nBTC 'Palou sJncco
r oJr o rnnulru !{n qcrqfl I P lu tod aql Pu9. ., .raf auered 1su1eE-r (
".,
prtto1-d a:e fo-1Ja rrolssa:EaJ aql Pue,,q.. t..s.. sraJalueJed (8)
.t.O - - ., :)., pue 9'0 - = ,. J.. JoJ palloldar B]BP aql slrotls (
7'B arnSIJ 'PaPJocal are .rof,Ja Paf eTcosse puB .,q,, PUB ,.e..
s.ratauered 'paleada: (S) PuB (S) sdals PuE,.c,,3u1sBaJJuT (
ue q] Trrl sauTl lef an as PalEInclBJar s] ( c+q) 3oT (r)
. qtTq Jo.rJa aqt pue PauTelqo sT lTJ f ood I
d 1an1te1a: e 'SL'O - = ..f,,. ql Tr1 'aldnexa 'rno uI 'uolssa:3a:
aql qtlt'r PaleJcosse roJJa aql alelncIPa'pue qceo:dde (
sa-renbs fseal e Sursn el"P aql q8no:ql aull rqSte:ls e lTJ (g)
... c,, Jo anlBA f caJJocuT c
uE sa1g1u31s rl auTT slql of aJnfBnJnc aq? aloN 'PalloTd
"Itl^1
pf ep asaqf sAoqs z'g af n8T.{ '(dTTeTlTu*T) EL'0 - = ? puB I
6L,O = ufuq alduexa rno uI 'b 3ot lsule8e (a+q) Eo1 lo16 (S)
(a + uTuq)3ol I
atelnrlec or alqlssodur aq uaql PTnoo 1T asneJaq uluq- o1
lenba Jo ueLlt ssa I aq ]ouuec ,, f,.. '(u Tuq) las lua&aJnseau I
a3: erlcslp aqt u1 a3e1s rDnrrrTuTrB aql snurln ueql f alea:3 1snf
tas aq plnoqs ,,c., d11e1r1u1 '(a+q) 3o1 pue b 3oT alnduo3 (l) l
't:ed qre€ ol sauTf aleredas lTJ Pue las PleP aPT^TS (g)
l
8.9 arn6l{ JO

c.. JaleuteJed aAJnc

t.0- 9.(

8.O

o'r ulnurutnl gg.o - :c

'Tl I
ot
z.L o
o
t.l o (

o
9'l o (

Q.l

(
t-98:e
o.z

olc+ul?
: Q (
uS13lrlvuvd 3AUnJ
I
arn6tJ
ootoo'r I
(

o
c
SI'0-.3 oo . . x.
,/t:
9.O - .3 xX .. C
t'0 - .J oo i!
'y'o x s.o- (ob
'r/ ,4t' x C
c / xx JLx,./\ o o
7 xx/ oo J

' "24
x
o o\{o:
..o
+
o
c
-*Y o

aao
aaa
C
..-.&K
o..,./-$z' ,1>' oo
ooo C
C
C
 L8

. stuaTalJJaoc
ssauq8no-r s,3u1uue14 puB s
,[,zaq1 ,{1an11cadsa.r o]p u pue c
.)JeJJns Jalelr aql;o adols feuTpnlyBuol aqt s] S
Tauueqc aql Jo:a1au1:ad palfalrl
r rr-J .{q PeP]^rP ea-rE lrroTJ aqt a] (snlpe: crlne:p^(q aql sT
u
,(lTcoTan ..roIJ ueau aql sT aJaqlA
^

su/J= elnn:oJ s, tzaq3

z / ts t/zd =A eTnrn.r oJ s, Eu 1uu e14

I

du ruuei,,t ol anp aJ e apIr.ruJoJ q?ns Jo salduexa u'ou)t lsaq aqI .ualqord

dzaq3 pue I
sl r' sauo.)rc,\o .ed:eqcslp aql a^lE qr eaJe (
,,orJ aqr dq pal1d1r1nu
PLr tr ' u irrtlj-r c)J Lr Jupt:, J sa"r r1oTJ p 3u1sn palplnDf ec ST ltoTJ
Jo dlTcoTan
aqt Llclqn u! poqtau ea:e_ado1s e Jo asn .sAoTJ qtTq (
re fauueqc aql
ro dJlauoa! a{l u1 e8ueqc pa1Jeu due:-og
funocce lquuea sAoTJ EnTpartr
Pue ..roT of parlrJ uoTrenbo an;nc 3u11e: e uo paseq uoTsualxg .ralal
poo.ri unulx*' aqr or aSreqcsTp paJnseau
c
1saq31q aqr uoJJ a^Jnc 3u11er
e 3u 11e1ode.rf xo Jo poqlau ,{:olce;slles .,{1a1a1dnoc
ou sT aJa..^tr c
uoltelode:txa@ C
E.s

-- n.,(ss'o - q)?.9€ C
:Bue:eq gundue) le sltofg go a8ue: eJTlua aql auTJep
or ruaTJrJJns sT 3u11e: auo (anoqe paJaplsuoe alduexa
C
aql ur

.a8uer q3ea JaAo padolanap

C
senlnc Bulle: eleredas pue sfuau8as auyl 1qB1e:ls
ro oAl otuT papT^Ip aq deru elep aqt,b tsuTe8e (t+r{)
a-rouc
Q
3o 1o1d
)rurqtr:e3ot aqt uo ouTt lt{EIB.r1s a13u1s e ulptqo
o1 alqlssod
liii L I i 'Poulelqo uaaq aAEq s-ratamered lsaq aql JalJe .JI
b^
C
(it)
e
 --
88

'palrnlrsuoc .rq uBJ anJn? 3u11e: e sa3e1s snoTJpn ]e arnpacord

slQl 3u Ilsadar fg 'Ba.lE noTl ar{t pup dtTcoTa^ aqt Jo tanpo:d aq1
rlloJJ punog s1 aS.reqcsTp aqJ .r(lTaoTan ao1;: atera^p aql ateuTlsa ol
eInu.roJ qceqsla6-dc:BO aqt otuT patntTtsqns sT anfe^ sTr{t pup XOIBoI
rsule8e 9771 go qde:3 aqt uroJJ punoJ sl J//I Jo anfB^ aqJ .U ataTn"fea
of pasn pue punoJ a-le.ralauy:ad pallaa pue BaJe AofJ aql ,1ana1
poolJ papJosal e -roJ rloTJ aqf als'Trsa oJ .x0I301 lsule8e eTnu.roJ
qJpqsrag-dc:Bq arlr uro-rJ palerncTBc 'J//I 3uT1ro1d .{q stulSneE .oro13
dIqcllts^e aqt uro:g dlleclqderB pa]ernTtsa a:e ,(aql leqt palsatEns sT
r I -rxatuoc tuasa.rd aqt uJ '(6L61 daH 'gL6t 3s:nqleg) TeT-ra1eu paq aql
Jo saltrado:d aTqBaJnseam.{11ec1sdqd o1 sluaTf,}JJaoc aseql aleTaJ ol
alqlssod s1 tT suoTteclldde arnos uI .stuaTeTJJaor a.rB q pup a aJaqA

(sq)0T3o1 c 1t
1

9013o1 Jo uoTfcunJ
rBaulI e se J//I sassa.rdxa uolqenba atTqI,l-{ooJ.qaToJ aW .J AuTlEoTtsa
u1 sI eTnurJoJ aqt Jo uoTtBJrldde uT ualqoJd uleru aql ule8y

-1 olcBJ uoTtcTJJ q"EqsTa6_.{c-req aqt s} J puB

d11rre:E ol anp uoTle-raTacce aql sT a
pauTJap (1sno1na:d se a:e S pup U araqA

r't)=,t.
(
E'o'suSg' -
(
s] BInru-roJ qf,eqsTa6-dc:eq aq; . sTseq
IElTlaroaqt punos B seq pue lcal.rof,.{11euo1suauTp sT l} sE aBTnEJoJ (
.,2aii..,l :o SuluupH aqt laqtTa ol pa::a;a-rd sT qJTq.tt,elnu:og aeuelsTsaJ
AorJ qceqslag-dc:eo aql roJ paleJrsllrT aq rrrA sTqJ .uollelodprfxa
(
foJ sTseq aqt sB sTqf asn pue sAoTJ paEneE go a8ue: eql.rano
tuaT"TJJaoc ssauqSnoJ aql afBrrrTlsa ol sT a:npato:d aATlBuralTp uV
I
.(OSOI) AoqJ aldurexa ro; ,aas .slualcygJaoe
Jo aTqet e qtTA
IauueqJ aqt Jo uosTJeduor B uo pasBq ,(1q1ssod {1ua1a1JJaoe ssauqtno: (
aql 3uluTut.ratap uT sT uoTtenba ue qcns 3u1d1dde uy Jo.rla Jo a"Jnos
u.rLrtl aL1,l .patatrTtsa aq ot tuaTclJJaoc ssauq8no: e pue (Bu1'poolgi JaIJB I
srr ) pa-lnspaE o1 adols 1eu1pn11tuoT aqf , pa{anrns
cur -1nors uo:3r
ol uoTJcas sso.r" aql a-r1nba: aefnw.roJ asaql laql uaas aq ITTn lI
aq
I
e
 (

.,,:t lrur I LJ e SuTIBrtr ul paJapTSuof, aq plnoqs dlqsuollela.r

I s^a p()(,
a3: et1.rs r p_a3ets aqt uT FtuTetJacun pa11dur1 aqt tnq poqlau eaJe
-ado1s dlr,l tuo.rJ auo aql aq ol pa.rapTsuoc aq Jsrun asec sTql uT afErlTlsa
-1 a:ltaq aql
.1_s€rn Zggz sl u
0.E te a8reqaslp aql etep tul8neE
I
molJ rurtrs rtll ol Prllll q( lFU)p = b u-roJ aql Jo uoTlenba ue moJd
(
I-s€E 86tZ
1'gE1 x ['t
vx^ = a8:eqaslq
I
T-su e '€ (
(9100.0 x gZ.E x 18.6 x 8) / x LO.,
I
: san13 uoTtpnba r{"Eqs1a6-.(c:eq uT paf ntTlsqns
(
qrTqA 'L0'' sT ur 9Z'S = d ot EuTpuodsa::oc Jl/I ,7.g a-rn3Td uofJ

e
u 9z's =- z'rrr aJoJa-raql
T'8S/ = u
_u T.R(/ PAJ€
c
no LJ aqt pue ur Z.rrl, s] :atatnT.rad pattan aqt E O. S go aEels P fV
C
sTuTod aql of PallTJ
aull tq3lerts aql pue XOI3ol tsuTege pattoTd J//T s6oqs ?.g a:nt;g c
'pafelncle.'- )l/r puB snlpe: cllnerpdq 'd11co1an aqt puB uoTlcas ssotrc
aqf uoJJ punoJ are .taJaul:ed pelta^1 pue pare noIJ aW .I.g aIqBJ e
u T se paJapJo pue palBInqBf a_re sJuaruaJnsearu aAJpqrsT? aqJ
e
aa8egs slqt 3e a8reqcslp aqt sT terll .tr 0.E
d11errdK1 sr rua^a TBnuuB rsaBJ.eI aqr rnq u / sT arTs
c
.I_r 'a^o
Irlt tlr p.rl)ro)crl Ta^aI pootJ unul]xettr aqJ r 9I00.O sT ,S ,aceg:ns
raleA dqt ;o adols leuTpntl3ol aql .u Sg.Z 07 u tg.g aflue.r afle?s e
aqt uT atgelTeAe aJe s8u13ne8 qeJBO le8ung fE TJql gueleg rog
a
aTdtrBxx
e
 06

I
(

eTnuroJ qreqsTafl-dc:eq vorJ J//I

(
:aqau1:a{ patta&/BarV = snTpsU clloerpdg
ea-rv/rlolc = ,t]T coTaA
(
uo]lcas-ssoJ." rtroJJ .ra1au1:ad pallafl pu" pa.rV
(
3u18neE ltoT; troJJ AoTJ pue a8elg

I
ch'7 60' e 7f
c>
.1
I L'6ZI r i-o_e 0' /09 ot'z I
tTZ nr't zs' r I' 6ZI ,'00, 0'809 rt'z
69' Z Oir'C dL L I'18I o'rE, 0'808 tL'z I
s6'a L> L 86'I | '1 Cl s'69' 0're6 98',2 (
sul
u stu
zuI
Itr ut
T- l-t
sn lPeu ia=Jau1.ra4

J/ /I a 1 1n e.r. pdg d1 1ao1a1 Pa11afi eaJy AoIJ aBe15 I

--
sluaruaJnseau ad.r Pt{cs Tp Itro.r J l//l 3u71BIncT93 I'g aTqel I
I
t.B arnotj

/ suberv \= lr
\:=6- ) T

I
ot (
qareq le6ung le lreH 6ue1eg ro; g 6o1 qdelg
#,o
I
I
(

I
 6t] I

(
.aluPualulsul uo-Jlels IPJauaS ul Pue suoTlcas-sso]c Ieuueq" afe.InccP
3u1u1e1qo padoldua di InJasn aq ueJ slDeaf ar{l Jaqleall .{:p 3o spo;:ad
(
au1:n11'sdse?qf,nsulTnlasnfsolua-resleoIJ'sJaATJuaPBTSTrqaP
Jo Sulr'rolJ tse; uI {11e1aedsa sleoq 3u1sn J9 ['JeF\' aq fnq aBpl:q '{q:eau
ro de6alqel e uoJl tafarn fuarrnr e 3u1sn lqE1u le aSne8 01 Paf,edard
I
aq p tnorls suBet aql . s3u1f e-r pa^TraP aql Jo luaEan o.rdu1 PaI.Ieu B uT (
3ull1nsa: aperx aq plnortl s8ulEneS poolJ duan:ncco daql .ranauaqd sPooTJ
e3ne3 of paredard aq pue suoTlels 3u13ne3 Jo.Iaqunu B of asol" PTaIJ
(
aqt u] paseq aq PInof, sueal aql JI '1la{ aleq sueal 3u18ne8 aql falJe
suoJtets 3u13ne8 aqt ssed slead pooIJ aql dsP aql uT a1e1 tulrJncco (
smroJs.raPunql af,e sITPJule:,(aeaq Jo dlTJo[Pu aql acuTS '3u1uana
-lo uooulalJe aql uT auror.l uJnJaJ puB suoTlels SulEnBE aJou Jo euo
ot rrr'!lJo IeJOI aqt uo.rJ {ep qcea a^Tf,P slneal 3u13ne3 fuasa:d fV
l
',(ce:ncee 3u11e: I
u't tuauanordul Plder e ot peal pTnoqs ler{f ac}lce:d c1:1auo:pdq u1
a8u eqc e pu auuo ce-r ol al I I pTnolr frodaJ.- sTql Jo s.roqf ne aqr f,8o1o:p,tU
I
poolJ ur atu€fJodur qcns fo sT s8ul?e: ;o 1uando1.""P aql acuTs
l
stuauJn@ ,'s'
,,(rtauorpfq uT
I
 aq ol elPP 'paqsrTqelsa
TPJor uo paseq poqraE
safqeua qceo-rdde uorreur:sr nna-, _
oQl dolanap ot pasn
ldcy: Idua ue 8u1aq .;i;:::
:";l:J"e'ro*qE1e:rs
e
I
auo aLr s1 anbruqrar
uolsse-r'a, t*
"ror;;;TJ;" I
sdJlllt?JJ luauqJrpJ 'q xauuv uT
snolr u,1, au1 vih,r? ParTerap aJ,,
r1;1cads aqg .uorr"uut ""ouT ol pesn scTlsFJalcaJeQc
3uauqcle3
TgruoJ u1 :o dltu .""JlsJJalcp.rpqf,
tuorlqrlp? Js faqlTa
asaql Jo sa"lpul
ot.{vH salPTaJ leQt padoTa^ap
oQ ot uo!renbo ue seTqpua i

srsdTru" uolut
pu r. ad erls i
ruau,Lrrrer ,drTsuap '(sa1e1) aBe:ors
::,t""""",
'odo1s 'ataxllr uo osle ";;r;;; ".:- "-"n Ru"1 's11os Pue d'oroa'
f,- ! poorr
lnq .;;: uo ruapuedap rsnf
renuup ueau
"u, ,"tlJ.;rj;;";;;"t"
":^-"u' asrnoc Jo 'sdlqsuoyrpla.r
rou
lua-ldJJlp Jo uoslreduc>c
a,,,paul,rarep aq "-::-
", ;;;";"1,,:,.;"1;::.:r";l
: '|r{}r'r'!) srsnlplrp uolSsarS0g
.JrqgCTldde l:r"l:";"r;:::l
-rcdu 1; lr;L' os slszileue a:e sanblUqJar uo}ssa:ga:
aJoJoq saIqerJpA
r] urrll ''rpaur[-uou s:ead.de -q!{r u-roJSUeJl 0r d:r"""""u
dlqsuol]era_r eHr
agr u] 'aulr Jr .asuas sarenbs_rsear".,
lanlrdo ue saplno:d qcTq6
dq paaoua: aq uec s1s,f1eue uolSSa:3a:
sT.ll 30 ssaue^rr3afqns EulSn
patuasa-rdar aq uec aqJ .a_rn'1,: aHf, uo
dlqsuortelaJ paAJasqo au11 e dq
due pup earp
lsule8e pa31o1d
;:,;.,]I.j"",ilj:^" ;":"fi'";":,;::"; pan:ase. uo paseq
'asrla:d e se EaJB :"",eA
T ll 1nq raPou' leclsdqd
pue dyr^{ uaa&raq
o1c1'rssod lou s1 dlqsu aql aqlrcsap ol
tI 'grBu", *r"rr",,::,"^:']""otterar
q'noqlTe Eare srT
dq uan rd s.i
"rjil";""] xap-r:1 a^Trpurarrp
ue
ii:*
dlqsuolrplar reurJoJ
e qsTTqersa or pue I'j;"T.,;':;]"jil:;
aqr pu' poorJ rua.qclpc aqr Jo cTrslJefref,eqJ
Jo azTS aqr qroq xepul or
due 3ro oq ot ,:anartoH .spoorJ dJessacau sJ
:aBB1q aaeq slua
rr asn

luauqJlef, JernJ]l;ed
uaao,laq dlqsuolteraJ
;l'il: :::l:1";1ffi"
aqt Js u.roJ
1e:aua' aur

';,;;i:ii} l,:'lJ] .', saqrrcsap xauu,

;;il:]J-:.
de or dvr.{ aqr Bullqeua
",;"'l::: ;ll morJ par'.nTrse
0"r"""1,'ot:r"::l;::;1"*Bqr
ue E:aldpqJ uI

uoitrnFor-lur
I.c
NOIJvobs ToIJvHIJsa
dvl{ FH.I c xaNr{V
 or11 aql uaarllaq uollelaJJoc qETq d:an e sT aJaql sv 'cYI^l t{lT!!
patelaJJor tsaq a.rp ('ISH pue VgW) saTqeTJa^ azJs ]Bql uaas aq uec lT
qJTqr't ruoJJ saTqeTJe^ pauJoJsueJt aql"Jo xTJlBltr uoTleTaff,oJ aql san13
I'J aTqel 'a8ets sTqt 1e 0 xauuY ol os.lnocaa 1no{lTa 1xa] 3u1ao11o3
aql psal ot alqp aq plnoqs sraPBaJ qSnoqlle'uoJlcas sTql uT PaqTJssaP
soltslJatf,eJer.{:) fuat[qc]ec aql ;o uo1ld]JcsaP IInJ B sanlE q xauuv
(
pooTt ienuue ueatr atlt :ol u=olle-nba uolss5i€au z'J

(i) JYI^I
(

se saTqeTJP^ (
I riLr r31Jc, irLl] JO surral u] passa:dxa aq ue) uollenba ue
qcns
(
Q) + ZX 0T3o1J + IX OIEoIg + V = gVll 0T8oT

(
ul.roJ aql seq IaPoul

aqJ dllecgnqrl:e3o1 tr aJ-an saTqPT-rP^ TTe (scTlsJratJe.reqc

pau.roJsuef,
tIX a:aqn
I
tuauqf,tea) saTqel.rE^ luaPuadapu'1 aql a.re "' 'ZX
(
+ZX"+IXq+e JVH
(
ur-roJ aql Jo Tapou
e SullrTJ Jo pBalsuT snqJ 'stTlsIf,afoe-rBqc luaBqclPe Pue dYH uaaAlaq I
d1r-1suo1teIa.t aqf ezT.lBauTI o1 Pa.rlnbar aq PTnolt saIqBTre^ aql Jo
uo'ilerB-roJsue-rf s lPql PaTBa^ar elep aql 3o s1s'(1eue '.t:eu1u11a:d y a
'pf,o?a.t aTqelTe^e Jo ql3ua1 aqtr I
o:t BuIplocJp, to I sraldeq3 Jo sPot]laul aql 3u1sn PalETn)IeD uaaq aAeq
spoolJ lenuue ueau aqJ .pooll Ienuup ueai! palB!r]tsa TTal't dlqeuosear
ca^Bqo]Pa.raPTsuocPuEs}fualnq"fecqceePuePassasseuaa(l
t
sEq fuauqcttsr qaea loJ (y xauuy uT PaIT;laP sB) snoIJ PooTJ:o; tu1lel
aqf Jo dcerncJB pue proJe.l aql Jo d]11enb pue ql3ual aql las sTql uT
t
uoTSnIcuT JoJ 'sasdlvue uolssa:3aJ TTe troJ sTSeq aq] uJoJ qcTqn Pue
dpnls sTqt u1 pasn elPP aql uleluo3 Z'Y pue I'V saIqBJ 'PauTEJalaP
t
dlalerncce aq tsnu anlen tl)€a pue alqlssod sT se (sf,TlslJ.afce-rBtlc e
ruauq)le3 PuP ,y14; ual T qf,Pa -roJ sanTP^ ;o aSue'r e alalduoc
sB ulBluoi tsnu elep aqJ, 'l] a^lJaP ol Pasn las eleP aql uo luapuadap
dllearB s1 uoTtstba uolsser3a: tullTnsaJ sLll 1o dfllenb aql
 96
z
o
r
o
5$gEgqgEHFFFE F-l
ts
ct
P
z c)
ol
i
::::.oi$$+roo
HnsE€SFE$$g$FE Fl
Ft
o o
ts
- A)
n
o
o-
a
gPpgotlrrr
FFsb;;;S^:-
,i..-\o;SHi$gEE :: h F
I
o n
- F I
-- X C

I
7
t=
h
HeHgggHs;ig 6
t- C
tll

-
:::e9?ppoc
Sg-:HHHFbb v) C
:D O

::l
J
-1-rtl
YYP:9ooo
e
r
s:s-isS$Sg
;-Nro-l
.A C
I
0)
-
+P+$Aoo
SgSSsFSg e
$P$$3o
HSFEsSg e
e
e
Alll
v u
^
.Y .i -'? o\ o
6d.'o-; Xg t-.)
d

nPoo ^I
(_o!\OX
o!@E (t

tt
9PP:
53Sg :<
F

rl ,J
\J
c
oA.;( \O-
vF
F
H
z
I
O

uo -tr

O
O
 .ITeJuTe.r Jo a.lnleu aqt ,alduexa :o3 ,a111un pT.to!4
aqr ur a:aq."rdue ssaco:d aues aqr sl sarpl dq spoo13
Jo uoTfenuarre
aql se dPrr sTLit uT aTqB-1 aJsuerl aq ol paJapTsuor sT
lueTclJJaoc
sIqJ .t:odap sarpnts pooTC )n aqt uorJ luaT3TJJeoc xapuT aIEI
aql
rcl c,pu ..r pap iJap stsn lT uoseal s.rql JOJ .1as
elep aql uT punoJ sa)ipT
''rrir JC) sLjoilr?cor dqr qrlll ruarsJsuoJ qSnoqlJe't{3T,-i dlqeuoseaJun
sT
Ll rlq^1 '0'z tr rurlJllJaoJ B uT palTnsa-r sTW .luefsuoc sluaTrTJJaOa
1; -
-raq10 aq1 3urp10q aITq& ular arer aqr Surppe f,q p"rp-rqTTes se& xapuT
alel e'luauqcrpc e uo aSerols arer Jo srJaJJa aHl:oJ rl0TTB OJ

' (SL6I't.rodag saTpnts

POOiC XO 3a) salpnls.laL{lo ruoJJ sanle^ iilJf lualsrsuoJ sJ luaTcrJJao)

')r'll puu lutodnarn 1eci3o1o:pd11 B uo-rJ alqeJrsap sT uorsnTcur

srl dllpcltsttets luprlJluBTs dlo.leq s] ,SI^IIS,xapul adols aqr
lo
r.'rsnilLr r aqt rtrLU! .\,'jiUV uo luapuadap JTaslT sJ qJTq.'r ltlal
VAUV aql
(
-r(';r 1Lr.iuod:i,, LrrqLlJen u saJnpcj-rrul slrir (e uo:1enba ur se) sarqeJJp^
Leu r31:o aqr jo srurar u1 '(z uorlenba uT Tx) arqel_re^ luapuadapul I
-rdrltrnJ e se Iapou uo-!ssa:Ea-l aql ul papnTsuT s] pu€ onTea (yguv)3o1
dtlt 3ur:enbs dq pau:oJ sl u-ral srqJ .ea.ra aBeuleJp ur asga-rJul I
ue lo ltaJJa aql 3u-rcnpal oeaJJsurrop sIa^e.rl lT se poolJ e uo asodurr
ueJ ruaurqJrec a3:u1 p leqr uorrenuarrp aqr etruase:da-r qclqo, .ZVSUV
'nr-lot Ea-r e puoDas e s] uoTssa:Bar aql .raf ua gt aTqETJe^ pJTqt aql
l
'xopu1 lpJur.e-r aqt 'g.Vgdv pup VAUV Jo uo1SnTruT aqt latJV .d1qe11a: (
Prl Btullsa aq ol elf,al e qcns Jo luaTcTJJaO" aql JoJ AeJ Oof aJall
aldqf sY Pall iulo a-ran I'O uPql :elea:3 xapuT a)iBT p qlT,.r sluatrqclec a
IIe llas elep pafDJJrsaJ e uo paoroJ|:ad uaaq an eq suorssa:3a-r
asau.T. 'aBels qrea te pappe Suraq aIqeT-reA luepuadapul BlJxa ue
tlt llt sTapou uor ssarBa_r Jo aruanbas e Jo sI Tetap sanlB
I
Z. C aTqeJ,
e
uollenba aqt uJ luatoaAo:dur1 fueclJluBls e dq pa1311snf
sr a IqeT-1 e^ e-I Jxe due ;o uoIsnTcuT aq] laqraqo' a3e1s qcBa 1e ssassp
c)l rlqlssod s! l! autrl e le aIgPTren auo Tapou uorssa:Ba-t aql Eulplrnq
dq.:enaooil .1111 rlJJIp uoT?enba uolssa:Ba-r aqtr u] uoTsnfJu] JoJ
sdrqurluA Jo ras-qns rsaq aqr Jo uolrcalas aql sar€In saTqpT-reA uaarlraq
uoTtela-lrol.ratuT sTqJ . oof azTS Jo xapul apn.rc e se slcB xapuT
edols aqr os pue (sauo rTsus aqr ueqr fallerJ ele sfuauqf,l'c ag:e1
dqr ras erep aqr u] 1eq1 SuTlBrlpuT parereJJoD dlenlleSau osre aJB
odols puE azls leql sT slql Jo uoTleueldxa eq1 . sfuanqclec :adaals
uro-tJ pal cadxa aq pTnoo spooIJ :aB:e1 spaJaq^r
.{vi^l ql Tr,1 pateTar.roo
d1a.r. lleBau TIE eJp saTqpTten adols aqt teql alou ol
Bu11sa.ralur
sr rr 'slsdleup uorssa.rSa-r aqr u1 InJasn oq rTTi'' auo fluo salqBTJBA
 LO

.(^) uo]tenba roJ papnlf,ur aJalr stuauqcJee

erlxa L '('',) ol ( l) suo!lu.ba u1 srsdleuB .roJ pasn aJaA sfuauqJlpo
E0 I '7

ol3oI 3u11e1 dq pau-roJSuB-rt aJaA saIqpTJe^

TTV .1 :satoN

961 'f) tr 1, l, 'l) fll. t- ttE'0 610'Z- 3)Y'i

sn 1d

an oqe sP JVH (^ )

89'i 01.0 '0 /r1'0 SHI S

z0'8 s0 t'0 Ctih' 7 wsdv

tq'I- 8r0'0 LZj'0- ZVsUY
91'I I /80'0 OZO' I Vf,UV
,oz'0 888 '0 ,8' L- 189'0 860 'S- lu el su oJ dYH (nr)

810'0 1e0'0- ZYgUY

6r'8 90e'0 ,08'z
(
UVSdV
9E'11 sB0'0 886'0
9OZ' O E88'0 TY L_ 0s9'0 rr6'r- lu€lsuoJ
VSUY

JV1^l ( TTT)
I
t8'8 862' 0 0r9'z wsdv C

802 '0 r88'0

'/ L

i8'1-
>L CEO

I E9'O
'O zs8'0
980' E-
YSUV
lu Pl su o3 JYI^I (rr) I
0t'61 Ee0'0 r/9'0 vau.v
9 LZ'O 88r.'0 /60'0 er9'0 luelsuoS JYI^I ( T )

luaTrTlJaoc Jo SaIqBTJB^ aIqeT-re^

'd'a.s orrS PrePuBls luaTcTJJaoJ
-u
L
-1
tuapuadapul luapuadag

JVH A]PIIITlSA O] suoTsserSau z.J 3.tsvJ

 86

.sPuaJl
ipuolSaJ rueclJruSTs ou sreanar pue uoTrenba aqr
Eo.lJ araJTrsa JO
ro'lra IetrorcpJ aqr Jo uorrnqrf,lslp
rpcTqde:30a3 aqr s^1.qs 7.3 a-rn31g
-.ruarlrs e {lrns
Jc> f,trxalduoc pasea.rcuT aql pa}lllsnf
pue _ro-r-ra
t, l r()rr'frp!)J :rrrrrrJIuSIS
e poJnpo-rd sa,rrreurarle asaqr
'(P''l:l o-1 or1 l{vgdv pue JO auou -raAadoH
vauv qroq) sanrp^ c}rsr-rarce-reqJ luaruqJrB)
ot BulproJte:o (e-rlsuns puE e^ef) d11euo1Ba.r
papT^rp se1r1 fas
erep arlr rlJTiiA u1 poJ.apTSuoJ a-raA
STapotr uolssarsal Jaql0 snor-rpn

.g _ratdeq3 u1
l),)qr..r )s.'p su uttsp 1BJol Jo asn aqr dq pacnpal
aq ualJo upc JoJJa aB:e1
slrlr rp.rl parou eq prnoqs lr '65'r.{q
paplalp anTe^ Tenrce pue (
saurl anle^ Tenlce aqt uaaAlaq aTI 6s.T
rTTA anle^ patsuTtse aqt ool Jo tno
sauTt g9 ul teer BuT.{BS se aules aqt
sT sTql saIqeT-rpA IsulB1_ro ai{l (
Jo
"t '(oot jo tno saolr gg) anTen
::t:' Parcadxe
(.)r . aq upr , rpntcp, aqt Jo 2.0 + uTrlt1o aq
(svi^r)oTgor parmTlsa aqr
2.3 alqer u1 paruasard se c
uoJrenba rLlr'Jo rn:o; 0T301 aq1 u1 'dyl,{
areuTrsa or uolrenba aql 3u1sn
go d:re:nJc" aql Bulssasse uT
TnJasn lsortr sT anfe^ fellpT sTqJ .OZ.A
Jo aleurTlsa Jo rol-ra prepuBts aQl 3u1a13 dq pue C
69.6 ,uolteuTurratap
'rldrrrnu Jo rLrr l rrJJaoc aQl 3ulronb (q dlleclrsJrers passa:dxa
aq uel) rettErs aqI .raqrnnu luaullclec.(q C
paIIaqpT s1u1od qlla
lds Ptep alalduroc aqt roJ dyj^l Jo sanT"^ pe^J.asqo
lsuleBe palEu]lsa C
soor.ls 1.3 arnBrg .auTI uolssa_rgaJ
aql punoJB f atlpJs alqe.raplsuoc
s] araqt addt slqt Jo uollenba due qtTn
paleadxa ag ol sT sV
C
vauv oI3o1 E17g'o - zo.T = A a.rarr^
C
qg.0-(3xv1+t) *.IT.osHrt * x Av[w *
nhr.zuvadv o_0I x 00.g .{VH
C
sI uollcnbe aIqeTJen anlg papuaEEora.t
aqJ q
 66
l'J arn6r3

ls/eul lVl,I po^Jasqo

o0t

(
m
@

J- (
*o
0)
A

3 (
-tl

oor t-< (

I
a

a
,r, te)
o,c.
"'' "i; /t' *"
t91'
/ t*'
e
,/
,,',/
C

e
C
C

uorlenbe uorssalooJ Aq paleutlsa JVy\l lsuleoe pallold

C
Jvtnl pa^rasqo
C
a
 00r

a a
o o
rm t-
m
@
o)

t
a I ot

; o
--.J

-o
o
o
o c
a
g
5 6
?c
o
gC
=

;c
o

o
;C
;
33
'n 'n
mO
oo
AC
a --
q)

;C
0r<
Iq
<6
11 J
3e
I
g
trC
-o
o<
rs

€c
.E
ox
ox
d'
@

o
o ge
o
@

gc
b-
===<=
'n 'n 'Tl 'n
A,r s F
'n
X
eC
=

lEig+,
-.'
-r1&6q;
I'6 C
\tg
xol
EoN
:ix
q)(,
!x
C
xt'Tl
3i'
>> -n
.Yl
C
'T'l
(o
C
IN
 TOT
E
C
arTtl pue ltrBLq qlTfr palnlTlsqns a'ran sdm Sulssru aw 'lealord
aq1
e
JouollB]npaq]f,o]VHdodqalqel}s^eaPeuSelleAe|JoPueTsTaqffoJ
sdeu asaqt Jo las alaTduroc lsoufP uY 'asod:nd
sTql roJ sdeu alqelTEAB
aS Sulddeq du:v sn) sdeo c1t1de:Eodo1
e
;-.1 .1 aQ1 Se (sa IJaS ( St{V) ac}Af

000'0q:1 aql ooJJ PaJnsPau se&

eaJe luauqDl€" 'sAef uI e
ztri u T
pa-rnsEaltr

SulxaPu] uJ cTls-tfalceJeqf,
e
] lead pooll aql Jo aPnlluSeo aql
e
S Y:IUV
tLr.lur{r?ec 1ue1:odur1 lso{u aql sT (vguv)
Ea're luamqclef,

rvEUV) earY tuaEqclec z'q e

'uolloe:lsQB rlaql -lol pa-rlnbar
aIqeI ul PalslI scTlslJalce'reqe
e
uo llPurJcrJttl Jtll t{l}l\1 rarltaSot I'0
aql 3o suo]1d -!rDSap JInJ a^1a .suollcas
SuTnoTTol aqI
e
dllluenb 1n33u1ueaut
e ueql laqlPr sdem uo'rl aIqPuTelqo
e
dl leclsdqd pue paleclldrloc aJoID

dllsealllslraf)e'Ie[qealdulsedqpaluasa:dalaJ1]sa'lnJeaJfuatuqe]eJ
.d1uo asuodsar fuauqclBC JO saclPuT Se Pasn aJe
e
duaur uosrla.r slrtt foc
salqe-t.re^ atll lxalu():) luasard aql uT'1apou fualDqalec PasBq
sclsdqd
e
t]uL]eadderqSlruleqlsfafaue:gdluasa:da:S"]?s}JaJcefeqcaq]
Jr; irLrros qSnoqt LV
-8u luoseal aATlTnfuT :O sa1d1cu1:d 1ec1s(qd 'Iaql 1e
of ullou)t are 1eq] dqde:3o1sdqd luauqrf ec
e
uto-1 I asuocl sar po() lJ acuonIJuT

Jc) s]cedse ]uasarda: sa1ro8ale3

aql .af,n]PaJ fuaBqcfec :e1nc11.red
aql sar'13 osTe qcTt{|,t I'0 aIqel
e
."- xapu
I ot pasn r]lslJaf cP.1 eqc clJlcads
eqr aqJ
u1 sE sa1:o8alec uan as oluT PaPT^1P aq uec scTlsJ'ral3e'1 Q

cvt^l IEuIJ aql u1 Sulreadde

uolfenba
asoql lsnl loli'dpn1s s]ql ur Pasn e
B sarr13 xauuY sTW
sr!ts lJ.alJeJ.Brlc fuaulLlcf Br aqf 3o uolldllcsaP TInJ
,Sapn]1u3eu PoolJ 3o uo}3PTJP^ aq] 3u1xapu1 uT ssaulnJasn IeTfualod
e
fuauqclec 3o sadf,r IB]a^as
3o Suraq se pauoltuaut aJe scTlSTJallsJPqt
padola^ap sr uoTlenba uolleuTlsa JVR aql qsltirl
uT J xauuv ur e
uoT1)ntolruI I'o e
SSIISIUATf,\TWEf, INAHEJIYf, O XSNNY
C
e
 zoI

.uoTlBnba uoTssaJgar
r10laq ua^T3 suolltras .)l 'u sTql SuTsn uaqrrl
I,rdo.rdde aql llnsuoJ PTnoqs
Tenuslll sTql Jo Jasn aqJ 'g :a1deq3
lo uo ! lenba uorssa-lBa-r 3t{l u1 s:eadde
cT I s TJa I Bqc aql I Er{l sa I eeTpu
"p.r r rlrr IsTfalsB uv

.soleiurl r,-;, iJu I pooLJ

?l| Oril sd.)uoriijr I
l.r or^ldu
I duu r?rli) aijl i (, od erls r;, (
._r
'xapu1 edeqs luauqJlpJ
LT' A AdYHS ad eqg

sL)-1 n:) l?ar I

xapu 1 duerrg 9I.O a
-tit,rLiJLl .j!1,:, ,.,Jrilo dHVIIS
:iJPu l
xapu] uoltetueTd
urj,-r).'uocl
so,l po()iJ .{j rpour s I 'a Nr.ld
xapuJ dppea tT'0 IQOYd
c
tiloq ui?r dsr) pueI dqJ
xapuT lsa.roc
tI'o JS3UOC a sn Pu e.I
C
spooIJ Jo uolterrudB
aql aDUanIJur uP)
sI los puI] slror
C
J0 sd]l xapuT TToS ot u
:ado:d JecrBolorpdq aqtr rtvJ
xapuT -dBo1oa3 T-;.^
rt u 'I0s9 addl pueT
C
spoolJ selBnudtlu u1e1d
IJ e.{l uo Jo SJTon-toSo-l
(xapu1 u1e1d pooTd) C
.ISH
Z0I - 0 :aao adols .ra^TU 'o
3oIe[ .{q .t.1 L{] 1a aBe:o1g
x.aPuT alef
OI
6 'A
olos
TXVT+ a3e: o1 g
C
Jo uolleBedo:d rdlspJ
spooIJ lsi,{ ZEg - 0 :ano adols ra^TU
B'C E80 S
C
'1SJ^i ZEB
- 0I r a^o adols .ra^ IU /'a
ol peal sadoTs doats
adols a1du19 9'q
ESOIS

SilI S+ ad o1g
C
I alqlldecsns
3u !po-oIJ

-j_t()ilt af u
i [tsJu1er dep
C
I ]uaurqclec
IeJu
rJ
luf, d^eeq luanba.rJ
cua r.r a dxa s luauqrl pO
ttrnur..txgu
rPnuue u eal^l ' E'q EVEdV+
G
IfBJUTBJ Tpnuue aBe:aay
t'a uw
spooi; raBBrq acnpord
a]E{trTT3
cq
I Q13ua1 I!ee.rlsuTE^l cu
orls sluauqrlec :a3:e1
eal e lu aIIILlcl eJ 7,n
I JJ'{

V3f,Y+ )ra6
!J
C+
'-
I

cvH uq aJUanTJuI
uo11d1:csaq uoTt cas loqudg d:o3a1e3
a+
*L
 prints obLainable from

Seksl Publikasi
Geological Survey of fndonesia
JL Diponegoro 57
Ba ndung

It is recommended that the user strould, if possihle,

refer Io the
original maps of thls AMS series to estimatt'ARF.A
since it is
sonetimes very dlfficult to draw catchment bounda ries on [he black and
white copies.

In Sunatra a ner{r set of l:50,000 6opographic

maps Is currently
being produced' At the rime (April
- M"y rgg2) rhat nnps were
required for this project the only ones availabre
from this new series
were for Banda Aceh, nost of surnatera utara
(North sumatra) an.
Lanpung, and a few in Sunatera SelaEan (South
Sumatra). Thr.se maps
can be obtained (with suitable authority) from

Bakosurtana 1

Jl. Raya Jakarta-Bogor Kro 43

Ct bi nong

JawaEan Topografi
TNI AD

Jl. Gunung Sahari

Jaka rta

The remainder of Sumatra was covered w.ith an old

series of maps
aE 1,100,000 scale and lhese were obEained for Ehis projecc
Irr:m
JawaEan Topografl (address above). It is rr)commended Ihat. il-
possible, the new I 50.000 series maps be used to obtain
catchmt.nt
area in Sumalra.

A11 topographic naps of Java and Sumatra usecl

by [his pro jcct iir(,
held ar DpMA ln Bandung.

The rlver basin area (AREA) should be measrr:r,(l wiLh a pJ-anirjr,.Ltr

and expressed in units of square kilometres.
t
) D.3 Main strean length (MSL)

Ilaln strearo length (MSL) is def ined as the Length of rhc .J

orrs,-:; r
rlver channel upstream of the gauging station as defined on
[he
I:50,000 topographic
'^ps. The nain stream length is measurcd witlr
dividers set to 4 mn' Elsewhere, on the l:100,000 topographic
rniips.
dlviders are set to 2 mn. For both map scales MSL rs calcurared
thus:

MS.L NDx0.2 km

where,

ND Number of dlvider sceps from gauglng station to the top of

the longest trlbutary as defined on the map.

Dividers should be used to estimate MSL in preference to

curvimeEers as the daEa set uded in Ehe regressions
was based on MSL
(and also slope measures) obtained with div.rders.
Dividers shourd
lnitlally be set as close to 4 mm as posslb.le and checked
against a
nlllinetre scale over at least 100 mm before and after
use. A
correcEion may then be applied to ND to a110w [r)i- s€tti.ns
errors.

D.4

CaEchment average annual rainfa 11, AAR. was obtained from 'yean
rainfallin Java and Madura 1931_1960, (Inst: itute oI Meteorolog_v anC
Geophysics) which contains a l: I ,000,000 sca le
map oF.la,,.a witlr
isohyets of average annual rainfall and frour
'Mean raintall in the
islands outside Java ancl Madura I93l_1960, ( Insti
tufe of Meteorology
and Geophysics) whlch contains a sirnilar map
for Sumafra at a scal p of
1:3,000,000.

For Java the procedure used Eo obtain MR was

ti rstly to enra rr,.
photographically the rsohyetal map to 1:500,000
onro transparenr
paper' seeondly the L;250,000 Eopographic *,ps*
containine the

ttl: operarions Graphic (Joc) Daps ar"

and ',t-3;390.j:ir:. u*i*'J tro"r"il"'"
i,., ot".u
Use of the col0ured original rEpsr if avatlable, is 00."""
more sarisfacrory
 catchnent boundarles, were reduced to 1:500,000.
Cabchment botrnriari Pq
were Eransferred from the topographic
Eo the rainf all nrap and rr rej_rs
between isohyeEs estimated by planlmeter.
D

t rn surnatra the procedure wa.s similar

boundary overlay was photographically
except the catchmenI
redtrced to the scale of rhe
t rainfall map (l:3,000,000). A1 Lhough
for large catchrnents, the srna11 scare
ooR was estlnared by pranimerer
of the ralnfarl nnp uade ir
impractical to use this raethod on
) other catchnents. AAR was estimared
in these cases eitlier by eye or by
counting squares on mm graph
paper' The technique for counting
D squares is ir rustratc-d by rhc
example for ApBAR be1ow.

) D.5 Mean annual_gg$_ngu,_SgJglgg!

I day raj.nfalt (APBAR)

) Mean annual maximum catchment

I day rainfa1l, ApBAR, is
calculated by nulciplying PBAR, Ehe
) nean annual maxiurum 1 day point
rainfall for the catchnent, by an areal
recluction factor (ARF). pBAR
is estinaEed as follows.
)
An isohyetal rnap of mean annual
) maximu.m l day rai nfa l I (PRAR)
lli s
been reproduced from Irish (l9gl)
and appears in this report ()n
Figure l.l for Java aE a scale of 111,000,000
) and Figure 1.2 F,rr
Sunatra at l:2,000,000. ConLours
of pBAR are at 20 mm intervals. Tht,
recommended procedure is to
draw the catchment bound,lfy,tn t)ri5
n;,p
and obtain the average catchment
value of pBAR as [o]lows.

(a) Draw the catchnent boundary

on ths 6ppropriate i :25{,r, O0()
Joint OperaEion Graphic (JOG) ser-es
map. Blacii arrrl whi,.o
copies are available from the Geo1ogi.cal
Survey of
rndonesia (address in Section D.2)
uut use of tirr. ,rrigin;rrs
is uuch more satisfactory. It is necessary
to draw blre
catchmenE boundaries on these
1:250,000 scale maps since i t
is not practicable to reduce the 1:50,000
or I:100.000
catchnent map to l:1,0OO,000 or I:2,000.000

(b) Reduce the ropographic map, rhe ca tchme nt bounda

ry a nd , i f
posslble, a lenguh of coastline to l:1,000,000
for Java or
l:2,000,000 for Surnatra eiCher photogral-,hicaJ. ly
or hv s,rr-,e
 b
D
t (c) pBAR is deternined by the welghted
contour values where the
average of the rnap
proportion of the c;rtchmt,'t
D betneen each contour
is used at; weights. Thrrs
area
givr.,. irra|
per cent of the caEchment 4O

I 100 and t20 mn and

falrs; withi n trre c()nrour
the renalnirrg 60 per cent
the 120 and 140 mn band,
ba'ci ,i
fa-1 1s within

I ttn
'v A t tn -f bU
- rru
PBAR is compured
as;

I
x 130
100 - LZZ mm.

(d) rn certaln areas

D the contours o1: pBAR are
widely spaced and
partlcularly for sma11
t completely in the area
catchments, interpolation
to obtain the best value.
For example a catchment
between the 120 mm
is requi re.
lyin,e
contours (ie 130 nrn band) and 140 mrn
) but closer to the 140
shour'd be given a value mn Iine
of pBAR between 130 mm
the actual vafue depending and 140 mrn.
) on thr: posiLion of the
catchme.t.
rf the catchnent is large
) a planimeter: may be used
to estimate
the catchment average .BAR.
rf the catchment is snall
counf squares on mlllimetre.,graph ir is berter to
) paper.

which refers to point

) 'BAR, rainfarl, is conVcrLe.rj
to catchment
areal rainfall, APBAR,
by multiplication b;z an
(Arr; ' To date little areal reduction Factor
) work has been done o11
Tabulated areal reduction ARF,s i n In<loncs ia .
factors givr.n in r-he Binni(l
'ReporE on Hydrology' (rgs0) a.ci partriers
are fron work try Dr Doern.
on a 130 km2 area situated drrring rg:3_25
near Jakarta. This table
0 to 2oo kn2 for duraEions covers []r,: range
of 30 ninutes to 24hours.
be described as hardly I{hirst rrris.r ;rn
satisfactory for Lhe purposes
wlth catchnen.s up ro about o{. fhis 5;51161,
20,000 kr2 i"
available at the tine ".:;, ,, al I rhar w;r.:
and pref erable Eo inported""s
since ARFrs a.re 's rures For ,AR.D
rela,ionsn,o, ,nj,l rainfa,, r.gi n,, ,r
;::;j;:::":j."i":::.'::;:1, is

that they are perhaps not ::":;"",-,'i":;ll:,

unreasonable. The ARF,s ",
used in this stuci'
are therefore based on
the work of Dr Doerma.
of any inaccuracres in In any case the effect
these ARF's w111 be eliminaLed
regression equation foi when using the
esElEatlon of |,{AF provided
the same ARF's are
 used in design as were used i n the devel o pment of the regress i orr
equation

CaEchment area ARF

)
KM

1 10 0. 99

I l0
JU
30
30,000
0
I
.91
1q') - 0' I 233 log I OAREA
t The relationship [or caEchment areas between 30 km2
ard
I 30,000 km2 gives a range of ARF,s between 0.97 at 30 km2
30,000 km2.
to 0.6 at

I Exampl e

t As an example of estimating consicler catchment 6l0 shown on

APBAR
Figure 1.2 (Air-Ketaun at Tunggang). This is locatecl across
) the
120 nn conEour of pBAR. rf.the catchmen. is
traced onro milrimerre
graph paper the followlng informaLion is obtained:
D
Nunber of rnillinetre squares in 110 urm pBAR band
= 200*
t Number of mlllrmetre squares in l3o mn pBAR band
*These f lgures are sub_Ject Eo sma11
= 40*
esE i.marion errors drre to r:i ne
thickness and personal interpreEation.
)
PBAR for the catchment is calculated thus:
)

) PBAR6 t0
(200 x 110) + (40 x 130)
(200 + 40)

) PBAR6 I13
1 6 mn

) '
The ARF ls calculated as follows (ca Ichment a rea 94r: tm2)

)
ARF6 1 9 1.152 - 0.1233 Iogys 946

ARF6 tO 0. 785
I
)
Hence APBAR for caEchuent 610 is estiunterl as the multiple of
)
PBAR619 and ARFOtO

)
APBAR6 1 6 I 13 x 0.785

)
APBAR6 16 89 mn

)
u.o River slope (SIMS)
)
Four indices of stream slope were considered. The fi rst oi tlese
) is ca11ed sinple slope, (SIMS), and is the,difference in height
between Ehe point of interesc and the hlgtrest point above bhe end of
t the nainstream divided by che rnainscream length (MSL). The ,highesr
pointt is the highest point on the catchment divide in the r.,ic j ni tv ot
) the source of the longesE fributary. L.inear interpolatjon of confo.rs
crossing Ehe river is used to estirnaLe the elevation of the point oF
) interest (naps as for AREA). The uniEs of SIMS are m km-I.

) D.7 River slope (S1OB5)

) The second measure of river s1ope, s1og5, is carcuraIec as s]r>pe

over the distance between lo"l and g5% of Lhe mainstream length
t rneasured upstream from the point of incerest. S1085 nr.t.r be
c.nsi,Jcrr.rJ
to be more represenlative of the basin as a wirr-rl c lhan lj IirlS h,,r-.,,,.--,,
t excludes extremes oF slope inherenE i.n SIMS. Sl,tEl,r r;.rs:thsLr-,-rf t*i l
F

a sinilar manner to that described for SIMS above. The rrnirs or Sl^q(
) are m km-l.

) D.8 River slope (5085)

) The thlrd mea-.sure of rive.r s1ope, sog5, is carculated as tht

s10pe over Ehe disEance between the point,rf interest anrl g57
0f rhc
mainstrean length. s085 rnay be considered !o be a measrrre.F
catchment slope whlch is beEween the extrerre srMs and fhe more
acceptable SI085' S085 was inEroduced when regressions i ndicat, rl S lI{s
was a uore signlf icant variable in inf I uenr:ins MAF
than S 1085. .fh.
units of 5085 are m km-I.
 D.9 Lake index (LAKE)

storage for flood waters provided by lar<es and reservoirs can

signif icantly attenuace downstreax' f l-ood pear<s. The degree of
atEenuaEion depends on such factors as the p'sltion of the
storage
within Ehe catchrnent, its storage/head relatlonship and, for
reservolrs, the operaElng rules. As ic woulcl be impractical to allow
for all of these factors in a stnple regressi-on model of troodine. the
lake index used here ls simply a measure of the proportion of
catchmenE area draining chrough lakes and res;ervoirs.
The lake index was calculated using the formula

LAKE _ Total calchment area upstrean of lakes (tr2)

AREA

For Java the total catchmenc area upstream of lakes was ohtai n
fron publlcations glving infornaEion on dams over 15 m higlr and are
available ar DPMA in'Bandung (DpUA, l9g0).
'J

In Sumatra.; the lakes on the ca tchments used i nth is study werr-

all natural and the toEal catchment area uPstreaul of a lake was
obtained from the Eopographlc naps (Section D.2).

The regression equaLion shoulcr not be used if LAKE is,ereati:r

than 0.25. Arso, if the total surface area of [he lake is ress
t]ra'
L% of the catchment draining through
the Jake, LAKI] is insiqni f ir.r,rr
and seE Eo zero.

' The range of LAKE is

therefore 0 to 0.25. However, as
logarithnic Eransforms of the lake index are requi red in
the
regression, zero values cannot be accepted anrj it is necessary
to add
a cons.ant to LAKE. rn this study the term w.ich appears
in the
regresslon ls (1 + LAKE).

D. l0 Flood plain index (S010)

This is ln fact a measure of river s).ope and is ca lcrrlatecl as

tlrr_
slope between the point of interest and l0Z of the mainstream
1er.rgth.
S010 was introduced as an experimental variable
under the hypothesis
Ehat a flood plain was more likely to occur where the river just
upstream of this point of interest is very f.lar. The unjrs of Solo
are m .Km _l

D.ll Geology index (GEOL)

InformaEion on geology 1n Java and Sumatra was obta j rred f rom the,
1:2,000,C00 scale map produced by Lhe Direcktorar (teoloej Inclonesj,r
and the UniEed States GeologicaJ Survey.

unfortunately the description of rock types on the maps was

insufficient to categorize accurately each type according to
perneability. Furthernnore. at the time of the analysis in the uK
no-one was available.-with suitable knowledge of rndonesian geology ro
provide assistanee. However. an attempt was made to classify the rock
types into three classes of permeability and the fraction of each
within each catchment estlmat"d by
"y". The geology index. GEol., wAs
ealculated thus:

GEOL = (3 x I) + (2 x M) + (1 x p)

where

I = f raction of catchnent area i mperme,aLrJ e

M = f raclion of catchroent area morler.r tel r' pt rm'rbl t-

P = f rac tion of ca tchnent a rea pe rm(.ab I e

D.L2 Soil index (S Lr_Ll

Soil maps at a scale of L:250.000 were ()braineci Irom

Lembaga Perelltian Tanah (LpT)

J1. IR. H. Juanda 98
BOGOR

Experience elsewhere suggests Ihat a merrsure of s:oi I tyl,, is ;r

useful but not highly significant variable i r the regressiorr

equation. Classification of the various soj Ls into groups of rrrnoIl-
b
b -
potential required considerable further
D Unfortunarely this was outside Ihe scope of
speciaJ sed work
the presenf s tudy
t therefore a soil i ndex has been omi tted.
and

I D. t3 Ig."9l_.rr9S" (I_qEI!l)

t Land use inform:tion for this study

was obtained as a series of
I:50, 000 maps detailing Land U"" tf..rroughout
Java and most of Sumatra
The maps. which are in black and
o white. were obtai ned from: -

I Departenen Dalam Negeri

Direktorat Jendral Agraria
' Direktorat Tata Guna Tanah
D Jl . Si s i ngama nga ra
t
-.ia
Jakarta

I An overlay of the boundary for

each catchmenr was prepared on
transparent paper from Ehe r:50'000 series
rnaps and posirion.cr o. tir,.
t Land use mps' The new series
of r:50.000 scarc ropographic rnaps for
parts of sumr tra (section D-2) so contain
some rancl rrse
^1
infornation' This was
) 'sed in preference to the Lancr Use maps
mentioned above whenever possible
as it was easier to absIract ancr
also rnore uptodate.
)
Land use informat'ion was traced
) through and the totar area o1.
forest determined by planiroeter. The
foresI index js c,rlclt.ttFrd Lrsi ng
the formula
)
F'REST _ Torat a,rea of fore:lggj)
AREA

ranges from 0, for no foresL cover

FOREST
, to I for compl ete
forest cover. In order co al1ow Jo65arithmic
r ransforrultion of tl,r,
FOREST lndex a constant of
I was added in the regressions

D.14 paddy ind€x (pADDy)

The paddy index was calculated using the [ormu.l.a

PADDY =
Total area of paddy (kn2)
AREA
)

The total area of paddy was obtalned from thc samc solrrce
an,l
estiurated in a simllar manner to rhe forest jndex. FoRFST cr.c..strri5t,,l
in section D. 13.

The tern used in the regression analysis was (r + pADDy).

for rhe
reasons given above for FOREST.

D.15 Planration lndex (pLTN)

The plantation-lndex eTas cercurated using the formula.

pLrN = fglg-l-='::-"jjlg!l:Eig! (t
'2)
AREA

The total area of plantation was obtained from the same sorrrcr.
and estimated in a sioilar manner to the forest index, FOREST,
described .!,n sect.lon D.13 and a constant of l added in the
regresslons to give an index (f+pLTN).

only abstracted for stations in Srrm,etra .anrJ tlrerofor-,.

PLTN was
could only be considered in regressions on that srrb-set of sIati.ns

D.16 Swamp index _(SI^rMp)

The swaurp index was calculated using the formu.la:

SWAM' _
Toral area:i_:grp jf.r2)
AREA

The toLal area of swanp was obtalnecr from Llrc, same sourct
ancl
estimated on a si.milar rnanner to the forest index. FoRLlsr. cJescrlhec
in section D, 13.

As with the other land use variabres the term usea in

[he
regression had I added (1 + swaup) to avoid zero valrres in logari
rhirr it
transforrnatlon.

D. 17 Catchnenr_shag_ r$Sx_ l_Qt4!I)

'rt night be expected

thaE the shape of a catchment wourd
influence efficlency of flood qeneration within
D
D
t other basin pararneters being equa1, a long thin catchmerrt oifer:, mrr,:
possibi. lity of f lood atEenuation than one of con1..;1q I slrape. 'lJtr. .slr; Ir.,

t index used in this study was.

AREA
SHAPE
D MSL'

D SHAIEis effectively the raEio of catchment width to length. MSI

is the nalnstream length as defined in Section D.3.
D

t
D

t
D

t
t
)

t
)
 ANNEX E. TEE PEAKS OVER A TERESEOLD HODEL

r,,r Introduction

Chapter 4 described the application ot the POT model ro estimare

the mean annual flood-. The roode-l described was one in which a
Ehreshold qo rrTas chosen and all peaks exceeding this threshol d j n
Ehe complete years of data were abstrac[ed. The resu] ti ng pOT serj es
consisted of M floods, Qi, froni N years of data. The theorv behjncl
Ehis model is considered in this annex. rt should be noted, however.
Ehat this is only one of unny possible POT models several of which are
described ln detail in the lJK Flood Studies Reoort
(NERC 1975).

Two variationsEo the basic rnodel are considcred: f irstl_v wtrer

data f rom incomplete years are al so avai lable anri st'co1rj lv wlrc-re a
historic series of events has been recorded.

r-,2 Ineorv

The POT serles of flood magnitudes are drau"n from a conditional

distribution as only floods greafer fhan a threshold, go, are
lnc1uded. From this dtsEribution it might be observed that 1cr. of
floods exceed a higher va1ue, q. but jt would be wrong ro state that
LO7" of. all floods are greater Lhan q. The conditional statement rhat
LO7. of floods greaEer than qo are aJso greater
than q, is mrrch less
useful than an uncondiEional statement relating to a1l floods. The
nethod of deriving the unconditional statement i s basical iv s i n;,1 r

Suppose tha t in a gi ven por sample ser ec ted to exc eed a ,i0c0 rn js '

threshold an average of 3 peaks per year are inclucled and rhat oF

these floods l0Z exceed the higher threshold of tr500,n3.-1. There is
a probabllity of 0.3 that 4500 m3s-I will be exceeded in one _veirr or
thar the return period of this event is 3.33 years. This c()nc{apt ('.iln
be expressed more formally by making ass.mnt i ons abo.t the
dlstributions.inherenE in the p0T model

The dlstribuEion of flood magnirr-rdes in the POT series is asstrmod

to be exponential. Thus, the conditional probabi'l ity statement that
Ehe probability thar a flood Q exceeds q, given that q is greater Lhan
the Ehreshold qo, can be wrltten
Pn(Q > qlq > qo) = e-(q-qo)/0 (r)
b
It
It where B is the sc3 le parameter of the exponential
distribution an<j
the rhreshold, go, 1s the location paraueter (the mean
of the
t distribution is given by qo + B). For convenience this probabi
wil-1 be written as PR(A lB)..
li tv

I Glven tha-t i froods exceed the threshold ln

any year rhe
I probablllty that r of these exceed the value q is glven
binomial distribution: -
by the

t PR(r peaks t qli peaks) = (i) (pR(AlB))r(l pR(Alr;yi-r

-
This conditionaL probability can be expressed in an
(?)
unconditional
e forn by assuming the probability of i floods occurring
given by pf. As 1 ( i the unconditional probability
in any yt,ar is

t q occur in a year is
rhat r peaks

t PR(rpeaks>q)=;
1=r
pR(r peaks t q
li peaks ). pi (3 )

I The probabiliti.es, pi, of i fLoods occurring

assumed to come fron Ehe poisson di s
in a year can be
t ri bu tion
I pi = e-tr Ii
i! (4)
where L is the nean number of exceedences per year.
D

t Combining equations 2, 3 and 4 gives

pR(r peak, t q) = .,-\ ii

t -l-(r)(PR(AlB))' (1-PR(ala;'
1=r
Redefining rhe limi ts of the summation
'
ii

t .,- l. f.i1 r
( i+r) !
i4r f\J i+r\
) Nore (J'^ ) -- -rl ^ /.
|

r II

) ; rj(t_PR(AlB))j
j=o il
Note ez = f z]
) j=o il

= e-\ \r (pR(Aln))'
1j/:
) e)'(1-PR(AlB))

_ "-rpn(AlB) [rpn(els) ].
r! (5)
Comparing equations 4 and 5 shows that the djstribution of peaks
exceeding q is also a Poisson distribution with mean \PR(A ln;. Tn ,r 'l'

year perlod TtrPR(AlB) floods > q would be expected to occr.rr; wlrere

only one flood t q is observed in T years q is then tlre T year flood

r\PR(Als) = r

Cornbining with equation I, where q now represents {l(T), the T year

flood, gi ves

a(T)=eo*Bln\+BlnT (6)
which allows the estination of any flood Q(T) from the POT serj es .

Using the POT rnodel as outlined above it has been assumed rhat
the number of threshold ,exceedences per year is dlstrtbuted
accordingly to a Poisson dlstrlbution and that the floocl magnitrrdes of
the POTseries are disirtbuted exponentlally. NeiIher of rhese
assumptions ls stlctly true; however, the discrepancy they j ntroduce
is likely to be small for low return periods where the flows are not
very much greater than the Ehreshold value. One such flood is the
mean annual flood.

The distribution of Ehose annual maxima which exceed the

Lhreshold qe ian be deduced fron the POT model- and shown tr> be from
a lype I extreme value distrlbutlon. On the assumption thar rhe
entire annual maxima have the EVI distrlbLrtion the mean is

p= MAF = 9o * Bln\ + 0.57729

It will be noted Ehat this implies the MAF to have a return period of
1.78 years whereas in the Seetion 1.1 it was noted that the l,lAF From
the annual naxlmum series had a return period of 2.33 years. The
difference arises fron the fact that annual maxima approach ignores
all exeept the biggest flood 1n each year but the POT method can
include several floods fron a single year or no flood if the aunua'l
maximum is less than the threshold. The annual maximum method
therefore only considers inEervals becween years with floods of
specified magnitudes rather than the lntervals between the floods
Ehenselves. The POT approach is in fact the correct one, although in
practice for large return periods the difference is slight. Thc. rwo
reEurn periods TPOT and TAM
 Tau f, I
Ll - exo(- ___ /rl-t
J _
Tpot

E'?
L ' J _lgpeUggJsere_"J_le!.

s It is often the case

there will be many breaks
that over the period of
in the continuity of
operation ot a gauge

a model as outlined
above only the conplete

i::"'::'",in';:":":;"'i""""j;
the record. In the pOT
years of data rrere
Although bv carerur
rrsed ancr
choice or fhe srarr
record can be reduced,
=r wasted. In the pOT
a sreart;:;r":;';:r::.i;i;;
J:i::i"rj:,::
F estimared rron rhe
exceedence
model desc:
series or ,:j:::
of the threshold' is
r;r:"rj:;;"r..,.u1
.;: .f,,"."o"
lnclusion of peaks froor unlikely to be affected hv tho
incomplete years of
=D estinags6 more reliably data; in fact
if more peaks are used. parameter it shorrr(l h{_
3 other hand could be greatly
over a rhreshold-occurred
influenced 1f it r,r'as
\ on the
assumed that no pc.aks
auring a period for
record' rt is recomnended which rhere is no
therefore that peaks
=r are used to estlnate
F but,that I, is estinrated
from the entire recorcr
iD f rom the compl ete
_vea rs
;:.'il using
.;1i,.,i. :::;';.:'-:::::'"s oo,n'..'o,r ,e vea rs onl y
e L.a nrstoric floods
---.---_.---::
I Historic floods are
t srone' or building'
often recorded as flood
The base of rhe
of as a threshold exceeded scone
marks on a I lood
o. our ldi ng ca n h{. arorrgirf
by a1J- the marked
two flood series are f loods. rn srrch
D available, the historic a case
over the high threshold series of n,exceedences
qj (corresponding to
I and the recorded serles
of n exceedences over
the lorrest possible mark)
the .lower tl. re:_:hr> l t,, .
rh such a case the
parameter B can best
) be estjmatcc-.r h,.,

;'' "in(ci 1*#_'",

'F, -
I nr* n.

l. should be estimated
frorn the recent
serJes onl
 4 variation of t_!.
lpr merhod

In the method outli

chapter o antot n"o in the previous
e threshold. ., secf ions -rnd
in
likely to 6s.-,. ' Yo' was fixed at a ,
se'l ned
Fron rhe '|ha'|
t""',;":il::r€rcir'lrsen'
""""';J;;:::ffTj":::: :'""'o*u'""'
exceedenc." er
l::
rear' and jn" o"ranerers )., ,r".
variatlon o, "lit;:., B' the mean exceede
are c.slir&rted'
uethod the numbet In a
and Lhe rhresr "r;;"';"","te per vear is ch'sen

modiri""'ion"; :;jt;:;"'. tj-;;r"

;:: ;":"1.,'*'"::;"*t,";1':;:'"-::':;:",';:'""
or rhe uor'*nt
or a rhresholdction
":;: .;:,.,",
j:;,;i:j,
ff: ;:-rj
J* **njj *"Tj:TJjn :; ; :. " ; :;
re'suJ ts rrom
under the headj;",J;t;.:r';":lt "t rhe a 0.,.". peaks and then'
variarton of ar:---::^t,analYsis on a resrric."o't"o
tu'nod is used thr
year is reducuo" ":i":ia" s
::":::::"
nerhod
ting rhe.rr,'",,"'J,
or res,.;^::-1"'."q"' ;1i"t.,:
rhis is ,r":,::.:.;
dirrerence ir a "".r""i"r; ;;""I.rre
" i'r kes'j i'['|'1 e
ma

;:::j',;;j;i;i',i".;"'::i':
:;::":"::..n j".: onsidered
to be:l:::'::;:'*'ijT"Xj;::j
rirsr case.
::

=t
il
t
I
I
I
)
a
rtt
*.

b ANNEX F. GRCITTE FACTOR ANALYSIS

b F.l Introduction

b Thls annex conslders the analysls

f requency growth; factors recommended
behind rhe design flood
1n Chapter /.

a rndividual f100d frequency curves relating

period nay be draqrn for any station for
irood peak to rerurn

e daEa exist ' This has been described

whlch a number of years,
in chapter 6. Howerrer few
sEatlons in rndonesia have a long enough
e flood frequency curves to be drawn wlth
year reEurn period' For f100d design
record to enabl e i ndi r.,j 1q3
confldence above the l5_20
61 1

3 purposes, however, engi.neers

are commonly lnterested in return periods
ln excess of 20 years.
How then can estlnates of
e high return period floods be obtained?
This is achieved by poollhg all the
dara avatrable and obraining,r
consensus on the behaviour of catchments
a at high rerlrrn periods.

As f100d frequency curves differ greatly from catchment tr)

a caEchnent it is desirable to scale
the,indlvidual curves prior rrr
pooling' This is ac'ieved by using
non-dlmensionar frood freqLr.rrc'
a curves (growth curves) in which rhe f100c
uragnitude scale is divided

t by an index flood' The index flood

oEher return periods by dinensionless
factors' The index f100d (rtre mean
is then relatecl to floods of,
rrultipliers or growtlr

I annuar f100d, MAF, in tl-ris

study) is assumed to take into account
catcirm.rt variabres sr-rcrr,.rs

t area rainfa11, s10pe eEc. However,

'
r*y sti11 have some dependence on
the growth fac to rs themst l vc:s
the catchment var.rables.

t l-h. r.
A coopromise is therefore requi red
in the pooling process s..lr

I
( l) Sufficlent catchnents a re grouped
to enable tire precJict irtn
) of high return perlods floods

) (2) Any signiflcant differences in growEh

tactors dn, I L) t
nature of catchroenEs are not hidden.
)
r€
€
The approach adopted
for this sttrdy was tirscly
€ single overall dimensionless to constrrrcr a
growth curve from all
stations in.rava
F and sumatra' This curve
above where all stations
satlsfies the first criterion
are pooled to enable estimation
m€,ntior,.,]

F return period floods, but

does not pernit variation
curve wlth external factors.
of higfr
of the growbh

F.2 pooling of growth.,curver

=t,
€ This sectron describes how
indivrdual station frood
curves vrere pooled to form treqrre,r,_v
an ,average,growth
stations used in this study. curve for a1l
=fit
E conbined growrh curve
for al1 starions was consrrucr.d ,rs
a"rr.::,
IED
(l) For each station a non
dlmensional growtlr curve
fron Ehe f100d frequency was con:iJ:rrr.Fr.ci
=E record by rhe MAF. In
curve by dividing each
f100d on th,,
each caSe the growth
€ a series of points _ reduced c urve was s trt 66..1
,
variate and assoc iated "
O/lfAF
(2) An average growtrr
=a varlaLe and mean
curve I,Jas Drori,,^ aJ L,-
th e mea n redr r- ,:d
Q/MA. fr;; ;::'::;:r::r':::;:
of reduced variate. The n each intervit
=D intervals of reduced variate
ir -1.5 to -1.0, _1.0 to _0.5, _0.5
to 0 erc,
ugcj e;. 1.,.

a (3) with the individual

staE,ion record 1r:ngIlrs
years, the smoothed average rangi ng rr-om .:
growLh currrr,was rrelL
defined
LU

e a return period of about

insuf ficient for unny
' r'/as extended
100 years. Becatrse
thjs is
design purposes, the growth
crrrrv"
Lrp to

e the data set

by considering the five
largest Q/uar rrnlues

e
ll
a
I
 and plotting the
f ive Iargest
independe", 'nu va I rr,.-s in ,, .srr.posedly
""ro;:*:t
(4) A general extrem
Lst 5, secEion, ";j;"rJ::Jj func rion (Flood Sr ud i es Repo
. ;, rt
(2) and (3) above ro the points obta .

such that ined in sIeps

Q/Mar u * a rt:jal

where,

v reduced variate
u lntercept of fltted
curve
scale paraneter
k curvarure or,rr,lo'::.::tted curve

ParameEers u, c and k were

obta
approach. The squa res
ro r I owl nB pa
r
;; ;:";" :'" :" J:" :j
., -: :::' ;"":_::. had the

u = 0.g4g
q, - 0.219
k = -0.214g

It should tu*"mbered thi

Points whicir tted rh rrt 11g
"o,o"
T:;:: ;:iT*:J':'l:,,":: ;:,,:
f 1

rerurn periods, trlarl_y at hiolr

1t was fitted. [his curve and fhe
Furtherrnore, the points to wh i r. ir
ornp5|1 C L i-,,./( .

)
 ry
a
a
showed conslderable varlation abour the mean. The
reader slrould
bear i n ni nd, therefore, th.a
this and other smooEh growth
C

whlch appear in thls annex in curves

a conslderable scatter.
fact represen[ a group of points
wiIh

e F.3

a The signlficance of any variation

catchnoent cha racteristics (criterion
of growth curve shape wi th

e deternined as follows:
(2) rn Secrlon F. l) was

Er A list was drawn up of Ehose

llkely to lndex the shape of
charactt,ristics consjdereci mo,;

t' (l)
[he growth sg11,,._;

Locarion (The two geographicalry

t Java and Suma tra).

convenienr regions of

I (r\
\!/ Catchnent area (Anee)

Ir (3) Average annuat rainfal1 (AAR)

e (4) Mean annual maximum caLchmenE I day ralnfall (ApBAR)

I i nto
For each of the above
two groups (Java and
cha racIeristics,
Suma tra, Iarge
catchments were d i ded
AREA and smalI AREA
'.'i
etc ).
D using the same procedure as
described above in Section F.l
pooled growth curves were
D produced for each of the
tested to see 1f Lhey were two gr.ups.r.ci
signif icantly dif f erent. .A
indicatlon at this stage resulted posi rl vt,
D in the catchments bci ng d i vi
agaln (lnto 4 groups) and dt,rl
the test re_appJ i t,d. Thus
relationship between any the
D characteristic and growtlr curve
be tested at its nost elemenlary slrape rrr-rrrr,J
level (2 groups) and if Iorrnd
signlflcant, further divisions
) of the data set reveared thc.
which the ,relatlonship 1 i mi r ro
could be adequately defjned.
)
The flrst sub-grouping of catchnents (accordlng
l0catlon) therefore had one to catchmenc
pooled growth curve for
pooled growth curve for Java and one
sunhtra' These curves are
sh.wn [oge[her
 with the ,a1l
station, 3urve
sta'ist. ically on Figure
l

significar dif I/hether lhere

,

and sunarra is f erence o",t't'een Srowtlt is a

rd 1n secrr."
"on"ia".utt curves in Java
a.r.

€
F c
Fig

::::-#i:;l;'"".i"'i;
a, c hme. .,,i;:: ;; r:,,:#*.
"n""'n" res,.L.c

ij:,,}":'""""-".:,I;,:J
ji:, :;:,,
Figure F.4 is
€ more ir
suggests smaller 1n th; i t shows
Hr catchments. The
revea-r-
catchnen,ntut""ting
result" of
have a
"r"uo".'t
o" the significan
the difference
6lrowth
a trencJ which
curve than larger
r.'srs in section
E "r,u.n.l-t'^:"::lts
rarge catcr hments,"
0..""."';.;;tL'
out"ten growl
F'5
ticalJ.y slgniar""'lt.Cttrr'"''t 'rn sm;; I I .r.d
"r"ar"-,n""
E F.4 _signiflcance
tests

tit There
applied to ,ttt "
i
number of

tr
I

"il. re s wn i c h
rna v
d i s t ri b
u r,. j:;" j',:;": :T :'::;: ;
erence
be

"""' :, :"*:,.
-"u dtro Lynn' rc7q. of two
Parauetrlc x2 of
and Kornogoto'o-sti.r""";""a""tn,tttt""' the p6p-
they are ind,
E .,lJr.r J;i;:::':' ve'L h e ad va n'[a g e L ha'L
no"ition'
Q' i:';J; :ili*'l'
P wiEhou'
dlstribution"t The'se fests'
the 12 .."'"'
Ko ;::,,,"::' ;i,:l::
it resulrs"nd
when
Lynn, l97g),
#i,.: .:T:;,";: :f::::;:
growth curve
di f e.un,
"t'otn to gi '''o :r irni la r
a;:tJ^trt-to
n" Kolmogorov-Sn ' f !iaticrr' (ster'''irq rrd
ir the 12 r"",
,a ,nui'rtfoelr
5esr does
test is easy
visualise.
,'""urot",
:J,'j*iiji:';' ;-;ij':;:, "
I rhererore used
as the basis ":
of comparison
rr r;:: "",
.

I The Kolmogofov_Smi
dlstrlbutron"
_-^
rnov tes _
"::;:

t "torov-sni
::Tj;j: ; .l::;:: ::T*;Ti:*:#_1i':, ::,:.", J;,
each sample ",
ranse rhis
)
;:,Taji:jil." :::: i::.:'r: ';.. ';:
) roughlv ,nu
""," :;:",:.,:
:'r:';;;;J:';' "'""::i: ",
"i;J""'": varions 1n each
_:j;:*;""::;::
e
I Regionalization of growth curve
I
a GROUP NUMBER OF
STATIONS
STATION
YEARS

I ALL
JAVA
STATIONS 92
48
.too1

I
615
SUMATRA 44 \too

D
O DATA POINTS FOR ALL STATION
LINE WITH
NUMBER oF sTATroN yEARs
. coxr^,"ui,*o
D I N-8.33 points ptots tess tnan y= _; i'"'--
+
t FIVE HIGHEST FLOODS tN 1OO1 STATTON YEARS

l
t
)*
E2 .4
)--
F
o
)

50 100

RETURN PEBIOO

Reduced variate
Figure F.1
 Effect of AAR on growth curve shape

GROUP AAR BAND NUMBER OF STATION

STATIONS YEARS
mm

SMALL 1950- 2710 46 sl2

LARGE 27'tO- 4950 46 489
ALL 1950 -,1950 92 1001

rl ll
20 50 100
RETURN PERIOO lyears I

Reduced variate y
 _ro

Effect of 'APBAR'on growth

curve shape

GROUP APBAR BANO

NUMBER OF STATION
mm SIATIONS YEARS

SMALL 62-93 46 479

LARGE 93 -162 46 s22
ALL 62 -162 92 rool

It

=
F
o

t
,

)
lr
= ir"
RETURN PERIOD lyears
/

Reduced variate y
 'AREA'on
Effect of growih curve shape

GROUP AREA BAND NUMEER OF STATION

STATIONS YEARS
km2

SI||ALL o'43 - 600 46 537

LARGE 600 - 12430 46 464
ALL 0.43 - 12430 92 1001

MEDIAN CATCHMENT
AREA km2

lI.

=F
o

YP
/. f,
Ya

/,{6
:--

b
D
/,tlir,i
D
_l I
500 100'0

D RETURN PEBIOO lyears I

,,tld
) _-T
7

)
D

) of Q/MAF values less Ehan or equal to each interval are determined

for each of the two distribuElons. The two cumulative f requenc v
) dlstrlbuEions are obtained by calcu lating the proportion of fhe
total nunber of poinEs in each inte rval for each of the fwo
) distributlons.

) At each i,nEerval, the cumur.ative frequency distribuIions

Q/Map
are subtracE,ed, and. the absolute value of this difference obtained.
The naximuu of these 28 dlfferences is the Kolmogorov-snlrnov
test
staEistic "D".

Table F.l gives the values of "D" for t.lie pairs of

Q/Uaf
distrlbutions under comparison. The problem now is one of deciding
wheEher the differences between these distributions are signi
FicanL
or not' Norrnally it is possible to use sEandard tables [o obtain
d(0.05) (the 952 confidence limit above which the distributions:rre_,
dissimilar) if one set of observations is compared wtth an
independent, completely specifled, conElnuous dlstrtbutton. Tn our
case we are co*paring two discrete sets of non-independent
observaElons which invalidates use of tables. A simulaIi-on approach
was used to obtain estirDaces of d(0.05) and is orrtlined be10w:

(1) For each comparison a simulated series of annual maximun

floods was generated using the general extreme vafue
distribution function with the parameters u, a and lc
obEained from fiEting to a1l 92 srations (Section F.2).
This dlstribution function given in the LK Fr.od srud,,,
ReporE, Section I .2.4 Ls:

F(q)
"-[t-t(o-.|)/a)r/k
If this expression is lnverLed and F(q) replaced hy U, a
randon nurober between 0 and 1, the flow generaIion
function used in this simulaEion is obtained I
u+9 (l (- 1og" U)k)
k

whe re u = 0.848
= 0. 219
k = -0.2148
 €€

r?\
\-/ rne number of
€ maxina general
same as in
6crrerated for each
tho ,.. ]' s tal
distribution
^
unde. ";;,tt"" """ the
values o, " ,";;.'
€ o snnerated'J:t:;'i,j"'^i"-' example iB
represenr rn'ure
€ ;; :JT:. j:
;:;:. ::; ",;,iljTi:.":
€ The resul[ we therefore
t=ilt'-i
-.. u' uy L'OS x Q-
tmed
"i::r': basic data:
" o
ir u*.1_> J x{i,nud.
generared fro a si
€Gt "::":j: J
"."::J::
;:; r ::_
:a under test.
TJii:"T"'"Ti fnr#
(3) Having genel
eEt a Kolmogoro.tttuo
a series for both
r-Smi rnov distributions be ing fis

ct
tes L was ce,
Performed ancl ,,D,,ca1
(/,\ culated.
\-./ lntS prOCedr

€r
repeated loo
es rimates ,r'ToI. " rimes to give loo

Er \J,/ rne IOO Valr)es

a\l
, of ,,D,,were ranked
l
obtained. This and the 95th highes
.^",] becomes our t
rt is an esa',l-t:-then ^]^=^'e esrimate of d.,
€ raken. rr,";",'rr:":il;"oalv too samp1"" ;";.to;::l'
-^_-*'-l
E signif icance 1evel
expressed as
1eve., ac,. anY fi>ced "ttor of this es trmat!ci
va l rrc of
d can be

€ S.e. (n)
\Y / = ,z rP(_LC
'N./ ,
whe re ,

=lt
a
p rhe rrue signifi
N number of samp,""
"::;.,to""t (0. e5)

b Subs ti tut:.

a J,.ffij :ii:"i :"' ' and N gives a sta.d: r'r

":' ;;ji: i :T
e ;.,: ::J,
be
f:i ir
ob..rn"j ;_.:::: *jl*j1,,..Hi:
e5.z confidence
inrerva.
e32 to e7z.

and e rr r"r','.o1, ;,T";rr*,:""^ *,;; .;:

iD m "',',^Jr;;1,;;:",
the simulations. 917
(6) Table F.l
gi.,-- o(o'os)
r
obtained frorn
D procedure
,"-=-ea"h this slmuLation r

error "omiarison. ^t"; ;t;"'-s

one s t".,a".a' Also given is d(o.osl
t
D
 ?able F. t.
Kolnogo rov_Snr:l
rnov tes t
res_u_]--t_9
Conpa rlson Groups
Nunber )
-(u.05) d
1s.05) Accept
+
i, - S.e. Hypo th es
scations
.JA VA is
A n^
v.U25
Suna t ra 0, 056 0. 055
stat{ons Yes
0.059
Suall Area
Lar8e Area 0.080
0.060 0.057
0 .064 No

sqglt Area (t)

v. vO)
SmaLl Area 0. 086 0. 080
(2) Yes
0.090
Large Area (i) 0.072
Large Area 0.086 0.086
(2) Yes
0.089
Hypothesis:
1s no slgnrflcanr
,T""t
disrrlburr"-l-:: dlffe renc e between the two
Criterion;
D < d(o.ps)

t-
r
t

)
ru
a
qD
F.5 pf_t!_U:!I n gf_Igsut ts

a
Figures F.2 and F.3 show rhat
there is little dlfference in
growth curve grouped according
to the rainfalr indices AAR and

a
APBAR' on the o.her hand Flgures
F.1 and F.4 lnd'cate rhat there is
a posslbl1lty that tt'" t"o groupings,
regionality and catchnent
area, may have slgniflcant dlfferences

a distrrbutlons' Beeause the sinoulation

time consunlng, Kolnogorov-snlrnov
in rheir Q1/I{AF
procedure descrlbed above was

a tests rrere uncrertaker only on the

reglonal and catchnent area grouplngs

a conslder firstly the results

and Sunatra) shor.m in Table F.l.
of the regional gro.upings (Java
The Kolnogorov_Snirnov ,,D,, f rom
the comparlson of the Java and Sumagra
Er (0.025) is welt below rhe 952
distrlbutlons of Q1/MAF
signiflcance level of ,,d,,(0.056)
il carculated by simulation, even
standard error in d(0,05). The
arlowing for the rnargin of one
concruslon, 1s therefore, that
il ttttre is no significa.,a irfference,
Kolmogorov-Snirnov rest'in
according to the
the Q/MAF dlstrlbutions in Java
and
sumatra' An explanation for rhi.s
iD 20 years' where the bulk
ls that for reEurn periods up to
of the data occur, the two growth
il are very close*(Figure F.l). curves

I Above 20 years return period

that for Java. This is prlmarily
the SumaEra curve is sceeper
due to the three most
than

t floods 1n all 1001 station years extrene

data occurring i.n Surnacra :

I Ca Echoe

number
nt Ca tchue nt lrlape
Date of
Flood
Q/Mnr

a 4Jt Batang Agan at Titi

ry
I
Ja nua 1 931 4.598
818 Way Besai aE Banjar Llasin
l'la rch r98r 4.476
316 BaEang Anal at lGdang
Enpat

t
December 1979 4 .146
43 IGll Serayu at Gurung
Ma rch L9I6 J. d)d
23 Clkadueun at Cibogo

t These three extrene floods

No vembe

are inportant when constructing

r t97 | 2.845

pooled curve. the

Great welght is placed on
D these few high Q/|,IAF values
when fitting the growth curve above 50 years
return perlod.
D
I
t However, these points form onT!'a very sEa11 part of the total
nunber of Q/uer' in each gr:up and there are not enough of Ehem
co
) register as a signlficant dlfference in the cumuratlve frequency
distri.butions between Java and suma Era . Hence t.he re jec t ion of rhis
) grouping of cat*c'hmenEs by the Kolmogorov-smi r.o v E es r .

) Nowconsider the resurEs of the Kolmogorov-smirnov tesr for the

Q/M-lp grouped according ro'caEchnent area (Table F.l) in conjuncrion
) e/ith the growth curves shown in Figure F.4. unrike ttre regional
grouplng dlscussed above, grouping caEchments into those with large
AREA (greater rhan 600 kn2) and smalr AREA (1ess rhan 600
km2) r"
signlflcant' The Korurogorov-snlrnov D for Ehese Ewo distributions.
0.08, 1s higher Ehan the si.murated 952 significant "d,'of 0.06. even
allowing for one sEandard error in ,'d',.

Fron Figure F.4 it can be seen chat the growttr curves are
dlssimilar rhroughout the range of Qt/UAf, (excepr ar rhe MAF).
This is in contrasE to the- regional grouping, Figure F.l, w\ere
'divergence only occurred at high return periods. Therefore within
Lhe body of the Ewo cumurative frequency distributions,
the
Kolmogorov-Sroirnov test was able to deEect at leasc
one part where
the dlvergenee of the t.wo distributlons \ras greater than could have
been expected by charidii. rn o.her words Ehere vras a rarge
enough
difference in the number of points in one or more particular
Q/MAF
ranges (as defined in Section F.4) [o decrare the d istributions
dissimilar.

tr.Iith a positive indication that the growtlr curves oI the

two
AREA groups were significantly different,
these rwo groups were
further sub-divided according Eo AREA. Thus the previous group
of
46 sna1l caEchments was dlvided inro two groups of 23
carchmenEs
again according to caEchmenL, area. The original group of
rarge
catchnents was sinilarly sub-dlvided. The purpose of
Err i.s was to
see if the trend of small catchmenEs to have steeper growth
curves
than larger 'ones -courd be def-i'ned f urther. Tn orher words,
could
the data set support four raEher than two slgnlficantly differenr
groups of catchments?
 e.p
eD
s botn t*"
results of th '

i*':jT-,ii *,'';ffiT:i1.".; :':;: i: ;:i]; l;i rn

GD t'""pings bv car

el ,n" ,."'"r",1'ii:iJ
l.tJ ssenrla.ry
""tt"tricanL
/
0,.r.
th"
Lrrc
tna tr'" ;;;:ffJjr:t:;,
same musc
"'t"ntu be acc€ofc'l

+
cEf
I
I rt t" perhaps
^^-,
differerls betwqen
of 6oo *in., but
sroups
surpl
tnu'ttt"tng

.tn"'
o" derecred.
that'ntt:
rnain grouorn; "nould
be a srarisrlcal
no dlrfe"unJ""o"J#" divided at the
rhar,",::
sub-di,,i"rpn,
0.n".".
"r:'^:": """

+ F rhar in","n ;: ;"r"rrt;j::"" expla

n*""i""H;1,:i
na'L ion ror
*
rhi s is
smaller sub-divisions

+
growth
;
"JTj,".,:JrTi
*"o"tn ";:";n"
to
suirnov ""tt""ted j:*,::".:j
q T
s r. w, h

on the ho^;^
"
",td
.

0""t" of the
J:: *:: :" *::.j T
:fu ; : j **i,"
"

and the ,.r,u Kr

€ tsdiscussed earlle
E
r
€ ::ii "T::tr^'::""n"J"#:':'"ffJ:'"'""
@ents be accep€.dlvision "'
-*Jl
o ta set i""'-'"
station growt'
c trrves
'
t
e -:'=" d as a sound or"
basis for flood
F'6 Reconmend"o t";tff.r;:.
:nded growtlr
n,"^,.*. factors

The aborro ._^,

analYsis
has
significan-tu "''n]","n""n that there s a statistically
divlded rn.o'l""rence Factor if tt
t* groups catchments
considers no" "";:ttn to are
rhese resul"";otnB ""r;;r";t
reconnendarlo
ns glven 1,
t'"""*i"',"J', ";n" r t""

"n.oi"i
Tj;,;;.
"*
fn order co slnPllfY
curves for th.
use o' of t
reporapplication
; ff.:i,j:; ;":;:;'r;';"'
cabulated set 'n'"
,tl
>r srowrh r..;;;"'

Thus we ha
given the gro
range of usefulve
i;'i;"i:';"' ,:""' Qr/uer' r'r a
:::i;"".T;ff.TT:T".T,:l
discussed rac cors *.nu'l':;
i::T*:j:"":;i: .;";J""" j"
D
"o r,...n ng the growth
curves
t
 €

€
It is inpot
that the design
to apply .n, recommendatlons
€ dible' triith thls
i considered: "."t"nt in uLi nd rhree oo.r"l"t"r"o::n """t
€

Cil )
(I Sepa ra Ee deslgn
6oo ku2
ractors
r ot-"utchments
e less greater than
t "";-;;;"srowth
nedlan catchnent
Ehan 600
km2 bulng rhe
area or a1r sra:::j;oo
€ f (2) a contlnuous
rela
€ perlod srowrh facror-,
"ld ""."r,:::"::J :;::'ercne whole range rerurn
F I

I
area.

(3) As option (2)

bur- over only
-* vvsr

part of the range

of catchment

€ area. of catchment
j

Optlon I is
€ the easiest
whlch of two

; ':"111';: ff: ]".";ffi *"::ij';: ; :J':l T:::::

s
€

"s anr one carchuenr area. coute a dtsc.":"'";i;:T

s ;
more likely.
credibility r^option
i^:: ::
forn of cont
1' therefore,
or averag" does nor r"nuou"
;:::: ::r*
relationship is

s
tn" cri terion of
"-gg carchnenr"
", ,"j"'.t:tt
Options.

a
devel'pmen'c
;iTffijl {i": area bein'
;'i;",.'he and
""'"ntul'lt:::;:tT::;
relaEionshlp-rv 'i"tor ""., "

*r i i::iil,".llj','::: l1d .'"Tffi*H i{li*::,'

t'
5 is some,.",
area. oprion"jT;:i*:;
j:::" rwo op*on" o,'on
decrease in growrlr
since
a
tha
rinear
r

t
Lhere
, t"""u" this ,-^t:"t*"t""
range of car"nr "":t.:::ing. trend ,o ou'lJn'rrntu"tor with catchment

il ;:ffi :::;,lli:':,-",.;:l:i""t:"'Jt"'*T:""'::;::::i':;"
i

the fact thar: the rour

falled to producand sub-;:"j::";Jt;;T';t
e a slsnlricanr
was rejecred. dlfferent growth
E curves, option 2

t Oprlon 3, which
over a lldted
pernits a
range of car"nr","ontinuous change in growth factor
nt areas, was considered
)
most credible and developed as follows lnto a set of design
recommendations :

fl\^/) The median catchment area \{as found ln each of [he

followlng three grouping of catchments:

(a) Snall carchments - 180 knr

(b) A11 catchnents 600 krn2
(c) Large catchments - 1500 km2

(2) Table 7.1 was construcEed by linearly inEerpolating

belween the growth factors associated wirh the three
catchment groups in '(1) above.

(3) When the caEchnenE area is 180 kn2 or less, the 'SMALL'
growt.h curve (or the firsr column of growth factors in
Table 7.1) ls used.

(4) When the caEchment area ls 1500 kt2 ', r rI)() rr', t'lrr- 'T.ARCE'
growth curve (or the last column of growfl) Fac f ors; I rr T;rbl,'
7.I) is used.

(5) If the catchment area is between 180 km2 and 1500 km2
linearly interpolaie between two adjacent columns. For
example the 1000 year reEurn period growth factor for a
425 km2 caEchment is calculated thus:

Qrooo/uer'(3oo trnz) = 4.58

Qrooo/MAr (ooo trn2) = 4.32

Qrooo/uer (425 kn21 = 4.32 +199L19] x (4.58 - {'3:)

(6oo-3oo)
Qrooo MAr ( 4zs unz) = 4.47
 RETERENCES

BaEhurst,J.c. 'Flow resiscance of rarge scale roughness, Journ.

Hydraulics Div. , A.S .C.E r , Hyl, December 197g.

Binnie and Partners et al. 'Report on Hydrology,.

For the Central
Plannlng consurtancy, Dlrectorate of planning
and programmlng.
River Basin plannlng Dlvision, D.G.W.R.D., D.p.U.,
Republic of
Indonesia, August lgBO.

Chow, V.T. tOpen Channel

Hydraulics', McGraw_Hill. 195g.

D.P.M.A. 'penyusunan rnvenEarisasi Danau dan waduk

di pulau Jawa
Tahap I - Jawa Tengah. Seksl Banganan Air, f9g0.

D'P.M.A- 'penyusunan rnventarisasi Danau dan waduk

di pulau Jawa
Tahap rr - Jawa Barat-Jawa Timurr. seksi
Banganan Air, rn press.

Hey, R.D: tFrow fesista.oce in gravel bed rivers ,

. Journ. Hvdraur ic s
Div., A.S.C.E., Hy4, ApriI Lg7g.

rnternaEional Engineering conpany et a1. ,sederhana

rrrigaLion,
Reclamation and Land Devel0pment project
- preliminary
Hydrological study ReporE on Estirnating Low and Flood
flows from
smalr watersheds'. For DGVRD, Republic of Indonesia.
Lg77.

rnsEitute of rlydrology and D.p.M.A.

'Flood Design Manrrar tor Java,
November 1981.

Instltute of Meteorology and Geophysics (rMG). ,Mean

Rrinf-irJ. l ancl
Mean Number of Baindays in Java and Madura,,
Met. Note No.E. part
I, 1931-60 and Volume g3A f96f_70.

rnstlEute of Meteorology and Geophysics.

'Monthly Rainfarl and
' Monthly Nunber of Ralndays'. Vol. 67 for 1951-52 ro Vo1.90
for
1977 .

rrish, J'L. 'ReporE on Ehe estination of

short duration design
ralnfall intensitles Ior Indonesla' part f ,One
, , <lay cl esign
rainfall anountsr, Draft, lggI
Lilliefors, W.L. '0n the Kolmogorov-smirnov test [or nornuility with
mean and variance unknownt ' Anerican Statistical Jor.rrnal. June
L967, pp 399-402.

Ir. Muhadi, Dipl H. 'Iul"enghltung Debit Maximuur Dgn. Cara Weduwen.

Melchior dan Rasionalr. Seminar Cara Perhitungan Design Flood.
D.P.M.A., Bandung, I976.

N.E.R.C. ?F1ood Studies Report' In 5 Volumes, avallable f rom Lhe

InstiLute of Hydrology, Wallingford, U.K., 1975.

RoberEson, A.I.G.S. tLogarlthnic plotting of stage discharge

. observations'. I^/ater Resources Board, Readlng, U.K., Repri,nted
OcEober 197O.

stevens, M.J. and Lynn, P.P. 'Regional growth curves'. InsIltLr te of

Hydrology Report No.52, October L978.

SuEc.liffe, J.V. 'Use of the Flood Studies Report Overseast. Flood

Studies Report-5 years on. Thonns Telford Ltd., London, U.K.,
r981.

World Meteorological OrganlsaEion. 'l'lanua1 on SLream Gaugingt .

Operational Hydrology Report No.13, Two Volumes, 1980.