Hydrological Sciences Journal

ISSN: 0303-6936 (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/thsj19

THE TWO-AXIS METHOD: A NEW METHOD TO

CALCULATE AVERAGE PRECIPITATION OVER A
BASIN / La deux-axes méthode: une nouvelle
méthode pour la calculation de la précipitation
moyenne d'un bassin versant

NEDAVIA BETHLAHMY

To cite this article: NEDAVIA BETHLAHMY (1976) THE TWO-AXIS METHOD: A NEW METHOD
TO CALCULATE AVERAGE PRECIPITATION OVER A BASIN / La deux-axes méthode: une
nouvelle méthode pour la calculation de la précipitation moyenne d'un bassin versant, Hydrological
Sciences Journal, 21:3, 379-385, DOI: 10.1080/02626667609491646

To link to this article: https://doi.org/10.1080/02626667609491646

Published online: 25 Dec 2009.

Submit your article to this journal

Article views: 4235

View related articles

Citing articles: 12 View citing articles

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=thsj20
 Hydrological Sciences-Bulletin-des Sciences Hydrologiques, XXI, 3/19 76

THE TWO-AXIS METHOD: A NEW METHOD TO CALCULATE AVERAGE

PRECIPITATION OVER A BASIN

NEDAVIA BETHLAHMY
Intermountain Forest and Range Experiment Station, Forest Service,
USDA, Ogden, Utah 84401, USA. Stationed at Forestry Sciences Laboratory,
Moscow, Idaho, maintained in cooperation with the University of Idaho
Revised MS. received 17 March 1976

Abstract. The two-axis method, used to calculate average precipitation over a basin, is based on three
assumptions inherent in the Thiessen and isohyetal methods. The method follows four steps: (1) The two
axes of a basin are drawn following a standard procedure; (2) for each raingauge location a station angle
is determined by either measuring or calculating the angle between lines connecting the station location
to the farthest ends of the two axes; (3) each station is assigned a weight (the ratio of its angle to the sum
of all station angles); (4) average area precipitation is calculated as the sum of the products of correspond-
ing station weights and measured precipitation. The method is fast, efficient, and is readily computerized.
La deux-axes méthode: une nouvelle méthode pour la calculation de la précipitation moyenne
d'un bassin versant
Résumé. La deux-axes méthode, utilisée pour la calculation de la précipitation moyenne dans un bassin
versant, est basée sur trois suppositions inhérentes dans la méthode isohyetale et dans celle de Thiessen.
La méthode comprend quatre étapes: (1) On dessine les deux axes d'un bassin versant en se servant des
règles normales de procedure; (2) pour chaque station pluviométrique on détermine un angle de poste ou
en mesurant ou en calculant l'angle entre les lignes qui joignent la station aux extrémités des axes; (3) on
donne un facteur d'importance (le rapport d'un angle à la somme globale des angles des stations) à chaque
station; (4) la précipitation moyennedans le bassin versant est la somme globale des produits des facteurs
d'importance des stations et de la précipitation mesurée correspondante. La méthode est rapide, efficace,
et se fait facilement à l'ordinateur.

INTRODUCTION
The depth of precipitation that a storm releases over an area may vary from a maximum
value at one or more points to zero at the storm's boundary. This variation poses problems
in determining the average precipitation that falls on a basin affected by the storm. The
average depth is commonly calculated in one of three ways: Simple average, Thiessen, or
isohyetal methods (DeWiest, 1965).
The simple average is the arithmetic mean of precipitation measured at a number of
stations. This procedure ascribes equal weights to all measurements, and is used when rain-
gauges are uniformly distributed and the topography is flat.
The Thiessen method weights the precipitation measured at each station by the area
that the station represents. It is appropriate when orographic effects are negligible and rain-
gauges are not spaced uniformly.
The isohyetal method uses the observed precipitation data as the basis for drawing con-
tours of equal precipitation (isohyets), and then weights the average precipitation of adjacent
isohyets by the area between the isohyets. The method is suitable for large areas, especially
those in which orographic effects may be present.
Although the somewhat complex and cumbersome Thiessen and isohyetal methods are
379
basically objective, the final results depend on the procedures employed by the analyst, and
on his skill, draftsmanship, and knowledge of the terrain and storm characteristics. Computer
programs for these methods are available: Diskin (1970) for the Thiessen method; Kwan et
al. (1968) for the isohyetal method.
Computer programs are also used in methods proposed by other investigators (Akin,
1971; Chidley and Keys, 1971; Edwards, 1972; Mandeville and Rodda, 1972; Salter, 1972;
Shaw and Lynn, 1972). The advantage of a particular method depends on the use of season-
al or annual precipitation over large areas in contrast to single storm events over limited areas.
Vahl (1972) pointed out that monthly and yearly areal precipitation are relevant to
water balance studies, whereas individual storm events are relevant to studies in flood hy-
drology.
An interest in flood hydrology prompted the author to propose a new method to cal-
culate the average depth of precipitation over a basin. The method is simple, yields unique,
reproducible answers, has the weighting advantages, of the Thiessen and isohyetal methods,
and is readily computerized.
The proposed two-axis method weights each raingauge station in an unbiased way. The
value of a station angle (expressed in degrees or radians) depends on the station's location
with respect to the basin's axes; station weight is the ratio of its angle to the sum of all sta-
tion angles; and the average depth of precipitation over a basin is the sum of the products of
corresponding station weights and measured precipitation.
Basic assumptions
The two-axis method is based on three assumptions. The first is that all raingauges are not
equally significant with respect to the average rainfall over an area; a weighting procedure is
required. This assumption is implicit in all methods except the simple average method.
A second assumption of the two-axis method is that a raingauge station located near the
centre of a basin should be weighted more than one located farther out. The assumption is
reasonable because a storm event at the centre of a basin has a greater probability of cover-
ing a larger area within the basin than an event near or beyond the border of the basin. This
assumption is also implicit in the Thiessen method. As an example, consider a circular basin
monitored by two raingauges, one at the centre, another on the perimeter. By the Thiessen
method, the raingauge at the centre represents 80.4 per cent of the basin, while the gauge on
the perimeter represents only 19.6 per cent. If the gauge on the perimeter were relocated
farther out, its relative weight would diminish until at a distance of one diameter from the
centre it would represent 0 per cent of the basin area, and the gauge at the centre would
represent 100 per cent of the basin. The principle of reduced significance with distance from
the centre also applies to the two-axis method, but the rate of reduction is far different.
In the example (circular basin monitored by two raingauges), station angles are 90° for
the central gauge and 45° for the perimeter gauge. Central station weight is 0.67 [90°/
(90° + 45°)] and perimeter station weight is 0.33 [45/(90 + 45)]. If the perimeter gauge is
relocated at a distance of one diameter from the centre, the central station angle remains
90°, but the outer station angles will vary from a minimum of 26.6° (if located on the ex-
tension of either axis) to a maximum of 29.3° (if located midway between the axes); and
the corresponding weights of the central and outer stations would be (0.75, 0.25) and
(0.77, 0.23).
A third assumption of the two -axis method is that the significance of a raingauge (its
weight) depends on its location with respect to the basin's two axes. This assumption is
implicit in the Thiessen method.
Consider a noncircular basin centred among three fixed raingauges. If the basin is rotat-
ed about its centre, then the relative significance of each gauge will change, depending upon
the position of the basin. The assumption is also implicit in the isohyetal method as is evi-
dent in the procedures followed by the knowledgeable hydrologist; in drawing isohyets, he

380
considers not only amounts of precipitation but also ground elevation contours, and these
contours are related to a basin's axes.

LOCATING AXES AND DETERMINING STATION ANGLES

Although the centre of a basin could be selected by eye, no two investigators would pin-
point the same centre, and a standard procedure is needed. In the two-axis method, the
centre of a basin is defined as the point of intersection of its major and minor axes. In most
cases, the axes may be located (Fig. 1) as follows: Connect the basin's outlet (0) to the
farthest point on the boundary (7); to this line draw a perpendicular bisector (AB)—this
line is the minor axis. To the minor axis draw a perpendicular bisector (CD)—this is the
major axis. This procedure will not fit all cases. If the basin is crescent-shaped, so that initial
line OT lies wholly or partly outside the basin's boundaries, then one should realize that the
minor axis is bounded by the basin, and that the major axis is the perpendicular bisector of
the minor axis. If the basin is broadly heart-shaped, then initial line OT should be drawn
across that portion of the basin that will maximize its length.

Fig. 1 - A standardized method to locate a basin's major and minor axes. Line OT connects the outlet
to the farthest point on the basin's perimeter. Line AB is the perpendicular bisector of OT and is the
minor axis. Line CD is the perpendicular bisector of AB and is the major axis.

A station angle is determined by referencing the station's location to the two axes; it is
the acute angle between two lines drawn from the station to the farthest terminals of the
two axes (Fig. 2). Thus, Station Pt is located above the minor axis and to the right of the
major axis; its angle is angle APXC. Station P2 is below the minor axis, left of the major axis,
and its angle is angle BP2D. The angle of a station located at the exact centre is always 90°
(the maximum possible angle), and all other possible locations have angles less than 90°.
The magnitude of a station angle is readily determined either mechanically with a pro-
tractor, or by computer if all points are referenced by Cartesian or map coordinates. A
simple computer program is possible because only elementary algebra and trigonometry are

381
involved. The length (L) of a line connecting any two points whose coordinates are (xh }>i)
and 0*2, .)> 2) ' s
L=[{xl-x2y+(yi-y2y}l/l
In any triangle ABC, angle A is
/I = arc cos {(b2 + c2 - a2)/2bc]
where a, b, c axe lengths of sides opposite angles A, B, C. Thus, in Fig. 2, angle AP\C is
arc cos [(IP, 2 +P^C2 - Â~C2)/2Â~FJ\C\

Fig. 2 - A station angle is the angle between lines drawn from the station to the farthest terminals of the
major and minor axes. The angles of stations P a and P 2 are angles APiC and BP2D. The weight of any
station is the ratio of its angle to the sum of all measured station angles.

STATION WEIGHT
The station weight (Wj) of raingauge station / is the ratio of the station's angle (Ai) to the
sum of all station angles:
Wi=Ail2Ai„ltn
Thus, if the respective angles of raingauge stations K, F, G, are 60°, 75°, and 80°, then the
sum of the angles is 215°, and the appropriate weights are the ratio of each measured angle
to this sum:
I % = 60/215 =0.279
WF =75/215=0.349
WG =80/215 =0.372
zWj =7xxx)

382
AVERAGE PRECIPITATION
The average precipitation (P a ) over a basin is the sum of the weighted precipitation of all
stations. Thus, if/'/ units of precipitation were measured at stations / with weights Wj, then
the average precipitation over the basin is:
P a = 2W(/>/
Figure 3 illustrates a hypothetical basin monitored by five raingauges located at stations
A to E, Precipitation totals for five storms are listed sequentially at each station. Each of the
five hypothetical storms is centred at a different station.
B+
(0,1.2,4,3)

Fig. 3 - Hypothetical basin monitored by five raingauges (A to E). Precipitation totals for each of five
storms are listed sequentially for each gauge. Average rain over the basin as calculated by the simple aver-
age method is 2.0 cm for each of the five storms. Analyses by the Thiessen, isohyetal, and two-axis meth-
ods yield different answers for each of the storms, but the order of magnitude of the five storms is main-
tained.

Table 1 shows average storm precipitation over the hypothetical basin, as calculated by
the three commonly used methods and by the two-axis method.
The tabulation shows that the simple average method yields identical values for all five
storms. In contrast, the methods that use a weighting procedure yield different answers for
each storm; but the order of magnitudes of the five storms is generally uniform and varies
little with the method used.
Gilman (1966) illustrates the isohyetal method with ah example that yields a 5.8-cm
average depth of precipitation. An identical answer is obtained when the same data are ana-
lysed by the two-axis method.

383
 TABLE 1
Average precipitation for five hypothetical storms calculated by
four methods
Method

Storm Simple
average Thiessen Isohyetal Two-axis
[cm] [cm] [cm] [cm]
1 2.00 2.86 2.51 2.38
2 2.00 2.01 1.84 2.14
3 2.00 1.95 1.80 1.81
4 2.00 1.76 1.76 1.87
5 2.00 1.43 1.46 1.80

DISCUSSION
The proposed two-axis method for determining average precipitation over a basin is based
on three reasonable assumptions: (1) all raingauges are not equally significant with respect
to the average rainfall over an area; (2) a raingauge near the centre of the area is more signif-
icant than one far removed from the centre; (3) the significance of a raingauge is affected by
the shape of a basin. Each of these assumptions is implicit in one or the other of the common-
ly used methods, excepting the simple average.
The true average precipitation over a basin can be obtained if the area is covered by a
multitude of raingauges. This method is clearly impractical except for very small plots. For
large areas the true average can never be known. At best, we can determine for large areas
an apparent average whose value depends on the number of raingauges. Investigators have
shown that increasing the number of gauges usually results in an increase in the apparent
average precipitation (Gilman, 1966); the reasons are threefold: rainfall over an area does
not decrease uniformly from a single storm centre; many spots of high rainfall concentration
exist; and the opportunity for sampling these spots increases with sampling density.
Although the isohyetal method is considered one of the best methods used to determine
average precipitation over an area, it cannot yield the true average. An isohyetal map does
not portray the true distribution of rainfall because the map is based on a limited number of
sampling points; the addition of a single raingauge results in a different map and a different
average. Also, even with the same data available, different but reasonable isohyetal maps can
be drawn, each yielding a different average. This fault (multiple possible answers from the
same set of data) also characterizes the Thiessen method. In contrast, the two-axis method
yields but one answer with a given set of data, the answer is obtained in a fraction of the
time required with the other methods, and the entire analysis is readily computerized.
Although the true average rainfall over a large area cannot be known, we can at best
sample properly and then analyse the data with efficient methods that are based on logical
premises. The two-axis method is recommended as a procedure that meets these criteria.

REFERENCES
Akin, J.E. (1971) Calculation of the mean areal depth of precipitation. /. Hydrol. 12, 363-376.
Chidley, T.R.E. and Keys, K.M. (1971) A rapid method of computing areal rainfall. /. Hydrol. 12, 15-24.
DeWiest, R.J.N. (1965) Geohydrology, chapter 2: John Wiley, New York.
Diskin, M.H. (1970) On the computer evaluation of Thiessen weights. /. Hydrol. 11, 69-78.

384
Edwards, K.A. (1972) Estimating areal rainfall by fitting surfaces to irregularly spaced data. Distribution
of Precipitation in Mountainous Areas (Proceedings of the Geilo Symposium, 1972), vol. II,
pp. 565-587: WMO Publ. no. 326 or IAHS Publ. no. 106.
Oilman, C.S. (1966) Rainfall. In Handbook of Applied Hydrology, section 9, (edited by Ven Te Chow):
McGraw Hill, New York.
Kwan, J.Y., Riley, J.P. and Amisal, R.A. (1968) A digital computer program to plot isohyetal maps and
calculate volumes of precipitation. The Use of Analog and Digital Computers in Hydrology (Proceed-
ings of the Tucson Symposium, 1968), vol. I, pp. 240-248: IAHS Publ. no. 80.
Mandeville, A.N. and Rodda, J.C. (1972) A contribution to the objective assessment of areal rainfall
amounts. Results of Research on Representative and Experimental Basins (Proceedings of the
Wellington Symposium, 1970), vol. II, pp. 120-128: IAHS Publ. no. 97.
Salter, Pauline M. (1972) Areal rainfall analysis by computer. Distribution of Precipitation in Mountain-
ous Areas (Proceedings of the Geilo Symposium, 1972), vol. II, pp. 497-509: WMO Publ. no. 326
or IAHS Publ. no. 106.
Shaw, Elizabeth M. and Lynn, P.P. (1972) Areal rainfall evaluation using two surface fitting techniques.
Hydrol. Sci. Bull. 17,419-433.
Vahl, Heiko (1972) Computerized calculation of areal precipitation and its accuracy. Distribution of
Precipitation in Mountainous Areas (Proceedings of the Geilo Symposium, 1972), vol. II, pp. 510-524 :
WMO Publ. no. 326 or IAHS Publ. no. 106.

385