Journal of Hydrology 237 (2000) 40–57

www.elsevier.com/locate/jhydrol

Gross rainfall and its partitioning into throughfall, stemflow and

evaporation of intercepted water in four forest ecosystems in
western Amazonia
C. Tobón Marin a,b,*, W. Bouten a, J. Sevink a
a
Fysisch Geografisch Bodemkundig Laboratorium, University of Amsterdam, Nieuwe Achtergracht 166, 1018 WV Amsterdam, The Netherlands
b
The Tropenbos Foundation, Colombia
Received 29 July 1999; revised 4 May 2000; accepted 4 July 2000

Abstract
The partitioning of gross rainfall into throughfall, stemflow and evaporation of intercepted rainfall was studied in four forest
ecosystems in the Middle Caquetá, Colombian Amazonia. Data on climate was collected automatically on an hourly basis
during a five-year period. Weekly measurements of rainfall, throughfall and stemflow were carried out during a period of two
years, while daily measurements, on an event basis, were carried out during two subsequent years. Throughfall, stemflow and
evaporation in each forest were checked for correlations with gross rainfall characteristics, canopy gap fraction, tree crown area
and bark texture. Canopy gap fraction differed between forests, ranging from 9% on the flood plain to 17% on the Tertiary
sedimentary plain. Rainfall was rather evenly distributed over the year, with one dry period from December to February. 92% of
the rain fell in single showers of less than 30 mm and most of the storms (56%) fell in less than one hour, during the afternoon or
early night. Throughfall ranged from 82 to 87% of gross rainfall in the forests studied and varied with gross rainfall in all
forests. It depended on the amounts and characteristics of rainfall, but differences in throughfall among forests, when comparing
similar rainfall events, clearly indicated that throughfall also depends on forest structure. Stemflow contributed little to net
precipitation (on average 1.1% of gross rainfall in all forests) and showed a power relation with gross rainfall. Correlations
between stemflow per tree, projected crown area and bark texture were very poor as indicated by the low coefficients of
determination. Evaporation during rainfall events exhibited a linear relation with rainfall duration and the ratio of evaporation
over gross rainfall increased with forest cover (1-gap fraction) in the forests studied. The structure of the forests seemed to vary
considerably and given its influence on rainfall partitioning it may explain both differences and similarities between results
from this study and those from most other studies within Amazonia. 䉷 2000 Elsevier Science B.V. All rights reserved.
Keywords: Amazonia; Tropical rain forest; Throughfall; Stemflow; Evaporation; Forest structure

1. Introduction climate (Salati and Vose, 1984). Using a modelling

approach, Eltahir and Bras (1993) concluded that the
The Amazonian rain forest seems to play an impor- atmosphere in the Amazon basin is an open system
tant role in the regulation of regional and global and that the net input of atmospheric moisture into the
basin is about 32%, about 68% of the gross input
leaving the basin. They also found that the recycling
* Corresponding author. Tel.: ⫹31-205257442; fax: ⫹31-
205257431. ratio of the Amazon basin is about 25–35%. This ratio
E-mail address: [email protected] (C. Tobón Marin). differs from those found by others, which were based
0022-1694/00/$ - see front matter 䉷 2000 Elsevier Science B.V. All rights reserved.
PII: S0022-1694(00 )00 301-2
 C. Tobón Marin et al. / Journal of Hydrology 237 (2000) 40–57 41

on the erroneous hypothesis that the atmosphere over 1982), which has been related to the large aerody-
the Amazon basin is a closed system (Molion, 1975; namic conductance of forest (Stewart, 1977). Addi-
Lettau et al., 1979; Salati et al., 1979). Eltahir and tionally, there has been an increasing awareness that
Bras (1993) furthermore concluded that deforestation evaporation of intercepted rainfall has to be investi-
of Amazonia would increase the surface temperature gated separate from transpiration, especially in very
and decrease the heating of the upper troposphere, humid areas (Hutjes et al., 1990; Shuttleworth and
which would result in a reduction of precipitation. Calder, 1979). Furthermore, most studies on forest
The foregoing illustrates that recent concern about interception showed that this interception is closely
tropical rain forest deforestation focuses on its impact related to gross rainfall amounts and characteristics.
on climate at regional and global scale (e.g. The However, the influence of forest structure on
Anglo Brazilian Amazonia Climate observation interception is poorly known.
study ABRACOS, Gash et al., 1996; the Large Most rainfall interception and water balance studies
Scale Biosphere–Atmosphere experiment in Amazo- in Amazonia were executed in Brazil (Ubarana, 1996;
nia, Nobre et al., 1996; Eltahir and Bras, 1993). Leopoldo et al., 1995; Lesak, 1993; Lloyd and
Current research concentrates on global circulation Marques, 1988; Shuttleworth, 1988) and only a few
models and on those parameters and fluxes that play in other parts of Amazonia (Hölscher et al., 1997;
a role in the global climate (e.g. Gash et al., 1996). Jetten, 1996; Wright et al., 1992; Jordan, 1978).
Nevertheless, it is often argued that additional work is Thus, in Colombian Amazonia, which represents
needed to explain and improve model predictions for one of the most humid areas within the basin, very
impacts of deforestation (Bruijnzeel, 1990; Nobre et little attention has been paid to the hydrology of forest
al., 1991). Evidently, local hydrological studies on ecosystems and to the effects of forest structure on
undisturbed mature rain forests would provide base water dynamics.
level information on initial conditions that might This paper concerns a study designed to address
allow for an evaluation of the presumed influence of this lack of knowledge by measuring rainfall and its
deforestation on regional and global climate. This partitioning after entering the canopy in four undis-
paper concerns such local study. turbed rain forests in the Middle Caquetá, Colombian
In the Middle Caquetá (Colombian Amazonia), the Amazonia. It focuses on the analysis of long-term
structure and species composition of the forest vary hydrological measurements of rainfall, throughfall,
considerably between the different landscape units stemflow, the resultant evaporation and the related
(Duivenvoorden and Lips, 1995; Londoño, 1993). structure of these forests.
This causes equally large variations in temporal and
spatial patterns of water fluxes in these units, leading
to local differences in water and nutrient stocks in the 2. The study area
various forest compartments (Vitousek and Denslow,
1986; Tanner, 1985). Thus, the partitioning of rainfall The study area is in Peña Roja (Nonuya Indian
into throughfall and stemflow leads not only to a more community) near Araracuara, Middle Caquetá,
diffuse input of water into the forest floor, but also to Colombia, (0⬚ 37 0 and 1⬚ 24 0 S, 72⬚ 23 0 and 70⬚ 43 0
local concentration around the base of tree stems, W; Fig. 1). Climate is classified as equatorial super-
which is known to induce spatial variability in soil humid Afi (Köppen, 1936). The research sites are
properties and soil moisture conditions (Waidi et al., permanent undisturbed forest plots, used by the
1992). In tropical forests, the abundance of epiphytes, Tropenbos Foundation for its research. They lie
climbers and aerial roots renders this partitioning approximately 200–250 metres above sea level and
much more complicated than in temperate forests form a sequence from the lower terrace of the River
(Longman and Jenı́k, 1990). Caquetá to the Tertiary sedimentary plain. Based on
It is well understood that in forested areas generally data from the manual meteorological station at Arara-
total evaporation is larger than in areas with shorter cuara (IDEAM), average annual rainfall in the area is
vegetation (e.g. grass) mainly due to the larger inter- about 3100, April being the wettest month and
ception by the forest canopy (Bosch and Hewlett, January the driest.
42 C. Tobón Marin et al. / Journal of Hydrology 237 (2000) 40–57

Fig. 1. Location of the research sites in the various land units in the Middle Caquetá, Colombian Amazonia.

Colombian Amazonia comprises 403,000 km 2 and located in the four main land units in the area: the
the major part of this area is covered by mature rain Tertiary sedimentary plain, the upland terraces of
forests, classified as ombrophilous tropical forest the River Caquetá (high and low terraces) and the
(Duivenvoorden, 1995). The research plots are flood plain. The vegetation is very rich in species
 C. Tobón Marin et al. / Journal of Hydrology 237 (2000) 40–57 43

(Duivenvoorden and Lips, 1995; Londoño, 1993) and branches; (2) manual measurements with two rain-
is typical for undisturbed mature forests in the western gauges per subplot suspended from cords attached to
part of the Amazon basin. The canopy reaches about two emergent trees in small gaps within the forest.
25–30 m above the forest floor with some emergent Throughfall was measured in the same rainfall
trees reaching up to 45 m in the rarely inundated flood subplots using 20 collectors per subplot, randomly
plain. There are three to four canopy layers, but large located in an area of 50 by 20 m (1000 m 2). Evapora-
tree crowns in the upper canopy form the bulk of the tion from the collectors was avoided by using an
vegetation. Lower canopies contribute far less to the internal plastic tube running from the funnels to the
forest cover. Small palms reaching a height of 2–4 m bottom of the collectors. To allow for direct correla-
constitute the lowest layer. The land units differ in the tion, all funnels for gross rainfall above the forest
total standing biomass, species diversity and tree canopy and for throughfall had an orifice of
density (Duivenvoorden and Lips, 1995; Londoño, 298.6 cm 2. Throughfall and forest rainfall collectors
1993). Other differences between plots pertain to the were calibrated against standard raingauges in the
structure of the forest canopy (canopy cover) and the open.
contribution of epiphytes, climbers and aerial roots. A Because of the large variability in throughfall due
more detailed description and vegetation classifica- to the forest structure (Jetten, 1996; Ford and Deans,
tion of the research sites is given by Duivenvoorden 1978) many readings are needed to study forest inter-
and Lips (1995); and Londoño (1993). ception. However, when using average values,
moving the collectors has a positive effect by reducing
the standard error of estimations (Lloyd and Marques,
3. Materials and methods 1988). Therefore, each month (after five measure-
ments) collectors were randomly relocated within
The areas for the present study were selected as all subplots during the period of ‘single event
being representative for the natural vegetation in measurements’, i.e. between 1996–1997.
the main land units from this part of the Amazon Stemflow was measured for 15 randomly selected
basin. Three subplots were selected in the Tertiary trees in each subplot. Collars, constructed from 8 mm
sedimentary plain (SP) and two subplots in the thick black polyethylene plastic, were sealed to the
high terrace (HT), the low terrace (LT) and the stems in an upward spiral pattern and the water
flood plain (FP), respectively, to measure gross diverted into bottle gauges on the forest floor. The
rainfall above the forest canopy, throughfall and opening of each collar extended only about 2–3 cm
stemflow (Fig. 1). In 1992, approximately 3 km from the trunk to avoid drips from the branches or
from the plots, in an open area of about 20 hectare leaves being collected by the collars. The amount of
(within an Indian community village), an water, which drops in a diffuse pattern around rough-
automatic weather station (AWS) was installed barked trees rather than adhering to and flowing down
to measure gross rainfall, temperature, air the trunks, was considered throughfall. For practical
humidity, incoming radiation, wind speed, wind reasons, only trees with diameter larger than 10 cm
direction and Class A pan evaporation. were selected for stemflow measurements. Where
Parameters were measured and recorded every palm trees were present in the subplots, depending
30 s using a datalogger (CR10 Campbell Scientific on their frequency of occurrence one or two of
Instruments), which additionally recorded means palms per subplot were randomly selected. In the SP
or totals every 20 min. and HT plots, two palms were selected, in the LT four
Rainfall in the open area was measured by a tipping palms in total and in the FP three palms. Stemflow
bucket raingauge with a resolution of 0.2 mm provid- measurements were rotated once by installing collars
ing information on the number and duration of around stems of new trees within the same subplots.
showers and on total rainfall. Gross rainfall in each Horizontally or downwards inclined branches of
plot was measured in two ways: (1) automatic trees may not direct intercepted rainfall to the centre
measurements with one tipping bucket installed in of the tree to be drained as stemflow. Therefore, the
the top of an emergent tree crown after clearing all flat area of the tree crown was mapped by means of
44 C. Tobón Marin et al. / Journal of Hydrology 237 (2000) 40–57

vertical projections from the edge of outstanding storage capacity of the forest canopy (C), which
upward branches to the forest floor. At least six have been presented thusfar (Klaassen et al., 1998),
projection lines were drawn for each tree. The mainly differ in the way of accounting for the drainage
crown area was found by integrating measured areas before the forest is completely saturated and the
of each triangle. Stemflow is expressed as millilitre of gradual saturation of forest layers with continuously
water coming in over the horizontal area of the tree proceeding evaporation. In the present study, C is the
crown and flowing down along the tree trunk. Thus, intercept of the regression of the estimated evapora-
each tree was considered as a single catchment area. tion versus gross rainfall. Therefore, only single, high-
Estimates for stemflow at subplot level were obtained intensity rainfall events of short duration (less than
by multiplying the average stemflow per tree of a one hour) were used. Late afternoon and night rain-
specific diameter by the total number of trees with falls were preferred, representing rainfalls under
approximately the same diameter within the subplot. conditions of a low moisture deficit. In total, 30–40
Data on sampled trees with a diameter of about 10 cm events were selected for each ecosystem. Free
were used to extrapolate the information to trees with throughfall is considered as an important parameter,
smaller diameter. Average stemflow per subplot was among others, in studies on nutrient cycling, since it
related to gross rainfall measured in the same subplot may represent the fraction of throughfall which is not
to determine the percentage of rainfall coming to the involved in washing out dry deposition, exudates and
forest floor as stemflow. released nutrients from leaves, branches and trunks.
Stemflow amounts from each rainfall event (ml) Free throughfall in the forests may be taken as the gap
and from individual trees were correlated with the fraction (ps). Therefore, it was estimated for each
measured projected area of the central crown of the forest from the set of digitised and scanned black
tree concerned and with some tree characteristics such and white photographs taken in each subplot and
as trunk surface area and bark texture. Consequently, from the regression coefficient of throughfall versus
the periphery of the trunks or surface area was deter- gross rainfall (pt), using data from small storms,
mined by measuring the tree trunk circumference at which were insufficient to saturate the forest canopy
breast height and the height of the tree trunk to the first (Gash and Morton, 1978). About 14 small storms were
set of divergent branches. Trunk area was determined selected for each forest to determine the value of pt.
by considering the tree trunk as being rectangular and Manual measurements of rainfall, throughfall and
applying a correction factor to account for the conical stemflow were carried out on weekly basis from
form of tree. Bark texture was classified according to December 1993 to February 1996, without moving
the roughness of the bark in a range from smooth to collectors. During these years, some measurements
very fibrous. Per plot, 40 hemispherical photographs were carried out on event basis during periods with
were taken to estimate the gap fraction of the forest limited or no rainfall (mostly in the dry season). Daily
canopy in the subplots where throughfall collectors measurements were performed from February 1996 to
were installed. The photographic camera was installed August 1997. Readings were made early in the morn-
horizontally at the same height as the throughfall ing (around 08:00 local time) or after the storm. Perso-
collectors. The black and white photographs were nal observation in the plots showed that the upper
digitised with a scanner and analysed with canopy dried out within 6 h after rainfall, if this
the Hemiphot program (Ter Stegee, 1994), which event occurred during the daytime or part of it.
calculates the gap fraction of a forest and roughly Therefore, the criteria to separate the single events
estimates the LAI from the forest cover. Although was that only those events where no rainfall occurred
the photographs were taken under covered sky with in the previous 6 h were considered. In those cases
diffuse sun light, they were corrected for the reflected where two or more events occurred during the night
light from leaves, branches and trunks. It is assumed time, since evaporation during the night is expected to
that the mean fraction of white pixels in the be low, they were considered as a single event. Most
photographs gives the best estimate of the gap analysed single events in this study are events,
fraction. preceded by a dry period of more than 10 h.
Most methods for the determination of the water Evaporation (E), which is taken here as the total
 C. Tobón Marin et al. / Journal of Hydrology 237 (2000) 40–57 45
2 2
amount of water intercepted and evaporated from the area was 62.2 m and the smallest 2.4 m , both in
forest canopy, is calculated on event basis from the the high-terrace plot. For the trunk surface area, the
difference between gross rainfall (Pg) and net rainfall correction factor of 0.5 was applied to trees with
(throughfall Th plus stemflow Sf) of the single rainfall diameter larger than 0.1 m. The apparent inconsis-
events. Cumulative evaporation during the rainfall tency in the flood plain data, of a large tree crown
events (Ew) is calculated from the difference between area in combination with the lowest tree trunk surface
gross rainfall and net precipitation plus the water area, is mainly due to the abundance of small trees
stored in the forest canopy after rainfall ceased (S), with a diameter of about 0.1 m. The data on the esti-
deduced from the estimated evaporation of single mated gap fraction of the canopy and the free through-
events minus the storage capacity (C) fall coefficient show that considerable differences
exist between forests, the largest values being
Ew ˆ Pg ⫺ …Th ⫹ Sf ⫹ S† …1†
observed in the sedimentary plain. Moreover, the
To avoid negative values in Eq. 1, although leaf area index appeared to decrease from the sedi-
they may reflect those events where the forest mentary plain to the flood plain of the River Caquetá.
canopy does not reach saturation, the value for S The estimated LAI values are within the range of
was inferred from the single rainfall events where values found for similar forests in Brazilian Amazonia
E was larger than C. (Roberts et al., 1996; Klinge et al., 1975). The
Statistical analyses were made for the entire decrease in gap fraction and increase in LAI towards
collected data. However, for the assessment of the the flood plain is in line with the results from Duiven-
effect of rainfall sizes on net rainfall and resulting voorden and Lips (1995), who observed that litter fall
evaporation, measurements of single rainfall events in the flood plain of the River Caquetá is highest when
are required. Therefore, throughfall percentages rela- compared to the other ecosystems in the area.
tive to gross rainfall and evaporation were calculated
from these single events. Moreover, to determine the 4.2. Rainfall characteristics
effect of forest structure on throughfall percentage and
evaporation, we used only data from single rainfall Gross rainfall above the forest canopy did not vary
events collected during the period of weekly considerably within plots. On average, the differences
measurements when collectors were not relocated. in the amounts of gross rainfall between subplots in
the SP were 5.5% (^5.6) with n ˆ 3; whereas in the
HT these were 3.7% (^3.5), in the LT gauges 3.7%
4. Results (^3.0) and in the FP 3.1% (^2.7), all with n ˆ 2:
Rainfall distribution differs between plots when
4.1. Forest structure examining separate storms, although annual totals
are rather similar. During the measurement period
Average values of measured variables of forest (1992–1997), the mean annual rainfall at the AWS
structure in each forest ecosystem studied are was about 3400 mm y ⫺1 and the average period with
presented in Table 1. The largest mapped crown rainfall was 616 h y ⫺1. In total, 1584 rainfall events

Table 1
Forest structure characteristics of four forest ecosystems in the Middle Caquetá, Colombian Amazonia

Parameter Sedimentary plain High terrace Low terrace Flood plain

Crown area (m 2) 9.5 (^7.3) 17.8 (^14.0) 9.8 (^7.2) 12.1 (^8.3)
Stem trunk area (m 2) 7.2 (^4.6) 8.2 (^5.2) 8.4 (^5.4) 5.4 (^4.6)
Gap fraction (%) 16.8 (^2.4) 15.4 (^3.4) 11.7 (^1.5) 8.2 (^1.5)
Free throughfall coefficient, pt 0.59 0.52 0.49 0.27
LAI 4.4 (^0.7) 4.9 (^0.8) 5.6 (^0.6) 6.6 (^0.4)
Forest storage capacity (mm) 1.16 1.28 1.32 1.55
46 C. Tobón Marin et al. / Journal of Hydrology 237 (2000) 40–57

Fig. 2. Rainfall characteristics at the research site in the Middle Caquetá, Colombian Amazonia.(August 1992 to August 1997)
C. Tobón Marin et al. / Journal of Hydrology 237 (2000) 40–57 47

Fig. 2. (continued)
48 C. Tobón Marin et al. / Journal of Hydrology 237 (2000) 40–57

Table 2
Throughfall percentages of daily gross rainfall for 5 storm classes, in four forests in the Middle Caquetá, Colombian Amazonia. (SP)
sedimentary plain, (HT) high terrace, (LT) low terrace, (FP) flood plain, (std) standard deviation of the means and (n) number of events

Rainfall ranges Throughfall %

SP HT LT FP

(mm) % std n % std n % std n % std n

⬍5 58.7 11.4 41 56.2 12.1 34 52.3 9.6 32 47.4 13.2 27

5–20 81.4 6.3 78 80.5 5.6 68 79.8 6.8 71 74.5 6.9 57
20–40 88.9 2.8 39 87.9 2.6 36 87.7 3.4 41 83.0 2.5 32
40–80 90.6 2.1 19 90.0 1.9 19 88.8 2.5 17 84.6 3.1 19
⬎ 80 92.8 – 1 92.2 0.8 3 92.4 1.0 2 88.5 1.2 5
Total 87.2 2.4 178 86.7 2.4 160 85.8 1.2 163 81.9 1.0 140

were recorded at the AWS in Peña Roja, with storms tage of mean throughfall) over gross rainfall varied
ranging from 0.2 to 161.6 mm and lasting between more than for major events. Furthermore, for some
20 min and 13 h. During the total period, 37% of the individual throughfall gauges values exceeded gross
incident rain fell in single showers of less than 2 mm rainfall (e.g. 29% of the individual gauge values in the
and 92.3% of these showers contributed with less than SP were larger than gross rainfall, whereas in the HT
30 mm. Rainfall intensity, calculated for the total this was 30%, in the LT 27% and in the FP forest
period with some rainfall, averaged 5.46 mm h ⫺1 21%), but the average (of 60 and 40 gauges) was
with a maximum of 78.16 mm h ⫺1 (Fig. 2a). Most always lower than gross rainfall.
showers (63%) fell during the afternoon and at night Throughfall was calculated as a percentage of gross
(Fig. 2b) and 56% of these storms fell in less than 1 h rainfall for five different rainfall sizes and from the
(Fig. 2c). Monthly rainfall distribution during the five- totals of the measured daily gross rainfall and
year period shows that there was a slightly drier throughfall during the study. Throughfall ranged
period from December to February (Fig. 2d). Compar- from zero, with events below 2 mm, to 95% in storms
ing our data on five years rainfall with data from larger than 100 mm, but mean throughfall varied from
earlier years in the Middle Caquetá (Duivenvoorden 50 to 93% depending on gross rainfall amounts and
and Lips, 1995), rainfall characteristics appear to be the type of forest (Table 2). The calculated value of
similar to the long-term average. total throughfall relative to total gross rainfall ranged
from 82 to 87% in the four forests.
4.3. Throughfall Although empirical regression equations provide
only site-specific information, they may indicate a
The variability of throughfall within a subplot was trend especially if explained variance is high. Therefore,
large, with the smallest variation in the FP forest,
although differences in average values between Table 3
Regression parameters of throughfall versus gross rainfall in four
subplots were small. The average coefficient of different forest ecosystems in Colombian Amazonia. (se) standard
variation (CV) of individual gauges in each plot was error of regression coefficient (Note: The equation for linear form is
0.285 (^0.10) in the SP, 0.306 (^0.07) in the HT, T ˆ a ⫹ bPg ; where T is throughfall amount and Pg is gross rainfall
0.279 (^0.09) in the LT and 0.225 (^0.08) in the (mm))
FP forest. The CV of the mean throughfall in each Landscape unit a b se R2 n
subplot was 0.062 (^0.058) in the SP, 0.043
(^0.05) in the HT, 0.046 (^0.04) in the LT and Sedimentary plain ⫺1.02 0.926 0.003 0.99 102
0.047 (^0.04) in the FP forest. As a general trend, High terrace ⫺1.02 0.918 0.003 0.99 97
Low terrace ⫺1.07 0.906 0.004 0.99 97
for small rainfall events the value of the standard
Flood plain ⫺1.48 0.887 0.003 0.99 84
deviation (std) of throughfall (expressed as a percen-
 C. Tobón Marin et al. / Journal of Hydrology 237 (2000) 40–57 49

Fig. 3. Average throughfall and its standard deviation (std) against gross rainfall in a forest ecosystem (Sedimentary plain) in the Middle
Caquetá, Colombian Amazonia.

regressions of throughfall versus gross rainfall 4.4. Stemflow

were computed from single storms for each forest
(Table 3). Average throughfall per plot was highly Large differences were observed in the amount of
correlated with gross rainfall in all forests (Fig. 3). stemflow of individual trees and among subplots. In
ANOVA analysis showed that the ratio of mean general, however, the contribution of stemflow to net
throughfall over gross rainfall in the FP forest was rainfall was very low. The average CV in each plot
significantly different from the other forests (at 95% was 0.295 (^0.12) in the SP, 0.207 (^0.12) in the HT,
level). 0.323 (^0.20) in the LT and 0.303 (0.26) in the FP

Fig. 4. Average stemflow and its standard deviation (std) against gross rainfall in a forest ecosystem (Sedimentary plain) in the Middle Caquetá,
Colombian Amazonia.
50 C. Tobón Marin et al. / Journal of Hydrology 237 (2000) 40–57

Table 4 Upon rainfall, in all forests stemflow increased

Summary statistics for regressions of daily stemflow against gross very gradually until a threshold of about 25 mm
rainfall, in four forest ecosystems in the Colombian Amazonia
gross rainfall is reached (Fig. 4). However, values
(Note: The power form is Ps ˆ c…Pdg † where c and d are the regres-
sion coefficients for stemflow (mm)) tend to scatter with increasing rainfall. The rela-
tionship between measured stemflow and gross
Landscape unit Regression coefficient se R2 n rainfall could be described with a power function
c d
(Table 4). Some rainfall events smaller than 3 mm
Sedimentary plain 0.0015 1.53 0.049 0.92 86 did not produce stemflow in most plots, which
High terrace 0.0020 1.467 0.038 0.94 92 explains the lower number of events (n) reported
Low terrace 0.0029 1.423 0.035 0.95 87 for stemflow regressions. For rainfall events with
Flood plain 0.0031 1.325 0.050 0.91 73 an intensity ⬎5 mm h ⫺1, stemflow showed no clear
relationship with tree trunk area or bark texture
forest. The percentage of stemflow in all plots varied …R2 ˆ 0:3†: Nevertheless, there seems to be an
from 0.2 to 3.2% of gross rainfall. The total average inverse relationship between crown area and the
percentage of stemflow relative to gross rainfall was amount of collected stemflow for each tree. We
0.85% (^0.46) in the SP, 0.94 (^0.51) in the HT, 1.45 also observed that lower parts of tree trunks with
(^0.88) in the LT and 1.12 (^0.56) in the FP forest. fibrous bark texture were slowly wetted during
Differences are mainly due to the higher contribution long storm events, which points to high water
of tree palms to the total stemflow per plot. For storage.
palms, high-capacity collectors (more than 35 l)
were required to measure the incoming water. In 4.5. Evaporation
subplots with abundant palms, these palms produced
about 43% of total stemflow. Evaporation of intercepted water by the forest

Fig. 5. Evaporation against gross rainfall in the sedimentary plain forest ecosystem, Middle Caquetá, Colombian Amazonia.
 C. Tobón Marin et al. / Journal of Hydrology 237 (2000) 40–57 51

Table 5
Statistics of evaporation from wet forest canopy in single rainfall events in four forest ecosystems in the Middle Caquetá, Colombian Amazonia.
Equations are of the form Ew ˆ e ⫹ ft ; (e in mm and f in mm h ⫺1) (Note: e and f are the regression coefficients for the linear function of
evaporation loss during rainfall events)

Forest Total Total Total time Evap. during Evap. rate Ew versus rainfall R2 n
rainfall throughfall stemf. (h) rainfall (Ew) from wet duration. Regression
(mm) (mm) (mm) mm canopy coeffficient
mm h ⫺1
e f 178

SP 3273.8 2853.7 32.4 557.0 190.1 0.342 ⫺0.424 0.46 0.86 178
HT 3293.0 2854.8 36.2 464.2 207.4 0.447 ⫺0.263 0.52 0.75 160
LT 3158.4 2711.7 38.6 487.8 201.1 0.412 ⫺0.351 0.52 0.82 163
FP 3120.9 2555.0 30.5 472.3 320.0 0.677 ⫺0.366 0.78 0.88 140

canopy is calculated by subtracting the measured that reason, the ratio of evaporation over gross rainfall
daily throughfall and stemflow from gross rainfall. was plotted against the mean forest cover (1-gap
Furthermore, it is related to gross rainfall character- fraction) established for each forest. Fig. 7 indicates
istics and forest system parameters. Following net that there is an increase of evaporation from the wet
throughfall trends, the percentage of evaporation rela- forest canopy with increasing canopy cover. This
tive to gross rainfall varied from 6 to 100% in all figure also shows an inflexion in the curve, i.e. a steep-
forests, depending on rainfall size. Mean evaporation, ing from the low terrace to the flood plain, which
expressed as percentage of total gross rainfall, also indicates that also other forest structural parameters
differed between the forests: 11.84 (^2.4) in the SP, (e.g. leaf surface characteristics) affected canopy
12.24 (^1.2) in the HT, 12.92 (^1.1) in the LT and interception.
17.15 (^0.96) in the FP forest. For small storms (less
than 2 mm), evaporation values were very close to
those of gross rainfall. For heavy showers, however, 5. Discussion
the relative value of evaporation became smaller
(Fig. 5). Storage capacity values of the forests studied on the
An assessment of the cumulative evaporation whole resemble the values found by Ubarana (1996)
during rainfall (Ew) was calculated with Eq. 1 for in the reserves Vale do Rio Doce and Jaru Duke in
the single rainfall events in each forest. Additionally, Brazil, which were based on linear regressions of
Ew was related to rainfall duration. The average throughfall against gross rainfall. However, Ubarana
evaporation rate during the rainfall varied from 0.34 concluded that this method results in an overestima-
to 0.68 mm h ⫺1 among the forests (Table 5) and Ew tion of evaporative losses. Since our estimates of the
exhibited a linear relation with rainfall duration for all canopy storage capacity were based on specific events
forests (Fig. 6). Negative values correspond to rainfall for which it was assumed that evaporation was negli-
events of short duration and low intensity, indicating gible, it might be that we somewhat overestimated the
that the forest canopy did not reach saturated condi- storage capacity by neglecting evaporation during the
tions during such events. selected events. The storage capacity of the flood
Though climatic conditions are similar in the plain forest is higher than the values commonly
forests studied, there is a clear difference in amounts reported in Amazonian rain forest studies. This can
of evaporation when comparing similar rainfall be explained by the fact that most of the latter
events. This implies that amounts of evaporation studies were executed in so called “terra firme” forests
from the wet forest canopy did not only depend (non-flooded ecosystem). As we studied a broader
upon gross rainfall and climate conditions. Differ- range of ecosystems, differences between parameters
ences in evaporation between these close-by forests values reported here and those reported in other
may be related to differences in their structure. For studies therefore should be interpreted in terms of
52 C. Tobón Marin et al. / Journal of Hydrology 237 (2000) 40–57

differences in forest structure between the ecosystems numbers of photographs have been taken and
studied. analysed.
The free throughfall coefficients of our forests, The range in our values for throughfall and stem-
derived from the data on selected small storms (pt), flow, expressed as percentage of gross rainfall, is
conform to the values found by Jetten (1996). similar to the range in values reported in earlier
However, these values which range from 0.27 to studies on rainfall partitioning in Amazonian rain
0.59 clearly differ from those derived from the forests (Table 6). This most probably also explains
scanned photographs (ps) and their application why such variability exist in the latter values, i.e. it
resulted in an overestimation of net rainfall rates in is probably largely due to differences in rainfall char-
all forests. According to Ubarana (1996) this may be acteristics and forest structure between the forests
explained by the waxy nature of the tree leaf surfaces studied. That coefficients of determination for our
causing the drops to splash off, which thus contribute regressions are significantly higher than most values
to throughfall before the canopy is saturated. Addi- presented in the literature can be explained by the size
tionally, our values were based on the results for of our data set, which is much larger than in earlier
small showers producing throughfall and this may studies. Our results also indicate that the partitioning
have influenced the estimation of the free throughfall of rainfall depends, among others, on the size of the
because of the low frequency of such events (for the rainfall event. Moreover, it is clear from the relation
research sites only 7–10 events were registered between throughfall and storm size (Table 2) that the
during the total period). Photographs taken from high CV of throughfall is the result of the large varia-
the forest canopy under non-direct sunlight and bility in rainfall classes.
with covered sky can easily be analysed, the white To define the total error (t.e.) of the mean through-
pixels reflecting the non-covered part. This estimate fall as a percentage of gross rainfall, we applied the
may provide better results, especially if large proposed formula for random relocation of n gauges

Fig. 6. Evaporation from the wet forest canopy (Ew) in relation to rainfall duration, in a forest ecosystem (sedimentary plain) in the Middle
Caquetá, Colombian Amazonia.
 C. Tobón Marin et al. / Journal of Hydrology 237 (2000) 40–57 53

Fig. 7. Fraction of evaporation from gross rainfall as a function of forest cover fraction (FC ˆ 1-gap fraction) in the four forest ecosystems.
Middle Caquetá, Colombian Amazonia.

by Lloyd and Marques (1988), although larger Whether throughfall percentage depends on storm
diameter funnels were used in the current study size remains to be established, as clearly stated by
p Lloyd and Marques (1988). Accordingly, we investi-
t:e: ˆ s:e:…1 ⫹ N=nm† …2† gated the effect of storm size on the variability of
throughfall percentage by using only those single
where s.e. is the standard error resulting from the events that were measured during the 20-months
random relocations of collectors, expressed as the period in which weekly data of throughfall and
best estimate of the standard error of mean throughfall gross rainfall were collected. During that period, the
in each collector, under the assumption that the speci- same methodologies were used with the exception
fic canopy structure is properly described by N that gauges were not relocated. Fig. 8, given as an
(number of grid points) and m (the relocation of example, shows that the variation in the ratio of
collectors). Based on the formula of Lloyd and throughfall from a single funnel over the average of
Marques (1988), the arrangement of 60 funnels with 20 funnels tends to decrease as storm size increases.
23 relocations in the SP forest results in a total error in This is a trend observed for most non-moving collec-
measured throughfall of 3.5% of gross rainfall, tors but also for the relocated collectors, as stated
whereas in the other ecosystems with 40 gauges and earlier in this paper, which suggests that storm size
23 relocations the error is 3.8% due to variation in also affect throughfall variability in our ecosystems.
canopy structure. These figures are lower than those We did not fully investigate the relation between
found by other authors, which can be explained by the storm size and throughfall (expressed as percentage of
continuous relocation of our collectors. Nevertheless, gross rainfall). Nevertheless, we conclude from our
our values for t.e. are larger than those found by Lloyd results that when the method of relocation of collec-
and Marques (1988). tors is used to estimate this throughfall, it is essential
54 C. Tobón Marin et al. / Journal of Hydrology 237 (2000) 40–57

Table 6
Partitioning of gross rainfall (percentages) in Amazonian rain forests

Location Forest type Throughfall % Stemflow % Evaporative loss % Reference

Venezuela Catinga 91 0.8–14 – Herrera (1979)

Brazil Rain forest 80.2 – 19.8 Franken et al. (1992)
Brazil Rain forest 87–91 1.8 8.9(^3.6) Lloyd and Marques (1988)
Brazil Rain forest 86–87 0.8–1.4 11.6–12.9(^5.9) Ubarana (1996)
Colombia Rain forest 82–87 0.9–1.5 12–17 This study

that this relocation is preceded by sampling of a wide Though the observed relationships may contribute
range of storms sizes with a fixed set of collectors, in little to the explanation of the hydrological processes
order to assess the combined effect of site forest at canopy level, the models nevertheless provide clear
structure and rainfall characteristics on throughfall. indications for the extent to which this partitioning is
Lastly, although litter fall in the ecosystems studied controlled by the parameters used. While linear func-
exhibits some temporal dynamics (Duivenvoorden tions produce better fits for correlation between
and Lips, 1995), no relation was observed between throughfall and gross rainfall, power functions
throughfall percentage and litter fall. Such relation produce better fits for such correlation with stemflow,
was reported for a Bornean rain forest by Burghouts in terms of the significance levels and standard devia-
et al. (1988). tion of residuals. The linear regression equations of
The variability of stemflow in mature tropical rain throughfall versus gross rainfall fit most points and
forest has been attributed to the high species diversity have a high coefficient of determination in all ecosys-
(Hutjes et al., 1990; Hertwitz, 1985) and this variabil- tems. Nevertheless, their application to very small
ity certainly is larger in tropical forests than in tempe- storms (lower than 2 mm) results in negative values
rate forest plantations (Lloyd and Marques, 1988). In for throughfall and they slightly underestimate
the present study, this parameter was estimated for throughfall for very high-rainfall events, which illus-
different tree species with different diameter. Stem- trates the limitation of regressions, which fit a curve to
flow values from this study ranged from 0.9 to 1.5% of a set of data.
gross rainfall, which is within the range of values Throughfall and gross rainfall were highly
presented in other studies on similar forest types correlated in all forests, which is probably due to
(Table 6). Although the contribution of stemflow to the similarity of our forests with regard to relevant
net rainfall was very low, it probably causes an impor- system parameters. However, the correlation between
tant input of solutes to the forest floor, concentrated interception values and gross rainfall is less promi-
around the base of trees. Results suggest that little nent, the coefficient of determination of the regression
water was stored in excess of the storage capacity of being distinctly lower (R 2 between 0.66 and 0.83). In
the stem elements, as indicated by the very small other words, throughfall percentages can be predicted
stemflow quantities collected once rainfall has ceased with a high accuracy based on data on rainfall
or during small storms. This can be explained by the amounts and characteristics, whereas for the predic-
presence of some tree trunks with hydrophobic bark tion of interception other parameters, such as forest
(personal observations 1992–1997) and of bark with structure, must be included.
fibrous texture. Upon rainfall, tree species with these Although our values are within the range of inter-
characteristics exhibited significant stemflow, even ception values reported in other studies from the
without being completely wet. However, once rainfall Amazon basin (Table 6), the values found for the FP
stopped, there was a sharp decline in stemflow. and LT forests are rather high compared to those
We found static models to be capable of describing reported in earlier Amazonian studies (Lloyd and
rainfall partitioning for the forests studied. The applic- Marques, 1988). This is in line with the higher canopy
ability of these models is most probably restricted to storage capacity of our forests as compared to those
the area and conditions during the period of research. described in these other studies (e.g. Lloyd and
 C. Tobón Marin et al. / Journal of Hydrology 237 (2000) 40–57 55

Fig. 8. The ratio of single gauge throughfall over average throughfall from 20 gauges, as a function of gross rainfall size.

Marques, 1988). Thus, even though climatic condi- tion. Although not evaluated in this study, it might
tions for our forests were similar, forest interception also depend on specific climatic conditions (e.g.
varied, due to differences in amounts and character- wind speed). Additionally, a distinct relationship
istics of the rainfall and in forest structure. The rela- seemed to exist with the forest cover fraction
tive proportion of evaporation from our forests was (Fig. 7). Although it should be realised that the
also higher than the values reported in these earlier number of forests studied is small and the relationship
studies. This can be attributed to the higher gross rain- is rather uncertain, it may serve for the estimation
fall in our forests, at least when compared with the of evaporation by a forest for which measurements
mean annual value of 2500 mm reported from central are not available. Provided that climatic conditions
Amazonia (Leopoldo et al., 1987). are similar, such estimations mainly rely upon an
In many studies on rainfall interception, it is adequate estimation of the gap fraction or LAI.
concluded that leaf surfaces determine the intercep-
tion storage capacity of woody plants (Hertwitz, 1985;
Gash, 1979; Singh, 1977; Rutter et al., 1975). Studies 6. Conclusions
in the research plots, using destructive methods and
derived regression equations for leaf biomass estima- Of the gross rainfall of about 3400 mm y ⫺1, most
tion (Overman et al., 1990; Alvarez, 1993), showed fell in small showers during the afternoon and at night.
that leaf biomass is higher in the flood plain forest The overall average rainfall intensity was about
(9.5 tonnes/ha) than in the other forests from which 5 mm h ⫺1. These rainfall characteristics largely
a higher leaf surface area can be inferred. Therefore, explain the partitioning of rainfall into throughfall,
the relatively high interception by the flood plain stemflow and ensuing evaporation in the forests
forest may be explained by its higher leaf biomass. studied.
Cumulative evaporation during rainfall events (Ew) Water fluxes in the forest canopy of four forest
from our forests strongly depended on rainfall dura- ecosystems in western Amazonia have been quantified
56 C. Tobón Marin et al. / Journal of Hydrology 237 (2000) 40–57

as a percentage of gross rainfall. Amounts of net to the Tropenbos Foundation, E-mail: ftropenb@co-
precipitation reaching the forest floor and evaporation lomsat.net.co
from the wet forest canopy varied for the forests
studied: the SP forest had the highest percentage of
References
throughfall relative to gross rainfall and the FP forest
the lowest. The observed differences in throughfall, Alvarez, D.E., 1993. Composición florı́stica, diversidad, estructura
stemflow and evaporation can partly be attributed to y biomasa de un bosque inundable, en la Amazonia Colombi-
differences in forest structure (gap fraction). Their ana. Ir dissertation, Universidad de Antioquia, Colombia.
range is similar to the overall range in these para- Bosch, J.M., Hewlett, J.D., 1982. A review of catchment experi-
meters as published in earlier studies from the ments to determine the effect of vegetation changes on water
yield and evapotranspiration. J. Hydrol. 55, 3–23.
Amazon basin, implying that the latter variability Bruijnzeel, L.A., 1990. Hydrology of moist tropical forests and
may very well be connected with differences in forest effects of conversion: a state of knowledge review. UNESCO,
structure. Paris.
Results from the forests studied provide some Burghouts, T.B.A., van Straalen, M.N., Bruijnzeel, L.A., 1998.
insight into the rates of evaporation from a wet forest Spatial heterogeneity of elements and litter turnover in a
borneam rain forest. J. Trop. Ecol. 14, 477–506.
canopy and strengthen the understanding of the Duivenvoorden, J.F., 1995. Tree species composition and rain
contribution of forests to atmospheric moisture. The forest-environment relationships in the Middle Caquetá area,
mean evaporation rate from a wet forest canopy Colombia, NW Amazonia. Vegetatio 120, 1–113.
during rainfall events in the Middle Caquetá, (Colom- Duivenvoorden, J.M., Lips, J.M., 1995. A land-ecological study of
bian Amazonia) was estimated at 0.47 mm h ⫺1 and it soils, vegetation and plant diversity in Colombian Amazonia.
PhD thesis, Landscape and Environmental Research Group,
increased with increasing forest cover. Moreover, this Faculty of Environmental sciences, University of Amsterdam
study of throughfall, stemflow and evaporation in a Tropenbos Foundation, Wageningen.
range of forests demonstrates the relevance of forest Eltahir, E.A., Bras, R.L., 1993. On the response of the tropical
structure for the evaporation of rainfall intercepted by atmosphere to large-scale deforestation. Q.J.R. Meteorol. Soc.
the forest canopy and for the net precipitation reach- 119, 779–793.
Ford, E.D., Deans, J.D., 1978. The effects of canopy structure on
ing the forest floor. The results show that within the stemflow, throughfall and interception loss in a young Sitka
scope of this research forest structure can be spruce plantation. J. Appl. Ecol. 15, 905–917.
adequately characterised by the gap fraction and Franken, W., Leopoldo, P.R., Matsui, E., Ribeiro, M. 1992. Estudo
LAI. These structural characteristics together with da Interceptaçâo da água de chuva em cobertura florestal
the rainfall amount and rainfall duration are the amazônica do tipo terra firme. Acta Amazonica, 12, 327–331.
Gash, J.H.C., Nobre, C.A., Roberts, J.M., Victoria, R.L. (Eds.),
main parameters determining rainfall partitioning in 1996. Amazonian Deforestation and Climate Wiley, Chichester,
the western Amazonian rain forests. UK (611pp.).
Gash, J.H.C., 1979. An analytical model of rainfall interception by
forests. Q.J.R. Meteorol. Soc. 105, 43–55.
Acknowledgements Gash, J.H.C., Morton, A.J., 1978. An application of the Rutter
model to the estimation of the interception loss from Thetford
forest. J. Hydrol. 38, 49–58.
We are grateful to Dr John Gash from the UK Insti- Herrera, R.A., 1979. Nutrient distribution and cycling in an Amazon
tute of Hydrology and to Dr Sampurno Bruijnzeel caatinga forest on Spodosols in southern Venezuela,
from the Free University, Amsterdam for their Dissertation, University of Reading.
suggestions and corrections of earlier drafts of this Hertwitz, S.R., 1985. Interception storage capacities of tropical rain
forest canopy trees. J. Hydrol. 77, 237–252.
paper. This work, forming part of a larger research Hölscher, D., de A Sa, T., Bastos, T., Denich, M., Fölster, H., 1997.
project on water and nutrient cycling in Colombian Evaporation from young secondary vegetation in eastern
Amazonia, was supported by the Tropenbos Founda- Amazonia. J. Hydrol. 193, 293–305.
tion in Colombia and the Netherlands and by the Hutjes, R.W.A., Wierda, A., Veen, A.W.L., 1990. Rainfall
Colombian Institute for Science and Technology interception in the Tai forest, Ivory Coast: application of two
simulation models to a humid tropical system. J. Hydrol. 114,
“Colciencias”. Collected data (four years data on 259–275.
gross rainfall, throughfall, stemflow and evaporative Jetten, V.G., 1996. Interception of tropical rain forest: performance
loss from western Amazonia) is available on request of a canopy water balance model. Hydrol. Process 10, 671–685.
 C. Tobón Marin et al. / Journal of Hydrology 237 (2000) 40–57 57

Jordan, C.F., 1978. Stemflow and nutrient transfer in a tropical rain Estimación de la biomasa aérea en el bosque del medio Caquetá,
forest. Oikos 31, 257–263. Colombia. Colombia Amazonica 4 (2), 135–147.
Klaassen, W., Bosveld, F.C., de Water, E., 1998. Water storage and Roberts, J.M., Cabral, O.M.R., da Costa, J.P., McWilliam, A.L.C.,
evaporation as constituents of rainfall interception. J. Hydrol. de A. SÁ, T.D., 1996. An overview of the leaf area index and
212–213, 35–50. physiological measurements during ABRACOS. Amazonian
Klinge, H., Rodriguez, W.A, Brunig, E.F., Fittkau, E.J., 1975. Deforestation and Climate. Wiley, Chichester, UK (611pp.).
Biomass and structure in a central Amazonian Rain Forest. In: Rutter, A.J., Morton, A.J., Robins, P.C., 1975. A predictive model of
Golley, F.B., Medina, E. (Eds.). Tropical Ecological Systems. rainfall interception in forests. II Generalisation of the model
Trends in Terrestrial and Aquatic Research, Springer, Berlin, and comparison with observations in some coniferous and
pp. 115–122. hardwood stands. J. Appl. Ecol. 12, 367–380.
Köppen, W., 1936. Das geographische System der Klimate. Salati, E., Vose, P.B., 1984. Amazon Basin: a system in
Handbuch der Klimatologie. Verlag Gebrüder Borntraeger, equilibrium. Science 225, 129–138.
Berlin (44pp.). Salati, E., Dall’olio, A., Matsui, E., Gat, J.R., 1979. Recycling of
Leopoldo, P.R., Franken, W., Villa Nova, N., 1995. Real evapora- water in the Amazon Basin: an isotopic study. Water Resources
tion and transpiration through a tropical rain forest in central Res. 15, 67–71.
Amazonia as estimated by the water balance method. For. Ecol. Shuttleworth, W.J., Calder, I.R., 1979. Has the Priestley–Taylor
Manag. 73, 185–195. equation any relevance to forest evaporation? J. Appl. Meteorol.
Leopoldo, P.R., Franken, W., Salati, E., Ribeiro, M.N., 1987. 18, 639–646.
Towards a water balance in central Amazonia region. Shuttleworth, W.J. 1988. Evaporation from Amazonian rain forest.
Experentia 43 Birkhäuser Verlag, CH-4010 Basel/Switzerland. Proc. R. Soc., London, B233, 321–346.
Lesak, L.F.W., 1993. Water balance and hydrologic characteristics Singh, B., 1977. The effect of rainfall characteristics and post-
of a rain forest catchment in the Central Amazon Basin. Water wetting synoptic characteristics on evaporation rates from a
Resour. Res. 29 (3), 759–777. wetted hardwood canopy. Climatol. Bull. 21, 12–33.
Lettau, H., Lettau, K., Molion, L.C.B., 1979. Amazonian’s Hydrol- Stewart, J.B., 1977. Evaporation from the wet canopy of a pine
ogy cycle and the role of atmospheric recycling in assessing
forest. Water Resour. Res. 13, 915–921.
deforestation effects. Mon. Weather Rev. 107, 227–238. Tanner, E.V.J., 1985. Jamaican Montane forests: Nutrient capital
Londoño, V.C., 1993. Análisis estructural de dos bosques asociados
and coast growth. J. Ecol. 75, 553–568.
a unidades fisiográficas contrastantes, en la region de
Ter Stegee, H., (1994). HEMIPHOT: a programme to analyse vege-
Araracuara (Amazonia Colombiana). Ir Dissertation,
tation indices, light and light quality from hemispherical photo-
Universidad Nacional de Colombia.
graphs. Tropenbos Documents 3. The Tropenbos Foundation,
Longman, K.A., Jenı́k, J., 1990. Tropical Forest and its
Wageningen, The Netherlands.
Environment. Longman, New York.
Ubarana, V.N., 1996. Observations and modelling of rainfall
Lloyd, C.R., Marques, F.A.de.O., 1988. Spatial variability of
interception at two experimental sites in Amazonia. In: Gash,
throughfall and stemflow measurements in Amazonian rain
J.H.C., Nobre, C.A., Roberts, J.M., Victoria, R.L. (Eds.).
forests. Agri. For. Meteorol. 42, 63–73.
Amazonian Deforestation and Climate, Wiley, Chichester, UK.
Molion, L.C.B., 1975. A climatonomic study of the energy and
Vitousek, P.M., Denslow, J.S., 1986. Nitrogen and phosphorous
moisture fluxes of the Amazon’s basin with considerations of
availability in treefall gaps of a lowland tropical rain forest.
deforestation effects. PhD thesis, University of Wisconsin,
Madison. J. Ecol. 74, 1157–1178.
Nobre, C.A., Sellers, P., Shukla, J., 1991. Amazonia deforestation Waidi, S., Wong, W.M., Douglas, I., 1992. Throughfall, stemflow,
and regional climate change. J. Climate 4, 957–988. overland flow and throughflow in the Ulu Segama rain forest,
Nobre, C.A., Coauthors, 1996. The Large-scale Biosphere-Atmo- Sabah Malaysia. Phil. Trans. R. Soc. London. B. 335, 389–395.
sphere Experiment in Amazonia (LBA). Concise experiment Wright, I.R., Gash, J.H., da Rocha, H.R., Shuttleworth, W.J., Nobre,
plan; LBA Science Planning Group, Staring Center-DLO, The C.A., Maitelli, G.T., Zamparoni, C.A., Carvalho, P., 1992. Dry
Netherlands. season micrometeorology of central Amazonian ranchland.
Overman, J.P.M., Saldarriaga, J.G., Duivenvoorden, J.F., 1990. Q.J.R. Meteorol. Soc. 118, 1083–1099.