The development of overland flow in a tropical rainforest catchment

The development of overland flow in a tropical rainforest catchment

Journal of Hydrology, 39 (1978) 365--382 365 © Elsevier Scientific Publishing Company, Amsterdam -- Printed in The Netherlands [1] THE DEVELOPMENT ...

1014KB Sizes 1 Downloads 38 Views

Journal of Hydrology, 39 (1978) 365--382


© Elsevier Scientific Publishing Company, Amsterdam -- Printed in The Netherlands



Department of Geography, James Cook University of North Queensland, Townsville, Qld. 4811 (Australia) Queensland Forestry Department, Atherton, Qld. 4883 (Australia) (Received September 27, 1977; revised and accepted April 1, 1978)

ABSTRACT BoneU, M. and Gilmour, D.A., 1978. The development of overland flow in a tropical rainforest catchment. J. Hydrol., 39: 365--382. A preliminary assessment is made o f the factors contributing to the rapid response of the discharge hydrograph in a tropical rainforest catchment. Overland and subsurface flows were measured by means of a trough system constructed at the lower end of three sites. Water was led from each trough into a tipping-bucket device with the number of tips being recorded mechanically using a hay-bale counter and electronically by a digital event recorder. Substantial overland flow occurred at all sites despite the exceedingly high average saturated hydraulic conductivity, k s values in the top 20 cm of the soil profile. The highest ratio o f overland flow to subsurface flow was associated with a plot having an upslope position. Subsurface flow volumes, however, were higher at the two remaining plots of lower-slope position. This was attributed to the presence of residual rocks in the deeply weathered clays. The occurrence o f widespread overland flow was explained by the occurrence of high-intensity rainfalls (particularly during the summer months). Such rainfalls frequently exceed the average k s values below 20 cm depth and therefore generate a widespread perched water table within the t o p 20 cm. This causes subsurface flow (particularly on steep slopes) through the highly permeable surface horizon. Additional rain causes the perched water table to emerge at the soil surface and hence widespread saturation overland flow results for the remainder o f the storm. This overland flow is tapped by a temporary dense drainage network, thus accounting for the rapid quickflow response. Spearman's rank correlation coefficient analysis emphasised the significance of rainfall intensity despite some detailed differences between " s u m m e r " and " w i n t e r " rainfall events. It is concluded that in this particular environment the variable source area concept of storm flow generation, associated with humid temperate areas, is not applicable mainly due to the characteristics o f rainfall intensity.


The aim of this paper is to make a preliminary assessment of the factors contributing to the rapid response of the discharge hydrograph within a tropical rainforest catchment in northeast Queensland. The majority of the


discussion will concentrate on the origin of overland flow, supported by the use of Spearman's rank correlation analysis, using data collected during the 1976 wet season. Some general comparisons are made with the drainage processes in humid temperate environments. There has been a considerable a m o u n t of w o r k accomplished in the humid temperate latitudes evaluating the origin of quickflow. Freeze (1972b), Dunne et al. (1975) and Ward {1975) have given detailed summaries of this work. The general consensus is that widespread overland flow does n o t occur in forested areas b u t is more locally confined to areas having thin soils, swales and the lower reaches of concave slopes adjacent to streams. However, in terms of contribution to storm r u n o f f the areas of thin soils and swales are only of significance if they occur at the f o o t of the concave slopes in the presence of a shallow water table. The surface emergence of this water table resulting from recharge by subsurface flow and rainfall leads to what Dunne and Black (1970a, b) termed saturation overland flow which expands upslope during the course of a storm. This forms the basis of the variable source area concept. The reason for this very localised nature can be attrib u t e d to the very high infiltration capacities which, under undisturbed conditions, occur in forest soils resulting from the continued incorporation of organic matter plus a high r o o t density. Therefore there is a consequential improvement in soil structure which is further assisted b y the presence of well developed A-horizons. Consequently, rainfall intensities are usually less than the soil infiltration capacities, and normal storm durations are insufficient to create ponding at the soil surface. This is in marked contrast with the more widespread overland flow found in certain semi-arid areas following Horton's {1933, 1945) philosophy. Thus for temperate humid areas most writers favour some version of the Hewlett variable source area c o n c e p t (Hewlett, 1961; Hewlett and Hibbert, 1967) of overland flow generation, despite few detailed attempts at field measurement (Ward, 1975). This has led to increasing attention being paid to the recognition and statistical prediction of such areas so that the hydrograph c o m p o n e n t s can be modelled (Dunne et al., 1975; Hewlett and Troendle, 1975). The only contentious issue is the principal mode of recharge of the water table whether it be interflow (e.g., Whipkey, 1965, 1969; Hewlett and Hibbert, 1967) or vertical infiltration {e.g., Dunne and Black, 1970a, b). But as Ward (1975) noted: "... n e i t h e r ... is likely t o be rapid enough t o m a k e a significant contribution t o s t o r m r u n o f f e x c e p t t h r o u g h the medium o f t r a n s l a t o r y f l o w o r t h r o u g h p i p e f l o w in s a t u r a t e d


Part of this debate depends o n the topographical--hydrological circumstances of slopes. The m o s t favourable circumstances for the variable source area c o n c e p t to apply are regarded to be where there are extensive saturated or near saturated valley b o t t o m s at the f o o t of concave slopes. Saturation overland flow is regarded as being less significant in steep, well drained and per-


meable hillslopes bordering narrow valley floors. This increases the significance of subsurface storm flow contribution to the hydrograph although volumetrically it is still regarded to be small. This was demonstrated by Freeze (1972b) using a mathematical model despite rigid assumptions. By comparison the humid tropics have received only passing attention during a consideration primarily of denudation problems. Consequently, there is no certainty that humid temperate processes are as valid in this environment. Perhaps the most commonly held belief is that expressed by Thomas (1973) when he stated that: "... rapid infiltration of heavy rain into the deep tropical soils prevents much surface flow taking place except on steep slopes ..."

and attributes this to be due to the kaolin clays being: "highly permeable, and the intensity of precipitation at ground level is much less than outside the forest."

Thus Thomas favours slower storm recharge to streams: "... by means of lateral m o v e m e n t or throughflow within the soil and regolith."

Whilst acknowledging the significance of subsurface flow Douglas (1973), however, noted that: "... under undisturbed forest cover there is little vegetation on the ground surface, and surface wash, derived from stemflow, may add to the transport of sediment towards stream channels by subsurface flow."

The surface wash component was particularly emphasised by Ruxton (1967) during a qualitative survey of the denudation processes occurring on slopes under tropical rainforest in Northern Papua. He pointed out that open rather than closed canopies occurred in this environment due to tree fall caused by monsoonal gales and mass movement. This created less protection from rainfall on the forest floor (c.f. Thomas, 1973), which made raindrop and waterdrop erosion more effective. He suggested that overland flow commenced as soon as the soft surface was sealed by fine particles disrupted by raindrop impact. Furthermore Ruxton did not regard overland flow as only operative when the soil was saturated from heavy, prolonged rain but like Douglas (1973), supported the role of stem flow. As he commented: "... in this way water is c o m m o n l y supplied to the soil at the tree base faster than it can infiltrate, and runoff may therefore occur after only moderate rainfalls."

The significant feature is that none of these writers is able to support their


views with quantitative evidence. Furthermore the matter of spatial organization and magnitude of overland f l o w remains inconclusive. DESCRIPTION



The site chosen for field experimentation was in an existing experimental catchment ca. 6 km east of Babinda (Fig.l). The climate of the area is typical of that experienced on the wet tropical coastal belt with high annual rainfall totals being an outstanding feature. The average annual rainfall on the site for the 6-yr. period 1970--1975 was 4 1 7 5 mm, and about half of this total falls during the wet-season months of January to March. The regular occurrence of cyclones and tropical lows associated with the intertropical convergence zone in the summer months produces rainfall intersities which are among the highest recorded in Australia -- daffy falls in excess of 250 mm are fairly frequent. Later in the year the majority of rainfall originates from moist on-shore SE winds associated with a firm high-pressure coastal ridge and the occasional eastward progression of upper atmospheric lows. In both cases rainfall intensities are usually much lower than in the wet METEOROLOGICAL A PLOT v





i i





!:.~:ii ~::



. :~'x :





TAS. ~, )

• ~,'50 m









STUDY/ ~ SITE ~ 12



150m 0






300 m


Fig. 1. The location of study area and experimental plots in South Creek catchment.


season proper, although daffy totals can still be quite high with recordings occasionally in excess of 100 mm. The experimental catchment viz. South Creek, in which the study sites are located is 0.257 km 2 in area and has steep slopes (mean catchment slope of 34%). The softs are yellow-brown clays derived from deeply weathered basic metamorphic rocks. Under the Northcote (1971) system, they are classified as Gn 3.74 and Gn 3.11/3.14, respectively (tt.F. Isbell and G.G. Murtha, pers. commun., 1976). The catchment is covered by tropical rainforest -- classified Mesophyll Vine Forest according to Webb's (1959) classification. The forest has an uneven canopy resulting from disturbance by tropical cyclones. A portion of the catchment (about 37%) was logged between 1967 and 1969. However, disturbance was slight as the area was only lightly timbered. The detailed study plots were located in areas undisturbed by logging. The forest floor exhibits many of the features described by R u x t o n (1967) and Douglas (1973) which suggest the occurrence of widespread overland flow. These include undermining of lateral surface roots causing r o o t steps and the exposure and alignment of finer roots in a downslope direction. Subsurface flow was also indicated by examples of caving beneath the r o o t mat at the edge of streams in non-meandering situations. The stream gauging station is located at the lower end of the catchment and consists of a V notch weir and a water level recorder. Meteorological data are recorded in an adjacent clearing. EXPERIMENTAL DESIGN Three sites were chosen for detailed study -- two sites (sites la and l b ) were located adjacent to each other on the lower slopes of the catchment and the third (site 2) was located in an upper-slope situation in an area where first-order streams are prevalent (Fig.l). The slopes of sites la, I b and 2 are in respective order 62, 50 and 42%. A trough system was constructed at the lower end of each site to intercept lateral flow from different depths in the soil profile. A pit was dug, and 2 m long troughs were installed on steps at the soil surface and at depths of 0.25, 0.50 and 1.00 m below the surface. The troughs were let into the vertical face of the pit and sealed with mortar to prevent leakage. Water from each trough is led into a tipping-bucket device and the n u m b e r of tips of each bucket is recorded mechanically using a hay-bale counter. Mercoid® switches attached to the tipping buckets are connected to digital event recorders at each site so that the r u n o f f is recorded with a time base of 6 min. However, problems with the recorders resulted in an incomplete record during the 1976 w e t season. A raingauge at the adjacent meteorological site also provides rainfall information with a 6-min time base. In addition, a n e t w o r k of tensiometers, piezometers and deep wells was established upslope from t w o of the sites (la and 2) to provide information


on soft-moisture conditions and saturated zones. The tensiometers were set so that a zero tension was equivalent to a dial reading of 200 cm. In this way small positive pressures could be determined. A soil field testing programme was carried o u t to determine the pertinent soil parameters to aid in the interpretation of the observed lateral flow and stream flow data. A ring inffltrometer was used to determine sorptivity and saturated hydraulic conductivity, ks using a portable constant-head permeameter (Talsma, 1969, 1974) at locations close to the study plots. The ks factor was also determined by the auger hole pumping test method (e.g., Kirkham and Van Bavel, 1948; Van Beers, 1958) where the permanent saturated zone could be located at site 2, otherwise the shallow-well pump-in m e t h o d (Boersma, 1965) was applied at the same sites previously used by the ring infiltrometers. This gave ks values for the 0--10 and 10--20 cm depths by the constant-head permeameter, and 20--100 cm (approximately) by the shallow-well pump-in m e t h o d . The piezometers located at depths of 0.5, 1.00, 2.00 and 3.00 m only had water in them for short periods during and immediately after heavy rainstorms. However, the unsteady nature of the potentiometric head prevented any reliable estimates of ks using the piezometer pumping test m e t h o d (Luthin and Kirkham, 1949). RESULTS

In comparing the ks results from the soil field testing programme it should be noted that the ring infiltrometers measure primarily the vertical component as against the horizontal and vertical c o m p o n e n t s combined by the remaining methods. Nevertheless Table I shows the expected decline of ks with depth. However, more notable is the rate of decline and the exceedingly high magnitude of k s in the 0--10 cm layer. The coefficient of variation and absolute range demonstrate the high variability of these surface soils. A mean ks value of 32.45 m / d a y for the top 10 cm of the soil implies that the infiltration capacity would be far in excess of any possible rainfall intensity which could occur. The mean ks value of the 10--20 cm layer of 1.54 m / d a y (64 mm/h) means that only the higher rainfall intensities would exceed the percolation rate. Thus the prospect of widespread overland flow originating from the surface 20 cm would seem to be ruled out. Even the much lower ks below 20 cm is still sufficient to a c c o m m o d a t e moderate rainfall intensities (~< 13 mm/h). The surface configuration and pedological--geological structure of all three slopes associated with these sites would favour predominantly subsurface storm flow on the basis of temperate latitude experience. That is, they would be categorised as steep, well drained and permeable, and bordering on a narrow valley floor in the case of sites l a and I b. However, this postulation is quickly dismissed when a sample storm is examined using information from the continuous recorders at site 2. Fig.2 shows that overland flow rather than subsurface flow is prevalent despite the upslope location. Several


TABLE I The v a r i a t i o n in s a t u r a t e d h y d r a u l i c c o n d u c t i v i t y w i t h d e p t h Depth (m)

Mean k s (m/day)

Absolute range o f ks (m/day)

Coefficient of variatio~ (untrans. formed data)

Coefficient of variation (square r o o t transformation)

Sample size, n

0--0.1 0.1--0.2 0.2--1.0 (approx.) 5 . 5 2 - - 5 . 8 8 (Well 1) (Site 2) 3 . 7 2 - - 4 . 4 8 (Well 2 ) (Site 2)

32.454 P 1.545 P 0. 318 s 0.435 A

2.819--99.398 0.079--5.394 0.053--0.625

84.6 73.81 55.83

45.13 39.82 30.3

37 33 27

0.106 A

0.409, 0.461

0.104, 0.108


(a) k s derived f r o m : P = c o n s t a n t - h e a d p e r m e a m e t e r ; S = shallow-well p u m p - i n m e t h o d ; a n d A = a u g e r h o l e p u m p i n g test m e t h o d (Ernst, 1950). (b) All k s values c o r r e c t e d for viscosity at a t e m p e r a t u r e o f 20 ° C. This is a b o u t the average g r o u n d w a t e r t e m p e r a t u r e f o r S o u t h Creek. (c) T w o sets o f c o e f f i c i e n t o f v a r i a t i o n are given because t h e c o e f f i c i e n t o f s k e w n e s s was s u b s t a n t i a l l y positive in t h e case o f 0--0.1 m ( + 1 . 0 3 ) a n d 0 . 1 - - 0 . 2 m (+1.2). This will h a v e some e f f e c t o n t h e s t a n d a r d d e v i a t i o n estimate. T h e square r o o t t r a n s f o r m a t i o n was f o u n d to be t h e m o s t s u i t a b l e a n d t h e c o r r e s p o n d i n g results are also s h o w n . However, this does n o t c h a n g e t h e r a n k order.

other factors are evident in this diagram, notably the rapid c o m m e n c e m e n t of overland flow soon after the c o m m e n c e m e n t of the storm and the very short lag response (6--18 min) between the peaks of rainfall and overland flow. Even more notable is the close relationship between the overland flow graph and the discharge hydrograph despite the difference in spatial location between the plot site and the gauging station. Previous work (Gilmour, 1975) has noted the rapid response of this catchment to rainfall events and the importance of the quickflow (or stormflow) portion of the hydrograph in contributing to the total runoff. It was calculated that 47% of the total r u n o f f occurred as quickflow. This compares with values of 15--20% for perennial streams in the temperate areas of the U.S.A. (Maki and Hafley, 1972). It is evident from Fig.2 that the quickflow component of the hydrograph mirrors very closely the overland flow graph thus highlighting the importance of overland flow as a major contributor to total runoff. Fig.2 is a typical example of the situation during that portion of the wet season when heavy rainfall predominates. It is interesting to note t h a t this storm produced an overland flow volume of 3242.9 1 compared with 105.6 1 from the 0.25 m depth. Therefore the ratio of contribution was 30.71 : 1 in favour of overland flow. It will also be noted that no reference has so far been made to the contributions from 0.5 and 1.00 m depths at this site. In









mmmm m l l









,Im,, ,im,lllllmm,,l . . . . I , , ,






STORM 0218hrs.

9.50 mm 1 5 . 2 5 .. 1 8 . 2 5 ,, 2 1 . 7 5 ,, 25.00 ,, 39.25 ,.




6mlns. 12 ,, 18 ,. 2 4 ,. 3 0 ,. 6 0 ,,



TO lO00hrs.



t!II.,,,; ,..,, .,..,, ,,,,,,, ,,,,.,.,..:

iI]l, I!ll l

| 25B LI~'R ES

, I,I l,,,,l,,l]J[,,, l , ,I,

S U B S U R F A C E FLOW ( 0 2 5 m )







EASTERN STANDARDTIME. L MARCH1976 Fig. 2. The c o n t i n u o u s r e c o r d for rainfall, o v e r l a n d flow, s u b s u r f a c e flow (0.25 m ) a n d discharge for a s a m p l e s t o r m at site 2.










< T



10 J




m ac'~5 0


~ ,00~

-~'" 1

, ~ 2°°1




"-I, 5 -






fact the 0.5-m bucket tipped only once during this storm yielding 2.8 I. There was no response from the 1-m trough. Sites la and l b cannot be compared on a continual recording basis with site 2 because of instrumentation problems. However, Table II showing the volume of flow per week from each depth indicates that substantial overland flow occurs at all sites. Subsurface flow also makes a major contribution to the total volume of lateral flow, particularly at sites la and I b. However, the magnitude o f overland flow is surprisingly large (considering the previously accepted explanation of drainage processes through forest soils) and it is this facet of flow which is the major subject of investigation in this paper. Certainly our visual observations during rainstorms do n o t indicate t h a t stemflow is the main source of overland flow as was suggested by R u x t o n (1967) and Douglas (1973). The nature of the uneven forest canopy permits a very high percentage of the rainfall to attain the forest floor as throughfall especially during heavy storms. Perhaps the most significant fact is that at none of the sites is there a very shallow permanent water table which is able to emerge at the surface to create saturation overland flow on the basis of the variable source area concept. All wells and piezometers were predominantly dry at site la except during rainstorms, and this also applied to the piezometers at site 2. Contact was made with the permanent saturation zone at site 2 but as Fig.3 shows for well 2/2, the water table did not come within 2.6 m of the surface and for the most part was below 3.5 m. This well was situated about 2 m upslope from the troughs. The m a x i m u m rise recorded by a second well, located further upslope, was 3.8 m below the surface but remaining records for the wet season showed t h a t the water level stayed below 5 m. Thus it is evident t h a t an alternative process must be considered. The principal differences between this environment and that in humid temperate areas are the characteristics of rainfall intensity. For example, in this envir o n m e n t m a x i m u m 6-min intensities for each week were c o m m o n l y between 6 and 10 mm during the 1976 wet season. If the soil ks properties are considered it is evident that the top 20 cm provide little obstruction to vertical percolation for most rainfall events. However, at lower levels a mean ks value of 0.318 m / d a y can also be interpreted as 13.25 m m / h or 1.325 mm per 6 min. A similar ks value in humid temperate areas, where the rainfall intensities are c o m m o n l y less than 13.25 m m / h , could be expected to encourage deeper vertical percolation. When occasional rainfall intensities exceeding this value occur for short periods, lateral subsurface flow could be expected. But Fig.2 demonstrates t h a t the majority of 6-min rainfall values exceeds 1.325 mm. Thus it is postulated that a temporary perched water table occurs soon after the onset of rain, and subsurface flow rapidly develops in the top 20 cm recharged by "pipe f l o w " infiltration due to innumerable macropores. The perched water table quickly emerges at the soil surface and subsequent rainfall moves either over the surface as overland flow or recharges the perched water table through losses by vertical percolation and subsurface


T A B L E II Weekly t i p p i n g - b u c k e t v o l u m e s (1) p e r w e e k at t h e t h r e e s t u d y p l o t s Date

Throughfall (mm)

Site l a surface

21.1.76 28.1.76 4.2.76 11.2.76 18.2.76 25.2.76 3.3.76 10.3.76 19.3.76

24.3.76 1.4.76 7.4.76 14.4.76 21.4.76 27.4.76 5.5.76 12.5.76 19.5.76 26.5.76 2.6.76 9.6.76 16.6.76

81.54 (la + lb) 70.3 (2) 56.58 156.34 27.58 98.99 65.86 52.37 153.73 (la + lb) 1 5 4 . 8 1 (2) 59.84 (la + lb) 5 8 . 7 5 (2) 96.89 145.03 244.35 113.13 89.93 17.57 60.93 58.75 65.13 99.71 93.33 18.88 10.18 71.59 15.55 28.3

(la + lb) (2)

(la + lb) (2) (la + lb) (2)


0.25 m ,

0.5 m 43.55


380.14 826.54 184.84 767.25 641.7 571.95 2197.13

520.65 1076.4 178.43 491.4 327.6 269.1 1325.02



488.25 850.95 7149.38 1541.48

976.95 1427.4 1064.7 239.85

331.65 , 402 30.15

16.13 , 548.25 16.13

0 41.85 ,

2.93 137.48 277.88

0 3.35 10.05

0 3.23 12.9





46.8 0 462.15 49.73 81.9

0 0 10.05 0 0

0 0 19.35 0 0

94.16 0 428.96 6.98 13.95

107.2 549.4 6.7 63.65 60.3 53.6 475.7

1.0 m


83.85 , 9.68 45.15 22.58 25.8 238.65 35.48

(a) Dates q u o t e d in m o s t cases r e p r e s e n t 7-day periods. (b) T h r o u g h f a l l c a l c u l a t e d f r o m t o t a l rainfall for a p p r o p r i a t e p e r i o d b y e q u a t i o n y = -- 0.7 + 0 . 7 2 5 x w h e r e y = t h r o u g h f a l l ( m m ) , x = rainfall ( m m ) ( G i l m o u r , 1975). T w t h r o u g h f a l l values are s h o w n f o r t h e a p p r o p r i a t e sites w h e n readings were t a k e n at d i f f e r e n t t i m e s d u r i n g rainfall. (c) Asterisks d e n o t e u n r e l i a b l e records.

flow. The mechanism has strong parallels with that p u t forward by Dunne and Black (1970a, b) who described it as saturation overland flow. The major difference is that in this study a temporary perched water table over virtually the whole catchment is the important factor in initiating overland flow whereas Dunne and Black postulate the rise of a permanent saturated zone adjacent to the stream channels. Thus the main implication is t h a t overland flow can occur over widespread areas of the catchment once the top 20 cm store has been filled. Furthermore, the temporary high drainage den-


Site I b

Site 2


0.25 m

0.5 m




245 412.5 35 255 195 240 1340

167.3 415.26 32.86 125.48 71.7 56.76 236

0 9.5 0 6.34 6.34 3.17 19




517.5 1337.5 2295 585

176.26 265.89 313.69 44.81

22.17 34.84 41.18 6.34

2.5 67.5 145

2.99 29.88 47.8



3.17 9.5




22.5 0 142.5 7.5 7.5

17.93 0 203.15 14.94 32.86

0 0 12.67 0 6.34

1.0 m 0

surface 861.3

0.25 m 112

0.5 m 0

1.0 rn 0 0 0 0 0 0 0 5.18

0 5.39 0 0 0 0 2.69

1650.83 • • 1513.8 991.8 809.1 4600.13

195.2 , , 185.6 160 60.8 201.6

5.58 13.94 2.79 0 0 0 8.36






2205.45 3830.18 , 1631.25

, 723.2 , 115.2

2.79 19.51 , 2.79

0 , , 0

0 0

61.99 512.21 694.91

12.8 76.8 67.2


2381.63 270.79 3.26 499.16 127.24 270.79

0 , 21.54 2.69

2.69 0 0 0 0



0 0

0 0




28.8 0 83.2 12.8 22.4

5.58 0 11.15 2.79 0

0 0 0 0 0

sity created b y the a d d i t i o n of n u m e r o u s e p h e m e r a l streams d u r i n g rains t o r m s e n a b l e s a s u b s t a n t i a l a m o u n t o f t h i s o v e r l a n d f l o w , a n d t o a lesser extent subsurface flow, to be tapped by the main stream. This accounts for the very high magnitude of quickflow response. To investigate this hypothesis further, weekly surface tipping-bucket v o l u m e s w e r e c o r r e l a t e d u s i n g S p e a r m a n ' s r a n k c o r r e l a t i o n c o e f f i c i e n t , rs (Siegel, 1 9 5 6 ) , w i t h 1 4 w e e k l y r a i n f a l l v a r i a b l e s w i t h s p e c i a l e m p h a s i s o n i n t e n s i t y . T h e y w e r e t h e w e e k l y m a x i m u m i n t e n s i t i e s f o r 0.1, 0 . 2 , 0 . 3 , 0 . 4 ,



oolEH I ~oo~ |



SITE 2. WELL No. 2


,,d3 LU ,-J ¢t" LU I---

LL 04T I--£1UJ £3
















Fig. 3. T h e well h y d r o g r a p h for well No. 2 l o c a t e d a b o u t 2 m u p s l o p e from the troughs at site 2.


0.5, 1, 2, 3, 4, 5, 6 and 12 h. In addition, the total weekly precipitation and the a m o u n t for the maximum storm were incorporated into the analysis after using a correction for interception (Gilmour, 1975). Spearman's rs method was used because its distribution-free property was considered more suitable when dealing with small samples. The period for this particular investigation was 14.1.76--16.6.76. The tensiometers indicated that the soilmoisture conditions were similar throughout this time interval with the suction in most cases lying between 0 and 100 cm. In addition, small positive pressures up to +40 cm were also often noted. Beyond mid-June a prolonged drying phase c o m m e n c e d creating a different hydrological situation so that any later rainfall events are n o t included in this analysis. Tables III and IV show the respective Spearman's rank correlation coefficients and their levels of significance (Siegel, 1956). It will be noted that the analysis has been divided into " s u m m e r " and "winter" seasons. The basis of the separation were the histograms for the short-term maximum rainfall intensities. Fig.4 shows an example of the separation of the weekly maximum 6-min intensities into t w o populations. Other short-term intensity histograms showed the same distinct bimodal nature. It was remarkable that in T A B L E III Weekly surface t i p p i n g - b u c k e t v o l u m e s vs. rainfall variables c o r r e l a t i o n c o e f f i c i e n t s (r s) for s u m m e r p e r i o d Weekly surface tipping-bucket v o l u m e s versus

Site l a n = 13

Site l b n = 13

Site 2 n = 9


significance level, p


significance level, p


significance level, p

0.05 0.05 0.05 0.05 0.05 n o t significant 0.05 0.05 0.05 0.05 0.05 n o t significant

0.64 0.72 0.82 0.76 0.71 0.51

0.05 0.01 0.01 0.01 0.01 0.05

0.43 0.85 0.90 0.82 0.80 0.85

n o t significant 0.01 0.01 0.01 0.01 0.01

0.67 0.73 0.76 0.74 0.80 0.63

0.01 0.01 0.01 0.01 0.01 0.05

0.85 0.90 0.93 0.87 0.92 0.8]

0.01 0.01 0.01 0.01 0.01 0.0]

0.01 0.05

0.82 0.71

0.01 0.01

0.73 0.88

0.05 0.01

Max. Max. Max. Max. Max. Max.

6 min 12 m i n 18 m i n 24 m i n 30 m i n 1h

0.62 0.62 0.63 0.58 0.54 0.42

Max. Max. Max. Max. Max. Max.

2 h 3 h 4 h 5 h 6 h 12 h

0.57 0.60 0.61 0.49 0.56 0.44

Total w e e k l y precipitation Max. s t o r m

0.85 0.54

The significance levels s t a t e d are for a one-tailed t e s t as physically the area o f i n t e r e s t is in in t h e positive s p e c t r u m , r/> O.


T A B L E IV Weekly surface t i p p i n g - b u c k e t v o l u m e s versus rainfall variables c o r r e l a t i o n c o e f f i c i e n t s (r s) for w i n t e r p e r i o d Weekly surface tipping-bucket v o l u m e s versus

Max. Max. Max. Max. Max. Max. Max. Max. Max. Max. Max. Max.

Site l a n = 8

6 rain 12 m i n 18 rain 24 rain 30 m i n 1h 2 h 3 h 4 h 5 h 6 h 12 h

Total w e e k l y precipitation Max. s t o r m

Site l b n - 9

Site 2 n = 10


significant level, p


significance level, p


significance level, p

0.20 0.46 0.40 0.99 0.97 0.92 0.90 0.95 0.95 0.98 0.98 0.65

not significant n o t significant n o t significant 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.05

0.43 0.66 0.58 0.96 0.92 0.94 0.90 0.94 0.94 0.95 0.97 0.60

not significant 0.05 n o t significant 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.05

0.46 0.71 0.67 0.92 0.85 0.83 0.80 0.91 0.87 0.88 0.90 0.76

n o t significant 0.05 0.05 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01

0.89 0.97

0.01 0.01

0.93 0.98

0.01 0.01

0.95 0.91

0.01 0.01

PERCENTAGE 40...........,.....,....


iiiiiiiiiiiiiiiiiiiiiiiiii!ii ~ ::::::::::::::::::::::::::::: :::::::::::::::::::::::::::::




iiiiiiiiiiiii;iiiiiiiii;iiiii ~ iiiiii!iiiiiiiiiiiiiiii;iii:: iiiiii?i;i?iiiiii?ilii?ii • .-z.':.'.','.'.'.'::.-





2 3.58

i11ii !i!i i!iiiiiii!2!iiii:l 212



3 7-58

I 9"58


I 11.6

Fig. 4. F r e q u e n c y diagram o f m a x i m u m 6-rain. rainfall i n t e n s i t y / w e e k f r o m 14 J a n u a r y - 16 J u n e 1976.


terms of time the first group incorporated the lower intensity "winter" rainfall dating from mid-April to early June. Conversely, the second group of higher intensity " s u m m e r " rainfall dated from the beginning of records in January to mid-April. It will be noted that the sample size for each site varies due to some loss in records. Referring to Table II the most outstanding feature is that the majority of Spearman's r s are significant for the summer period. The physical interpretation is probably that the short-term maximum intensities make a large contribution to the total weekly overland flow recorded despite their extremely short duration. Certainly, these intensities are far in excess of the ks values below 10 cm and therefore capable of rapidly developing saturation overland flow. Furthermore, if they occur once the perched water table has emerged at the surface, which is c o m m o n l y the case, then the catchement will be acting as an "impermeable" surface resulting in a very large proportion of this large short-duration input to move over the surface. The longer-term intensities, i.e. max. 2 h--max. 6 h, also include the lower 6-min intensities which as seen from Fig.2 are high enough to iniate overland flow. In addition, they will make a substantial contribution to saturation overland flow once the perched water table has emerged at the surface. The latter statement can also be applied to the significant rs values corresponding to the maximum storm and the total weekly precipitation. The small sample sizes in conjunction with random error effects make it difficult to evaluate major differences within and between the three study plots. In view of these statistical considerations any comparisons that are made must be regarded with caution. Nevertheless site l a consistently shows lower rs values which, when significant, are at a lower level, than sites l b and 2. Table II showed that site la had the most subsurface flow. It is also the steepest and differs certainly from site 2 in that borings for piezometer installations detected a scattering of residual rocks in the clay just below the surface. No borings were made on site I b b u t the trough exposures suggested that it was intermediate in composition between the other two sites. The shallow-well pump-in tests showed that any above-average ks values for the 20--100 cm zone were generally associated with rocky profiles. Consequently, this will not only lead to a much higher proportion of throughfall being transmitted either as subsurface flow or as deep vertical percolation b u t also will require much higher short-term rainfall intensities to cause the perched water table to emerge at the surface in such areas. In other words the effects of rainfall intensity do not always erase the influence of the lithological factor which probably contributes to the lower r s values. This implies that overland flow is possibly only generated during higher rainfall intensity events which can exceed the rate of removal of stored water b y subsurface flow. It is n o t e w o r t h y that the steepest slopes do n o t necessarily give the highest overland flow amounts, c.f. Thomas (1973). The winter situation presents interesting comparisons with the previous season (Table IV). The distribution of significant r s values for all sites shows


a much greater emphasis on the longer-duration maximum intensities. A notable feature is the sudden rise in the rs between max. 18 min and max. 24 min. The initial three rainfall variables, i.e. max. 6 min to max. 18 min are either nonsignificant or only significant at the lower level. A further point of note is the highly significant rs values for total weekly precipitation and m a x i m u m storm. It is not surprising that the results show less emphasis on the ability of the short~term m a x i m u m intensities to generate substantial overland flow because at this time of the year intensities are much lower than during summer. For example, Fig. 4 shows max. 6-min values ranged from only 1.58 to 4 mm. The characteristics of the winter rainfall in 1976 being of lower intensity but longer duration probably accounts for the high rs values between max. 24 min and max. 6 h. Nevertheless once the perched water table has emerged at the surface the higher-intensity parts of the storm, focusing on max. 24 min, seem to make a significant contribution to overland flow volumes. The longer-duration maximum intensities, as in the summer situation, will incorporate some of the lower short-term intensities which will also be capable of generating overland flow. The fact that max. 12-h has a much lower rs at all sites can be attributed to some weeks n o t including a storm of that duration. During the winter storms the rs values for max. 12 min--max. 18 min were lower at site l a than at the other two sites. It is possible that the short-term intensities cannot make much impression on site l a because of the mosaic pattern of above average ks created by subsurface rocks. Thus subsurface as 'well as overland flow is a major flow path. On the other hand, in the absence of such a factor (particularly at site 2) a significant overland flow contribution can be made. Geological differences seem to be less important from max. 24 min onwards possibly because the long duration of the storm is capable of filling up even the larger voids thus enabling the perched water table to emerge at the surface. This statement could also explain why site l a shows such a high rs value for the m a x i m u m winter storm compared with the m a x i m u m summer storm. The results from this study are in marked contrast with findings in humid temperate areas. In such areas, if overland flow is to occur away from the shallow water table areas, it is likely to be associated with short periods when very high rainfall intensities are capable of exceeding the infiltration rate. Selby (1973) statistically demonstrated this fact in connection with ungrazed grass and scrub on pumice lithology in New Zealand. He noted that the most significant correlation coefficients corresponded with short-term m a x i m u m weekly rainfall variables. This contrasts with the high rs values for both longas well as short~term rainfall intensities in this work. Other statistical analyses by Hewlett et al. (1977), based on long-term records connected with a forested watershed in the southern Appalachians, U.S.A., indicated that m a x i m u m rainfall intensity (hourly and minutely) were of little practical value in predicting storm flow volume or peak dis'charge. Thus they concluded that rainfall intensity was not a d o m i n a n t variable in stormflow generation in that environment.



This work has indicated that in this particular tropical rainforest environment the variable source area concept, associated with humid temperate areas, is not applicable due to the characteristics of rainfall intensity. Spearman's rank correlation analysis indicated that this applied to both seasons, despite some detailed differences, providing the assumption of a linear relation is correct. Thus the rainfall intensity is sufficient to exceed the average ks for depths below 20 cm and therefore create a widespread perched water table, and corresponding subsurface flow, within the top 20 cm. Once this upper store has been filled, its emergence at the surface develops an "impermeable" surface and therefore leads to widespread saturation overland flow which can be tapped by a temporary dense drainage network. This p h e n o m e n o n is thought to continue for the remainder of the storm. The only d o u b t raised by Spearman's correlation analysis was during the summer period when storms in 1976 were generally of shorter duration than normal and where a slope is underlain by residual rocks in the deeply weathered clays, e.g. site la. It is hoped to examine this problem further with more sophisticated statistical analysis once continuous records become available. The significance of subsurface flow contribution to the storm hydrograph, originating from the perched water table, becomes greater in the presence of residual rock. However, this does not preclude large volumes of overland flow occurring even on the steeper, well drained slopes in contrast with the situation in humid temperate areas. ACKNOWLEDGEMENTS

The Australian Water Resources Council, Queensland Forestry Department and James Cook University of North Queensland are thanked for their continued financial support of this work. Mr. H.O. McColl is also thanked for his assistance in data collection under at times very trying conditions.

REFERENCES Boersma, L., 1965. Field measurement of hydraulic conductivity above a water table. In: C.A. Black (Editor), Methods of Soil Analysis, Part 1, Agronomy No. 9, Am. Soc. Agron., Madison, Wisc., pp. 234--252. Douglas, I., 1973. Rate of denudation in selected small catchments in eastern Australia. University of Hull, Occas. Pap. Geogr., No. 21, 127 pp. (quotation in text, p. 35). Dunne, T. and Black, R.D., 1970a. An experimental investigation of runoff production in permeable soils. Water Resour. Res., 6: 179--191. Dunne, T. and Black, R.D., 1970b. Partial area contributions to storm runoff in a small New England watershed. Water Resour. Res., 6: 1269--1311. Dunne, T., Moore, T.R. and Taylor, C.H., 1975. Recognition and prediction of runoffproducing zones in humid regions. Hydrol. Sci. Bull., 20: 305--327. Ernst, L.F., 1950. A new formula for the calculation of the permeability factor with the auger hole method. T.N.O., Groningen (reported by W.F.J. van Beers, 1958).


Freeze, R.A., 1972b. Role of subsurface flow in generating surface runoff, 2. Upstream source areas. Water Resour. Res., 8: 1272--1283. Gilmour, D.A., 1975. Catchment water balance studies on the wet tropical coast of north Queensland. Ph.D. Thesis, James Cook University of North Queensland, Townsville, Qld. (unpublished). Hewlett, J.D., 1961. Watershed management. In: U.S. For. Serv., Southeast. For. Exp. Stn. Rep., pp. 61--66. Hewlett, J.D. and Hibbert, A.R., 1967. Factors affecting the response of small watersheds to precipitation in humid areas. In: W.E. Sopper and H.W. Lull (Editors), International Symposium on Forest Hydrology, Pergamon Press, Oxford, pp. 275--290. Hewlett, J.D. and Troendle, C.A., 1975. Non-point and diffused water sources: a variable source area problem. In: Watershed Management, Symp., Comm. Watershed Manage., Irrig. Drain. Div. Am. Soc. Civ. Eng., Logan, Utah, pp. 21--46. Hewlett, J.D., Fortson, J. and Cunningham, G., 1977. Effect of rainfall intensity on stormflow and peak discharge from forest land. Water Resour. Res., 13: 259--266. Horton, R.E., 1933. The role of infiltration in the hydrological cycle. Trans. Am. Geophys. Union, 14: 446--460. Horton, R.E., 1945. Erosional development of streams and their drainage basins: hydrological approach to quantitative morphology. Bull. Geol. Soc. Am., 56: 275--370. Kirkham, D. and Van Bavel, C.H.M., 1948. Theory of seepage into auger holes. Soil Sci. Soc. Am., Proc., 13: 75--89. Luthin, J.N. and Kirkham, D., 1949. A piezometer method for measuring permeability of soil in sites below a water table. Soil Sci., 68: 349--358. Maki, T.E. and Harley, W.L., 1972. Effects of land use on municipal watersheds. Water Resour. Res. Inst. Univ., North Carolina Report No. 71. Northeote, K.H., 1971. A Factual Key for the Recognition of Australian Soils. Rellim Technical Publications, Glenside, S.A., 3rd ed., 122 pp. Ruxton, B.P., 1967. Slopewash under Mature Primary Rainforest in Northern Papua. In: J.N. Jennings and J.A. MabbutL (Editors), Landform Studies from Australia and New Guinea. Australian National University Press, Canberra, A.C.T., pp. 85---94 (quotation in text, p. 91). Selby, M.J., 1973. An investigation into causes of runoff from a catchment of pumice lithology in New Zealand. Hydrol. Sci. Bull., 18: 255--280. Siegel, S., 1956. Nonparametric Statistics for the Behavioural Sciences. McGraw-Hill, Tokyo, 312 pp. Talsma, T., 1969. In situ measurement of sorptivity. Austr. J. Soil Res., 7: 269--276. Talsma, T., 1974. Infiltration in field soils. Trans. 10th Int. Congr. on Soil Science, Moscow, 1: 68--74. Thomas, M., 1973. Landforms in Equatorial Forest Areas. In: D. Brunsden and J.C. Doornkamp (Editors), The Unquiet Landscape, David and Charles, Newton Abbott, pp. 141--146 (quotation in text, p. 145). Van Beers, W.F.J., 1958. The auger hole method, a field measurement of the hydraulic conductivity of soil below a water table. Int. Inst. Land Reclam. Improv., Wageningen, Bull. 1. Ward, R.C., 1975. Principles of Hydrology. McGraw-Hill, London, 2nd ed., 367 pp. (quotation in text, p. 256). Webb, L.J., 1969. A physiognomic classification of Australian rain forests. J. Ecol., 47: 551--570. Whipkey, R.Z., 1965. Subsurface storm flow from forested slopes. Int. Assoc. Sci. Hydrol., bull., 10: 74--85. Whipkey, R.Z., 1969. Storn runoff from forested catchments by subsurface routes. In: Floods and Their Computation, Studies and Reports in Hydrology, Leningrad, UNESCO-IASH--WMO, 3: 773--779.