In addition, controlled experiments have shown that the conversion of forests to grasslands usually increases delayed flow (dry-weather flow) (Bosch and Hewlett, 1982; Cassells et al, 1987). During extended dry periods when the shallow rooted, herbaceous cover is under stress; the deeper rooted trees continue to access soil moisture or even groundwater. Thus removal of trees prevents additional water losses at these times and allows unsaturated/saturated flow to continue transmission to streams. In turn, groundwater levels rise in response to the decreased evapotranspiration (Bosh and Hewlett, 1982; Peck and Williamson, 1987). Another factor which contributes to these changes in water yield revolves around wet/dry canopy evaporation differences between forests and low crops. Where soil water is non-limiting, McNaughton and Jarvis (1983) noted that whilst the canopy resistance (or bulk surface resistance, representing stomatal control, rs; Shuttleworth, 1988) of forests was higher than grasses, therefore causing dry canopy evaporation over forests to be lower; the rate of evaporation of intercepted water over forests (wet canopy evaporation) was higher than the herbaceous cover. In climates where the canopy is wet for a significant proportion of time, then the combined wet canopy/dry canopy evaporation from forests can approach or even exceed the potential evapotranspiration rates of short crops (McNaughton and Jarvis, 1983; Morton, 1984). This particularly applies to locations near the edge of continents or on islands where it appears the energy attracted from the ocean is capable of enhancing the evaporation of intercepted water; and net time-average water loss exceeds the potential rate (Calder, 1985; Morton, 1985; Shuttleworth, 1989). The fewer studies undertaken in continental interiors indicate the the time-average total evaporation reverts to almost the potential rate (Shuttleworth, 1989). In the case of central Amazonas, evaporation losses were within 5 percent of the potential rate (time-averaged over 25 months), but as Shuttleworth (1988, p. 342) noted 'the time-average behaviour is therefore a balance between less than potential rate, with a dry canopy, and greater rate with a wet canopy. The near equality between average evaporation and potential evaporation in central Amazonas is ... in part fortuitous...' and continues '... it is, for instance, very probable that Amazonas regions with higher rainfall, where the forests spend a greater proportion of the time wet, will tend to have average evaporation rates greater than the potential rate'. The conflict then in water yield changes in the humid tropics revolves around the delicate balance between the complexities of the evapotranspiration process and changes in soil hydraulic properties connected with infiltration, a relationship also emphasized by Bruijnzeel (1989) and Chang (1989). If the net gains in groundwater accretion from lower evapotranspiration are more than offset by much-reduced percolation, and therefore reducing groundwater accession, then the depth of the water table may decline causing springs and stream flow discharges to reduce in dry seasons. In spite of these assertions, there is a dearth of supporting evidence concerning changes in hillslope storm runoff and soil hydraulic properties. There are additional factors which confound these widely held ideas, viz scale of investigation, controlled or uncontrolled drainage basin comparisons and the prospect of drainage basin leakages, all of which are closely interrelated (Bruijnzeel, 1989; Shuttleworth, 1988). As Morton (1984, p. 385) noted "... in a large basin the difference between the groundwater and topographic divide is negligible and the river has eroded into channel deep enough to intercept all significant groundwater flow systems. In a small basin this is not necessarily so". The underlying geology and different scales of investigation are particularly appropriate to the Amazon Basin where a review of the deforestation controversy showed that annual runoff coefficients ranged from 0.19 to 0.57 (Salati, 1987, p. 288) from different water balance studies. Shuttleworth (1988, p. 322) commented "... water balance experiments are prone to the uncertainty that ungauged, subterranean leakage forms a significant part of the total movement, a problem of considerable relevance in the free-draining soils of this region. Underestimation of drainage easily leads to a (proportionately enhanced) overestimation of evaporation...". Furthermore possible misleading conclusions emanate from uncontrolled experiments (no prior calibration between pairs of drainage before disturbance) as noted by Bruijnzeel (1989) in the case of work reported from central Java (Hardjono, 1980). The development of process-hydrology studies however, are not devoid of their own problems. In a recent commentary on the state of physically-based models, Beven (1987, 1989) highlighted the critical issues facing process hydrology and the inadequacies of such models. These included the problems of applying small scale physics of homogeneous systems to large scale drainage basins, eg. infiltration equations derived from laboratory experiments. As Beven (1989, p. 161) noted "... it is not possible to use small scale physics equations (such as the application of infiltration theory in runoff generation studies) at the grid scale (eg. 250 x 250 m in the case of the Système Hydrologique Européen, SHE model, Abbott, 1986a, b: Bathurst, 1986a, b); and that we should be developing more complex equations that account for the effects of such heterogeneity". These concerns also applied to temporal heterogeneity at a point within a drainage basin as well as spatial heterogeneity. Amongst several developments required, Beven (1989, p. 170) outlined the need for closer correspondence between model equations and field processes at different scales. It was in recognition of the preceding issues that major programmes initiated between the Department of Geography of James Cook University and the Queensland Department of Forestry in 1974 in the tropical rainforest of north-east Queensland, and in 1980 with CSIRO Division of Soils, Townsville, in an open eucalypt woodland within the semi-arid area of central-north Queensland. Both programmes were initiated to seek baseline information on the mechanisms of storm runoff for the development of rigorous land management guidelines in connection with disturbance by logging and clearing of tropical rainforest, and the future introduction of improved pasture species and more intensive grazing in the semi-arid area. These research programmes are still continuing and this paper will briefly review the published literature to date in each study and outline possible future developments. Special emphasis will be placed on the interaction of synoptic climatology with the runoff process in both studies. As this is a review, only a brief description of the physical background, and experimental and analytical methods for each study will be given. Further details are obtainable from the literature cited.
THE WET TROPICAL COAST
The problem
THE SEMI-ARID OPEN EUCALYPT WOODLAND
Details of the experimental methods have been discussed elsewhere (Bonell and Williams, 1986b, 1987; Williams and Bonell, 1987, 1988). The principal focus was a cascade system of five troughs, 10 m long and offset at 25 m intervals, which monitored a 100 m section of slope (Fig. 6) located 2.5 km from a subdued topographic divide. This design is based on the assumption that the overland flow observed on the upslope trough is an estimate of the runon to a 250 m2 plot and that the runoff from the plot is monitored by the downslope trough. Unlike bounded plots, the continuous systems allows the outputs and inputs of both water and sediment to be estimated without interference to the natural processes.
The runoff-runon (mm) for the plots between the troughs for a given period is given by X = Qd = Qu where Qu is the upslope and Qd is the downslope overland flow. Infiltration, i (mm) for a given period is calculated by i = R - X - dS - dV where R is the rainfall in the given period, and change in surface detention store and depth in surface water is given by dS while dV represents the change in water interception by vegetation.
It follows that X is positive for runoff and negative for runon. Where runon prevails, this adds to rainfall and contributes to infiltration. For the time intervals of 1 min used in the analysis it is assumed that dS and dV are zero. Consequently the plots behave as a constant head infiltration device in which ponded infiltration theory is applied to determine cumulative infiltration using rainfall and overland flow only, [dS = 0], from the commencement of overland flow.
The time to commencement of runoff, tr could be as short as 1 min when average rain intensities were high (equivalent hourly rate up to 84 mm hr-1) and inferred incipient ponding (Rubin, 1966) occurs even earlier. Rain intensities that generate Horton-type overland flow (Horton, 1945; Dunne, 1983) also showed a marked variability over time above 10 mm hr-1 due to the temporal variability of soil hydraulic properties (Bonell and Williams, 1986a, b). Such variability reflects in part the change in soil fabric from biological activity, raindrop compaction and dessication cracks of the surface crust.
The trend of runoff-runon is not consistent in the storms shown in Table 2. For most events Plots 2 and 3 are runoff areas, but the volumes generated remain small and are less than 19% of total rain. Plot 1 changes from a runoff to a runon area from 23.1.82 whilst Plot 4 is a consistent runon area, acting as a sink for excess runoff generated by the preceding plots. This internal redistribution causes only a small amount of overland flow ( 1 mm) to be exported downslope out of this 100 m transect, and it is less than 4% of total rain.
The infiltration parameter sorptivity, S and transmission, A of Philip (1969) and in situ field saturated hydraulic conductivity K* were determined over time from infiltration rings (0,3 m dia, 0.15 m deep driven into the soil 0.10 m), using analytical techniques described by Bonell and Williams (1986a). The same hydraulic properties at the scale of the plot were determined from cumulative infiltration as a function of time (Bonell and Williams, 1986b; Williams and Bonell, 1987, 1988), using the simple analysis described above.
The range and magnitude of the sorptivity parameter (Philip, 1969) in the infiltration rings were small (grand mean, 0.097 mm s-1/2 Bonell and Williams, 1986a) and were consistent with the low storage capacity (only 8-12% by volume of water held between matric potentials - 10 kpa and - 1500 kpa). The conditions of 'profile at infinity' (Philip, 1969) were quickly attained under ponded infiltration such that the transmission parameter, A in Philip's truncated equation, approximated K* (Bonell and Williams, 1986a) and cumulative infiltration (I) = K* t, where t is time from ponding.
A similar behaviour was observed for the runoff plots during ponding under natural rainfall (Bonell and Williams, 1986b; Williams and Bonell, 1987, 1988). Under a mature ground cover before fire, the same order of magnitude of K* was determined for the infiltration rings in non-vegetated surfaces (log mean K* = 65.8 mm hr-1) and for the runoff plots (log mean K* = 42.1 mm hr-1). This contrasts with K* determined from rings enclosing spinifex tussocks (289.6 mm hr-1). These findings are supported by partial correlation analysis (Bonell and Williams, 1986b), which showed that 1 minute rainfalls with an equivalent hourly intensity of 60 mm hr-1 or greater were highly correlated with measured overland flow where the largest volumes are produced.
Away from the surface, the subsoil K* (Talsma and Hallam, 1980) is higher (log mean K* = 100.4 mm hr-1) than the non-vegetated soil, but less than the surface soils associated with spinifex tussocks. The effects of raindrop impact are then evident at the surface.
Williams and Bonell (1988) later analyzed over thirty storms ranging from 5 to 55 mm in magnitude and to date all storms provided remarkably linear cumulative infiltration-time relations, including records collected over a two year period following fire. The analysis assumed that once the surface detention store was filled (approximately 3-4 mm), the flow depth remained constant with time and the infiltration process could be described by I = K* t.
The spatial variability of the 0.7 m2 ring estimates of K* was approximately 4-10 times greater than that for the large 250 m2 plots. This is consistent with Sisson and Wieringa (1981) who noted a substantial reduction in spatial variability upon increasing the diameter of infiltration rings. With respect to temporal variability of K*, the plot estimates of K* were again much less variable in time than those of the rings, although the differences were less than for spatial variability. In addition rings enclosing bare soil appeared to undergo less variation with time than the part-vegetated rings.
A comparison of the combined spatial and temporal variability of K* for ring and plot estimates is set out in Table 3. For K* the plots provide a 3-4 fold reduction in variability over the infiltration rings located in bare surfaces and a 10-12 fold reduction compared to rings located in tussock vegetation. More important, the mean values of K* for 4 rings located in bare soil between tussocks were approximately twice that of the plots whilst those rings located in soil associated with grass tussock was some six or seven times larger than that estimated from the plot. Apart from differences in the scale of measurement, the fact the bare ring estimates are of the same order of magnitude to those from the plots is encouraging particularly when it is recognized that the rings and plot measurements were not made at exactly the same time.
This leads to the concept of the 'representative elementary volume' (REV) as applied to soil physical measurements (Youngs, 1983; Baveye and Sposito, 1984). To incorporate the macroscale variability, the sampling scale must exceed the REV of the system. This leads to the additional concept 'repetitive unit' which is intended at regarding the behaviour of an inhomogeneous material equivalent homogeneous material if the length scale of observation is much larger than the characteristic length of the repetitive unit (Fig. 7). From mapping of the understory, it would appear that the repetitive unit has characteristic lengths of approximately 1 to 2 m. It follows that the plot or infiltration ring dimensions would need to be at least this size before a reduction of spatial variability would be observed and it indicates why the spatial variability of the 250 m2 plots is so small.
The problem of 'scaling-up' measurements made by rings to a plot level, disregarding a larger pixel scale as discussed by Beven (1989), are demonstrated by Williams and Bonell (1988). Whilst Youngs (1983) noted that there was no general solution for recombining hydraulic properties, Williams and Bonell (1988) weighed the K* values for each of the strata according to their relative area as follows. The area of soil associated with tussock grasses is 6 percent, the area associated with termite mounds is 1 percent, with the remaining 93 percent consisting of bare soil. Using the values of these strata and a value of 0.128 m d-1 for the termite mounds (Bonell and Williams, 1986b), results in an arithmetic mean K* for the soil system to be 2.82 m d-1. This figure is still considerably greater than the mean for the plots of 1.06 m d-1. A possible explanation is that the tussock areas are slightly higher in elevation than the bare soil surface and may not be ponded, in such areas infiltration would be determined by the rainfall rate only. The effect would be to reduce the estimates of K* for the plots biassed towards the bare soil. The fact that the arithmetic mean yields much larger values than the large plot estimates does serve to illustrate the intractable position if measurements are made at a scale smaller than the characteristics length of the repetitive unit.
The same work showed that the temporal variability in infiltration parameters that arise as a consequence of biological activity or the impact of rainfall cannot be removed by changing the scale of measurement. The temporal effect can however, be reduced at an increased scale of measurement through the influences of numerous compensating processes that influence soil hydraulic properties.
It is clear that the advantages of using bulk properties of the whole system are more useful than point soil physical properties (Youngs, 1983). The merit of measuring infiltration parameters from a hydrograph of a large unbounded plot or small catchment (eg: Burch et al, 1987) is that these soil parameters refer to a scale that appears to be representative of that landscape element.
It is apparent that this is a relatively simple environment for the testing of scale problems, physically-based infiltration theory and possible future application of physical process models.
HYDROLOGICAL IMPLICATIONS FOR LAND MANAGEMENT
The wet tropical coast the impact of logging and clearing of tropical rainforest on storm runoff hydrology erosion
It became apparent by the 1960s that unconstrained logging was causing spectacular gully erosion up to 12 m deep (Gilmour, 1977a), especially on logging or 'snig' tracks. This instigated two separate investigations within the Freshwater Creek basin near Cairns (Gilmour, 1971) and in the Babinda catchments (Gilmour, 1977b).
The Babinda work followed the classes approach of calibration of undisturbed, paired catchments (North and South Creek), followed by monitoring the impact of logging and later clearing of North Creek. Statistical details of the work were presented by Gilmour (1977) and further interpreted in Gilmour et al (1982), in the light of process studies concerning runoff generation. Particularly outstanding was no detectable change in quickflow volume, quickflow duration or time to peak after logging and clearing in North Creek. These characteristics imply that there was only minor changes in the storm runoff, in terms of process and source areas despite a significant increase in soil moisture content (Gilmour et al, 1982) and an order of magnitude decline in the surface (0-0.1 m depth) log mean K*. For example, Gilmour (1975) showed that when the soils were in their driest condition after clearing in 1973, South Creek required 291 mm of rain for the soils in the surface three metres to attain 'field capacity', whereas the treated North Creek catchment required only 94 mm for the soils to reach the same condition.
Determinations undertaken between 1984-1986 (over 10 years since logging and clearing followed by forest regeneration) showed that the surface 0-0.1 m depth K* (log mean = 184.0 mm hr-1, n = 34) was in the same order of magnitude as the corresponding K* for South Creek (log mean 842.5 mm h-1), but obviously the absolute value is a lot lower and more towards the log mean K* for the 0.1 to 0.2 m depth (60.0 mm hr-1, n = 60) in South Creek. In addition, there was no significant difference between the drainage basins at the depth intervals, 0.1-0.2 m and 0.2-0.5 m (*). It is not clear whether the significant decline in surface K* of North Creek is the result of either compaction from earlier logging and clearing coupled with a reduction in soil biological activity and raindrop compaction of the newly exposed soil; or erosion of the original top 0.1 m exposing part of the former 0.1-0.2 m layer in the undisturbed soil. Current work detecting Caesium-137 levels in the soil profile will assist in this interpretation. The fact that the subsoil K* for both North and South Creek, 0.1-0.2 m layer are comparable encourages the idea that compaction and reduction in biological activity maybe the more likely explanation. In this context it is interesting that the few determinations of K* for the surface 0-0.1 m layer in the remaining undisturbed area of the upper reaches of North Creek (log mean K* = 1144.8 mm hr-1, n = 10) are not much higher than the larger sample measurements in South Creek.
Despite these surface changes of K* in North Creek, it is evident that the subsoil K*, continue to act as the 'throttle' or impeding layer in the storm runoff process in a similar manner to that occurring in undisturbed South Creek. Consequently determination of just this one parameter helps considerable to interpret Gilmour's (1975) earlier water balance results.
Major changes however, were evident in the water quality of North Creek after the change in land use. Peak suspended sediment concentrations rose from 180 mg L-1 before logging to about 520 mg L-1 in the two years after logging.
Clearing brought a much more dramatic change in suspended sediment concentrations
in North Creek reaching values as high as 400 mg
L-1 despite dilution from peak discharges. From
the data on stream sediment levels in North Creek, Capelin and Prove (1983) estimated that
annual
(*) 0.1 - 0.2 log mean K*, North Creek, 57.3 mm
hr-1, n = 28; South Creek, 60.0 mm hr-1, n = 60 / 0.2 - 0.5 m
log mean K*, North Creek, 3.3 mm hr-1, n = 155; South Creek, 3.5 mm
hr-1, n = 219.
suspended levels rose from 4.8 tonnes ha-1 before disturbance to 10.9 tonnes
ha-1 after logging and 59.6 tonnes ha-1 after clearing. Bedload was not measured, but Douglas (1967) suggested this factor
may be as high as 50 percent of the suspended load which raises the total loss to about 90 tonnes
ha-1.
When the erodibility of various rainforest soil types is considered, the quoted erosion losses from the Babinda catchments are likely to be conservative if the same activity had been undertaken on other catchments of the same morphometry. Using the dispersion index of Middleton (1930), the most stable soils are those derived from basalt, basic metamorphic (Babinda catchments), colluvium and some of the granite members (Bonell et al, 1986). The remaining soils derived from granite and acid metamorphics were erodible below the surface layer where organic matter is not incorporated. Exposure of these subsoils by log haulage causes them to disperse easily from raindrop impact and concentrated overland flow especially on logging tracks. Thus overland flow can be a mixture of both saturation overland flow and Hortonian overland flow due to soil compaction.
Apparent from this Babinda study, is that the amount of vegetation and the conditions of the top 0.2 m of soil are relatively unimportant in terms of quickflow generation. But the same two factors are of great importance in controlling erosion. The frequency of widespread saturation overland flow in undisturbed rainforest due to the prevailing rainfall intensities and shallow impeding subsoil, ensures little change in the storm runoff hydrology even after varying degrees of soil compaction from logging and clearing.
The need for the development of watershed management controls was initially advances by the Freshwater Creek study. Gilmour (1971) presented evidence to show that the principal sources of sediment in the creek were from poorly located undrained roads and 'snig' tracks, and from earth and log filled crossings. In the light of these findings, several guidelines were put into operation and included:
1) Snigging and hauling through streams was prohibited.
2) The use of streamside buffer strips at least 20 m from the stream were adopted to reduce the supply of sediment from 'snig' tracks, roads and logging ramps. These strips also ensured stream bank stability from the high discharges resulting from monsoon storms.
3) Earth and log fill crossings were prohibited. The alternative was the construction of girder bridges.
Following the implementation of these guidelines, significant reductions in suspended load were evident. For example, the highest measured concentration was 188 mg L-1 after a 262 mm rainfall compared with c 780 mg L-1 after 66 mm storm before these initiatives (Gilmour, 1971).
Further refinements to Gilmour's management controls were subsequently developed in the light of understanding of regional hydrological processes developed in the Babinda Research Programme, and these have been routinely applied to all forest harvesting operations since 1981. The nature of these controls and the social and technical considerations were described in detail by Cassells et al (1984). Amongst the technical improvements were:
1) The allotment of specifications for buffer strip width based on watercourse width, soil type and catchment size.
2) Maximum spacing of cross drains along 'snig' tracks and roads for different slope angles and soil erodibilities to reduce concentration of overland flow.
3) Design and location of haulage roads, 'snig' tracks, stream crossings, landings and the general logging direction and methodology.
It is interesting to compare these findings with the controlled experimental study described by Nik (1987) and Law and Cheong (1987) referring to the Sungai Tekem Experimental Basin in Peninsular Malaysia. No soil hydraulic properties are reported, but Law and Cheong (1987) indicate the soils have high surface infiltration capacities. Following clearing, no significant change was observed in quickflow volumes, but peak specific discharges increased up to 65 percent more than that in the calibration period and time to peak decreased considerably. Following the complete establishment of cover crops, peak specific discharge reduced but still remained 30 percent higher than that in the calibration period. Significantly, suspended sediment loads of one of the partially deforested basins rose from 0.3 t ha-1 yr-1 to 4.1 t ha-1 yr-1 following initial clearing, before a decrease after establishment of a cover crop to 1.6 t ha-1yr-1. These losses are very low compared with Babinda and highlight the differences in rainfall characteristics, topography and soils between the two areas. For example, Law and Cheong report annual maximum rain intensities of 24 hr duration (range 64.5 to 149.5 mm) much lower then those experienced in Babinda (Table 1, Fig. 3).
Table 4 shows the aggregate fluxes in overland flow and sediment between field visits, rather than for specific storms (Table 2), before a severe fire passed through the experimental site. Aggregate erosion from rainflow transportation (Moss et al, 1979) along the 100 m segment of slope remains small (approx. 0.02 tonnes ha-1) considering the amount of overland flow being exchanged between plots and even deposition is indicated, as shown by the negative change, is indicated during period 6. Following the fire, equipment malfunction caused some loss of data, but that available showed no statistical significant difference (Bonell and Williams, 1987) in terms of sediment transfer. Following a combination of all records from the plots 1 to 4, a good relation was developed (Fig. 8) between runon-deposition and runoff-erosion and the nonsignificant regression intercept is particularly notable. From a conceptual understanding of the erosion process (eg: Kirkby and Morgan, 1980) such a result can be expected, but to our knowledge has not been demonstrated from field experiments using natural rainfall. The commonly used bounded runoff plots in hillslope hydrology prohibit this type of analysis.
Clearly on these low relief landscapes where the soil infiltration properties are in near-equilibrium with the prevailing rain intensities, the amounts of erosion are small in the undisturbed landscape. Any disturbance by way of soil compaction from cattle or tracks induced by vehicles, has the potential to reduce infiltration and increase overland flow, and in turn erosion judging by field evidence of degradation in these landscapes.
DISCUSSION
In both the humid and semi-arid environment overland flow is the major delivery component, but the process of generation is quite different in undisturbed conditions. Hortonian overland flow prevails in the semi-arid area, where the lowest K* occurs at the soil surface due to raindrop compaction. Perhaps the occurrence of this flow type is not surprising as it is associated with arid and semi-arid landscapes where vegetation densities are low (Dunne, 1983), and the Torrens Creek experiment provides a classic example. But the temporal variability of surface K* requires different rain intensities and in turn different times to runoff to exceed infiltration and depression storage. Therefore factors such as biological activity and raindrop compaction continually altering the surface soil fabric cannot be ignored.
By contrast, the highly transmissive surface soils make Hortonian overland flow impossible in undisturbed rainforest, and the 'throttle' to vertical transmission of rainwater occurs below 0.2 m depth, away from the dense surface root mat. High rain intensities, long duration of storms and low prevailing soil matric potentials (¥) result in the rapid emergence of subsurface stormflow, especially in the top 0.25 m, and in turn widespread saturation overland flow in the monsoon season. This vindicates Burt's (1985: 582) observation for humid environments that ' ... amongst the various conclusions reached is the agreed dominance of subsurface (storm) flow as the major contributor to stormflow in its own right...'. Both flow components are significant contributors to quickflow, but saturation overland flow dominates in the monsoon (December-March) which is a remarkable response for a forested environment, and in areas of South Creek comparable with study site 2 in the post-monsoon (April-mid-June). Subsurface stormflow becomes a more significant contributor to streamflow over remaining areas of the catchment in the latter season, with saturation overland flow restricted to the highest rain intensities in storms. In terms of peak runoff volumes and lag times between rainfall and stream discharge, South Creek has a hydrological response more akin to an environment where Hortonian overland flow is the dominant contributor to stormflow rather than subsurface stormflow generally associated with humid forested lands, particularly in temperate areas. For the same catchment area of 0.257 km2, Dunne (1978, [Figure 7.18 and Figure 7,19]) estimated a peak runoff rate of 0.6 mm hr-1 and a lag time of 12.99 hours from experience in humid temperate catchments (some of which are forested) where subsurface stormflow is claimed to be the dominant contributor to stormflow. This rainforest catchment has recorded peak flows up to 74 mm hr-1 and average lag response times of 0.4 hrs (mode 0.3 hr, range 0.2-0.9 hr) during monsoon storms (January to March, 1976 and 1977). These responses are more towards response patterns of environments dominated by Hortonian overland flow, where a similar-sized catchment can be expected to produce peak runoff rates of 150 mm hr-1, with lag times of 0.32 hrs (Dunne, 1978 [Figures 7.7 and 7.8]). This highlights the significance of saturation overland flow as a contributor to stream flow in this environment and contributes to the high annual and individual storm runoff coefficients. About 65% (2515 mm) of the annual rainfall (4009 mm) appears as runoff and 47% of this amount appears as quickflow, mostly between December-May (Gilmour, 1975), thus incorporating the monsoon and the bulk of the post-monsoon season.
The disturbed North Creek would still favour saturation overland flow, but the lower surface K* could also produce infiltration-excess, Hortonian overland flow during the most highest short-term rain intensities, especially in the summer monsoon season. A saturation overland flow is also capable of developing almost instantaneously following commencement of storms in the undisturbed rainforest, so there should be little difference in the timing of both overland flow mechanisms which explains why only minor changes in the storm runoff response occurred between the two drainage basins.
By contrast, Nortcliff and Thornes (1981) noted the occurrence of overland flow was rare in a hillslope study located in the Amazonas, near Manaus. The possible reasons will be discussed later. In the same area, Salati and Vose (1984) reported only 19 to 26 percent of the annual rainfall (c 2000 mm) appearing as total runoff (quickflow and delayed flow) from Amazon rainforest basins near Manaus. Elsewhere in undisturbed equatorial rainforest, Nik (1987) reported low annual runoff coefficients ranging between 9.0 and 14.3 percent across three drainage basins in the Sungai Tekem Experimental Basin Project in Pahang, Peninsular Malaysia where mean annual rainfall was 1722 mm. Law and Cheong (1987) noted quickflow accounted for only 21 percent of annual runoff due to the high infiltration capacities of these drainage basin soils. No data was presented on the hillslope hydrology, but the work was reviewed by Bruijnzeel (1989) who suggested that the relative absence of increased stormflows even following forest clearance, indicated that significant overland flow along hillslopes still did not occur. Other separate studies in lowland or highland equatorial rainforest for example, in Tanzania by Edwards (1979) and Lundgren (1980), and in the reviews by Lal (1981), Walsh (1980) and Douglas and Spencer (1985) all point to saturation overland flow being relatively unimportant except in highly localized areas in some of these environments.
The significance of rain intensity is central to the description of runoff generation in these studies, and was shown statistically by cross-correlation and lagged regression analysis referring to the rainforest (Bonell et al, 1979; 1981), and by simple regression and partial correlation analysis in connection with times to runoff and overland flow volumes in the semi-arid area (Bonell and Williams, 1986b). By definition, the role of rain intensity in producing Hortonian overland flow makes this conclusion perhaps predictable, although the temporal variability of rain amounts required for ponding between discrete events adds another perspective due to soil fabric changes. The evidence from the rainforest, however, differs from other environments. For example, the comprehensive studies by Hewlett and co-workers (Hewlett et al, 1977, 1985; Hewlett and Bosh, 1984) statistically showed that rain intensity had no appreciable effect on quickflow volumes and only a small effect on peak flows. The data base for that work was drawn from temperate areas where reported rain intensities, for example max 1 hr, and response ratios (quickflow/gross storm rainfall) are low compared with South Creek. The much lower rain intensities experienced in the Coweeta Experimental Forest project in the southern Appalachians are reinforced by the later summary of 40 years of rainfall data by Swift et al (1988). They noted that the reason why Hewlett et al (1984) strongly questioned the value of intensity in that environment was because '... nearly three-quarters of Coweeta's precipitation falls with an intensity of less than 10 mm hr-1, and only 10 percent of storms have maximum intensities over 50 mm hr-1 (Swift et al, 1988, p. 43).
It is our intention to analyze the long-term rainfall records for both North and South Creek in the same way as Hewlett et al (1984), but based on the meteorological seasons, monsoon (December-March), post-monsoon (April-mid-June), winter (mid-June-September) and pre-monsoon (October-November) (Bonell and Gilmour, 1980) rather than lumping all the storms together. Much of the discussion in this paper has concentrated on the first two seasons when 85 percent of annual rainfall occurs.
The measurement of soil hydraulic properties and the use of elementary soil physics has been fundamental to both studies in the interpretation of runoff process on the lines called for by Burt (1985). The establishment of field saturated hydraulic conductivity as the dominant parameter in both studies holds significance for extrapolation to other 'similar' areas. A preliminary assessment of other geological and soil types has already been made on the wet tropical coast (Bonell et al, 1983b). In the meantime work has just been completed measuring K* on a 75 m2 grid in both North and South Creeks to assist interpretation of the long-term rainfall-runoff records and for use in O'Loughlin's (1986) topography-catchment wetness model.
Despite the subsoil ( > 0.2 m) of the rainforest impeding most of the rain, the transitional 0.1 to 0.2 m layer is capable of causing ponding during the highest short-term rains. It is ironic that the K* of this horizon is in the same order of magnitude as that of the non-vegetated areas in the semi-arid eucalypt woodland, and that maximum 1 minute rain amounts between the two environments are not too dissimilar (unpublished data). There are some parallels then in that each layer can act as a 'throttle' to downward movement of rainwater and this in turn produces overland flow. The point of genesis, however, makes this component hydrologically different between each environment.
The occurrence of widespread saturation overland flow and subsurface stormflow in rainforest, particularly in the summer monsoon, represents part of the extreme 'wet' hydrological situation in the context of Hewlett's variable source area concept. The proportion of South Creek contribution to quickflow depends on the transit distance saturation overland flow has to take before it is tapped by organized drainage lines, either perennial or ephemeral. The soil hydraulic properties, the intensity and long duration of monsoon storms, the high drainage density and the steep catchment slopes favour widespread contributing areas to quickflow. There is pedological evidence for natural erosion by saturation overland flow on the steeper slopes with only remants of the A2 and A3 horizons surviving. Research on hillslope erosion coupled with statistical (Hewlett et al, 1984) and topography-wetness modelling (O'Loughlin, 1986) will further check this conclusion. Furthermore new initiatives during the 1989-90 wet season will attempt to reconcile the earlier hydrometric study with the stream hydrographs utilizing the environmental isotopes deuterium, oxygen-18 and radon-222 to identify runoff sources (eg.: hydrograph separation into 'old' and 'new' water) on the lines described elsewhere (Bonell et al, 1989; Rodhe, 1987; Pearce et al, 1986; Sklash and Farvolden, 1979).
In the semi-arid study, the soil hydraulic properties and short-term rain intensities favour the classical Horton view of widespread areas of a catchment contributing to flood peaks. However, only a very small proportion of rain is transferred downslope to organized drainage and most overland flow is redistributed. This area differs from the Babinda study in that the storm durations are too short to maintain surface saturation for long periods, the slopes are too shallow for rapid transfer of overland flow, and the drainage network too poorly developed to capture most of it. Following the variable source area concept, the contributing area to flood peaks would be dependent on such factors as the duration as well as the intensity of storms. For most rain events, this area would seem limited to the valley bottoms.
Much emphasis has been placed on rain intensity and how it makes these environments hydrologically active in comparison with other areas, especially with regard to the tropical rainforest. There is a link here between hillslope hydrology and synoptic climatology. Bonell et al (1986) reviewed the principal rain-producing systems affecting the wet tropical coast, but such descriptions have some relevance to the Torrens Creek area as well. In that review it was noted that north-east Queensland is located on the southern rim of what Ramage (1968) termed 'the maritime continent'. This area extends across from the Indo-Malayan archipelago, through Papua New Guinea to the scattered islands of the west Pacific, and incorporates a broad region of sea temperature maximum and large scale release of energy from convective activity. As McAlpine et al (1983, p. 1) commented, the maritime continent ' ...acts as an important global heat engine driving not only its own internal atmospheric circulation but extensive regional circulations to the north and south as well'. The geographical position of north-east Queensland in relation to this 'continent' makes it one of the most meteorologically active areas of the tropics. There is a marked concentration of high intensity rain in a few months of the year associated with the inter-hemisphere monsoon exchange of air (as represented by the monsoon trough and defined by Sadler and Harris, 1970); and the poleward draining of latent heat energy and moisture liberated from convective activity in the maritime continent by low pressure troughs in the upper westerly circulation (as defined by McAlpine et al, 1983). All four rain generating mechanisms of WMO (1983) are represented on the wet tropical coast (convection, convergence, orographic and cyclonic) (Bonell et al, 1986), but what makes this area outstanding is that circular disturbances can be almost stationary for several days, embedded in the monsoon trough over the warm waters of the Gulf of Carpentaria or western Coral Sea. Such disturbances are more commonly tropical depressions (as defined by Lourensz, 1981) rather than tropical cyclones (as defined by Bureau of Meteorology, 1978), and the nature of this activity means that rain occurring on a few days makes up a large proportion of the annual total. For example, 48 percent of the annual 1981 rainfall total of 5324.5 mm occurred between January 3-17, inclusive.
As discussed elsewhere (Bonell et al, 1986) a significant proportion of tropical rainforest, for example in Indonesia, Malaysia and the Congo basin, is associated with the maximum cloud zone (as defined by Davidson et al, 1983; Sadler, 1974) of the summer equatorial westerlies (Atkinson and Sadler, 1970), or orographic uplift of winter easterlies where rain occurs from deep convective cells and not from well organized circular systems in either season. The Amazon is different, with the main influx of atmospheric moisture coming into the basin only as easterlies from the Northern Hemisphere trade winds (Atkinson and Sadler, 1970; Salati and Vose, 1984). However, the geographical location of the area prevents the formation of tropical cyclone or depressions, and convective cells are once again the rainfall source as in the equatorial westerlies. Differences, then, can be expected solely on meteorological grounds between the hydrological response of these non-tropical cyclonic areas and the hydrological response for the wet tropics of north-east Queensland based on rainfall intensity (Bonell et al, 1986). In a survey of tropical rainfall, Jackson (1988a, p. 112) noted that ' ... lack of data for tropical areas, particularly in the case of conditions for periods less than 24 hrs makes it impossible to give indications of highest falls and intensities experienced'. However, he concurred with the view that thunderstorms may produce the highest falls for periods of 1 hr or less, especially identified with the low latitude areas like the Amazon. In turn Jackson (1988a, p. 112) noted that 'for longer time periods, high totals will be associated with major organized disturbances of which the most extreme case is the hurricane especially a slow moving, declining system. Since hurricanes in particular do not usually occur within about 5° of the equator, it follows that maximum totals are found in higher latitudes'. When comparing daily rainfall over northern Australia with other tropical areas, Jackson (1988b) noted that northern Australia rainfall stations (including some on the wet tropical coast of NE Queensland) tended to record most concentrated rainfall, i.e. fewer raindays and higher mean daily intensities, compared with most other areas of the tropics, thus emphasising northern Australia as a highly energetic environment. However the simple distinction between tropical cyclonic and non-cyclonic areas maybe inadequate, because an earlier study (Jackson, 1986) showed stations in central Africa having the closest rainfall characteristics to those found in northern Australia. Other factors such as orographic uplift over high topography may be as significant and need further investigation.
The more remote position of Torrens Creek, in relation to well-organized maritime disturbances, means that scattered thunderstorms are the most common source of rain. This contrasts with the wet tropical coast where thunderstorms are less frequent. The Gulf of Carpentaria is a major source of moisture into central-north Queensland induced by the eastward passage of upper meridional troughs, or more occasionally from the equatorial westerlies associated with the southward movement of the monsoon trough.
The clearing of forests in tropical uplands has often been misconceived as the major source of severe flooding, despite recent scientific reviews showing that there is no strong evidence for this belief (Hamilton, 1987; Bruijnzeel, 1987). Perhaps the commonly quoted example is just outside the tropics, concerning deforestation in the Middle Hills of Nepal being blamed for increased flooding in the lower Indo-Gangetic plain (eg: Nautiyal and Babor, 1985) even though preliminary studies have argued the contrary (Gilmour et al, 1987).
The process hydrology studies described here have shown that widespread saturation overland flow can occur within tropical rainforest where prevailing rain intensities are high and subsoil K* are very low, away from the surface horizons containing biological activity. Under these circumstances, floods can occur from pristine forested drainage basins on the same scale as from disturbed areas. For example, Bonell et al (1986) describe in detail the storm runoff generated from 2560 mm of rain between 3-17 January, 1981 in the Babinda drainage basins. During this period daily rain exceeded 200 mm on 6 days, with a maximum total of 433 mm on 12 January 1981. Total discharge for North Creek (disturbed) and South Creek (undisturbed) was respectively 2096 mm and 1880 mm and so did not differ greatly. In more detail, seven hydrograph consecutive separations were analyzed over this 14 day period and quickflow percentages of total rain ranged from 10 to 75 percent for North Creek and 18 to 64 percent for South Creek. The small difference between these treatments confirm that pristine forests do not act as infinite 'sponges', but can generate flood-producing runoff.
Discussion so far has concentrated on the effect of rainfall on lowland tropical rainforest. Further west on the mountain range, more impressive rain amounts are recorded. A rain gauge on Mt Bellenden Ker Top (1561 m), immediately to the west of Babinda, has a mean annual rainfall of 7664 mm, 1974-84 incl. (median 7210 mm, range 6305 mm (1974) to 11,346 mm (1977)). It is from such mountain areas covered in 'pristine' forests that a major contribution to the recurrent disastrous flooding is made by affecting the lower reaches of the Mulgrave, North and South Johnstone, Tully and Murry Rivers (Bonell, 1983). Annual runoff coefficients can also range between 0.58 and 0.90 (Bonell, 1988).
When the K* results presented by Herwitz (1986) are examined for a sample plot on the Mt Bellenden Ker range, it is evident that a similar trend exists as that described for the Babinda study in the order of magnitude decline in K* down the profile, despite differences in parent materials. The soil horizons below 0.20 m are once again the persistent impeding layer (K* range 2 to 7 mm hr-1), although the 0.05-0.20 m layer also would be impeding to maximum short term intensities in excess of 100 mm hr-1 (K* = 108 mm hr-1, 0.05 -0.20 m layer). Consequently, widespread saturation overland flow would occur during the summer monsoon season, but Herwitz produced results to show that a combination of high intensity rainfall and the funnelling effect of trees would produce 'localized', based stemflow fluxes capable of exceeding the surface K* and thus producing Hortonian overland flow.
Under these wet conditions, it can be concluded that forests respond to extremely high rainfalls similar to other land use conversions and do not act as infinite 'sponges'. One of the critical differences between tropical rainforests in say, the Amazonas, and NE Queensland concerns the vertical changes in K* down the soil profile. The results presented here show that the transmissive layer is very shallow and confined to the top 0.2 m of soil. Irrespective of any reductions in surface K* from compaction, the chief 'throttle' controlling the disposition of soil/rain water remains in the shallow subsoil ( 0.2 m depth). Good quality soil K* from other rainforest environments is scarce, but Nortcliff and Thornes (1981) presented K* for Amazonas Oxisols near Manaus which showed that those soils had a deeper transmissive layer down to 0.90 m depth (K* range 61.3 to 156.7 mm hr-1) before K* lowered to 21.7 mm hr-1 (0.90-1.15 depth) which is in the range of the more permeable soils of NE Queensland, eg. derived from basalts and colluvium (Bonell et al, 1983). The K* of the top 0.15 m of the Amazonas soils (K* = 921,3 mm hr-1) differed little from those reported from NE Queensland. The effects of surface compaction, however, in the Amazonas might have greater ramifications on the runoff process because the predisturbance subsoil 'throttle' is more remote from the surface (unlike in the Babinda study), and the upper transmissive layer is much deeper which acts as a buffer to overland flow. Nortcliff and Thornes' (1981) had noted that the occurrence of overland flow was rare. Consequently, the effect of transferring the 'throttle' layer to the surface in relation to short term rain intensities on disturbance, might be more significant in causing changes in the delivery mechanisms and magnitudes of storm runoff. The need for detailed K* measurements on the lines described in the Babinda study is highlighted to substantiate this argument.
CONCLUSION
The tropical rainforest presents the paradox of high surface infiltration rates typical of
humid forest environments but a runoff response that is not typical of those environments. Given the
high field saturated hydraulic conductivity of the surface rainforest soils, Hortonian overland flow is
clearly not possible. But the high prevailing rain intensities and the low K* subsurface values ensure
the frequent presence of widespread saturation overland flow in the summer, and cause this catchment
to be highly responsive to rainfall inputs. Similarly, the surface permeability of the open
eucalypt woodland soils is comparatively high despite raindrop compaction producing sealed areas.
Ponding and subsequent Hortonian overland flow only occur because short-term rain intensities are high.
There is a parallel between these contrasting environments in that each has a 'throttle' layer to the
prevailing inputs, viz at the surface in the semi-arid study and the subsoil in the rainforest. This is caused
by opposite trends in the order of magnitude of change in K* down the respective soil profiles.
Field saturated hydraulic conductivity is a significant hydraulic property in both environments,
where temporal as well as spatial variability has to be considered, especially in the semi-arid study.
In contrast to the rainforest example, most overland flow is redistributed within the
semi-arid study because of the very shallow slopes. Such redistribution, however, is sufficient to provide a
high erosion potential through rain-flow transportation (Moss
et al., 1979) following disturbance of
these landscapes. The runoff characteristics on the wet tropical coast cause disastrous erosion to occur
on disturbance as shown by the Babinda study. In addition, average annual losses of 100 tonnes
ha-1 have been estimated for 11,000 ha of sugar canelands on various soils north of Cardwell (Capelin and
Prove, 1983).
There are, then, close links between storm drainage and synoptic climatology in
north-east Queensland. Such links have been largely ignored in other runoff processes studies because
most research has been undertaken in humid temperate areas of western Europe and the eastern United
States where rain intensities are much lower in magnitude (examples: Hewlett
et al, 1977; Weyman, 1973) and where streams have a lower responsiveness to storms. The tropical rainforest study, in
particular, emphasize this link. This means that, when considering synoptic climatology in addition to soil
and slope factors, differences between this area and other rainforests can be expected. In a review of
runoff process and models in the humid tropics, Walsh (1980: 181) noted that of all the tropical
rainforest areas so far investigated, the Babinda catchments have ' ... the only
distinctively tropical runoff process pattern'. Clearly the high rainfalls in the Babinda catchments outweigh the lithologic influences.
Otherwise the soil hydraulic properties would have suggested a localized saturation overland flow
and subsurface stormflow model found elsewhere in humid tropical and temperate areas (Kirkby,
1978; Walsh, 1980). Detailed knowledge of the interaction between synoptic climatology and
runoff hydrology is clearly required.
Finally the Australian experience has emphasized the need for detailed measurements of
soil hydraulic properties, notably K*, at different scales of investigation in line with Beven's
(1989) comments and the need for an open, cascade trough system to monitor overland flow rather than
the commonly used bounded plot studies. This new design forms the basis for future
investigations concerning the possibilities of extrapolation of measured parameters to other ungauged areas.
Future research on these lines is suggested to resolve the continuous debate whether conversion of
forests increases storm runoff or not in a wide variety of humid tropical environments. In the meantime
an explanation is offered for the dichotomy of views in terms of the effect of land use conversion on
the runoff process.
ACKNOWLEDGEMENTS
These research programmes have been supported by the Australian Research Grants
Scheme, Australian Water Resources Council, National Soils Conservation Program, CSIRO (Division
of Soils), James Cook University (Special Research Grant and University Research Grant Schemes),
and the Queensland Department of Forestry.
| Return period | Duration (min) | |||||||||
| (years) | 10 | 20 | 30 | 40 | 50 | 60 | 70 | 80 | 90 | 24hr |
| 2 | 23.3 | 37.5 | 46.5 | 56.2 | 62.0 | 66.1 | 67.3 | 69.5 | 69.8 | |
| 10 | 30.6 | 49.3 | 61.2 | 74.4 | 82.4 | 87.9 | 89.5 | 92.5 | 92.7 | 117 |
| 25 | 132 | |||||||||
| 50 | 143 | |||||||||
| Return period | Duration (min) | |||||||||
| (years) | 6 | 12 | 18 | 30 | 60 | 3hr | 24hr | |||
| 2 | 12.7 | 23.1 | 29.3 | 42.2 | 67.5 | 120.9 | 335.8 | |||
| 7 | 20.5 | 28.3 | 40.3 | 58.5 | 90.7 | 209.3 | 497.0 | |||
| 14 | 24.7 | 32.0 | 45.5 | 64.5 | 103.9 | 244.0 | 660.9 | |||
| DIFFERENCES IN OVERLAND FLOW (mm) |
| Date | Total rain | Plot 1 Trough 2-Trough 1 | Plot 2 Trough 3-Trough 2 | Plot 3 Trough 4-Trough 3 | Plot 4 Trough 5-Trough 4 | Net Change Between Plots |
| 21.1.87 | 14.8 | 0.14 | -0.14 | 0.20 | -0.12 | 0.08 |
| 20.1.82 | 9.8 | 0.04 | 0.12 | 0.45 | -0.44 | 0.17 | 21.1.87 Storm A | 10.8 | 0.05 | 0.14 | 1.05 | -0.94 | 0.30 |
| 21.1.82 Storm B | 12.0 | 0.04 | -0.09 | 0.04 | -0.03 | -0.04 |
| 23.1.82 | 16.4 | -0.63 | 2.24 | 1.20 | -2.53 | 0.28 |
| 2.2.82 | 34.0 | -2.86 | 6.36 | 1.82 | -4.54 | 0.78 | /tr>
| 18.2.82 | 23.6 | -0.97 | 4.04 | 1.37 | -3.44 | 1.00 |
Source: Bonell and Williams, 1986b.
| N | Mean | Median | Max | Min | Stdev | |
| A parameter (m d-1) | ||||||
| Plot | 21 | 1.08 | 0.95 | 2.16 | 0.35 | 0.59 |
| Ring, bare | 24 24 | 2.78 | 2.76 | 10.02 | 0.52 | 2.04 |
| Ring, vegetation | 26 | 7.06 | 4.45 | 22.03 | 0.52 | 6.70 |
| S parameter (mm s-12) | ||||||
| Plot | 21 | 0.0070 | 0.0170 | 0.0860 | - 0.0740 | 0.0452 |
| Ring, bare | 24 | 0.0455 | 0.0316 | 0.3430 | -0.0450 | 0.0666 |
| Ring, vegetation | 26 | 0.0391 | 0.0280 | 0.1940 | -0.1490 | 0.0885 |
| K* (md-1) | ||||||
| Plot | 21 | 1.08 | 0.86 | 1.90 | 0.52 | 0.52 |
| Ring, bare | 24 | 2.63 | 2.42 | 7.69 | 0.86 | 1.53 |
| Ring, vegetation | 26 | 6.19 | 4.45 | 15.29 | 1.64 | 4.67 |
Source: Williams and Bonell, 1988.
Fig. 1 The location, experimental sites and physiographic features of the Babinda catchments.
Fig. 2 The distribution undisturbed/disturbed tropical rainforest in North Creek.
Fig. 3 Rainfall intensity - frequency - duration analysis for the Babinda experimental drainage basins (1971-83)
Fig. 4 The location of sampling points for K* determination for the impeding layer 0.02-0.5 m depth in South Creek and preliminary permeability.
Fig. 5 The location of sampling points for K* determination for the impeding layer 0.02-0.5 m depth in North Creek
Fig. 6 The experimental layout at the instrumented slope near Torrens Creek in central north Queensland, Australia
Fig. 7 Diagram to illustrate the concept of the 'repetitive unit' which is aimed at regarding the behaviour of an inhomogeneous material as an equivalent homogeneous material if the length scale of observation exceeds the characteristic length (L1 and L2 ) of the repetitive unit (after Bear, 1979).
Source: Williams and Bonell, 1988.
Fig. 8 Erosion-deposition versus runoff-runon for the pre- and post-fire periods. The equation shown is for a zero intercept. The constant for the fitted equation (y = 0.4543x + 0.546) was not significantly different from zero (p > 0.05). The ANOVA percentage variance accounted for is defined as 100 x (total mean square minus residual mean square)/(total mean square). The symbols A to D represent individual observations from plots 1 to 4 respectively. Where more than one observation is coincident this is shown numerically. The + symbol close to the origin indicates 28 almost coincident points.
Source: Bonell and Williams, 1987.
REFERENCES
Abbott, M. B.; Bathurst, J. C.; Cunge, J. A.; O'Connell, P. E. and Rasmussen, J. An introduction
to the European Hydrological System Système Hydrologique Européen, 'SHE', 1. History
and philosophy of a physically-based, distributed modelling system.
Journal of Hydrology, 87, 45-59, 1986a.
Abbott, M. B.; Bathurst, J. C.; Cunge, J. A.; O'Connell, P. E. and Rasmussen, J. An introduction
to the European Hydrological System Système Hydrologique Européen, 'SHE', 2.
Structure of a physically-based, distributed modelling system.
Journal of Hydrology, 87, 61-77, 1986b.
Abdul, A. S. and Gillham, R. W. Laboratory studies of the effects of the capillary fringe on
stream flow generation, Water Resources Research,
20(6), 691-698, 1984.
Arnold, G. O. and Fawckner, J. F. The Broken River and Hodgkinson Provinces. In: Henderson,
R. A. and Stephenson, B. J. (eds.), The Geology and Geophysics of North-east Australia,
Geol. Soc. Aust. Inc., Qld. Div. July 1980, 175-189, 1980.
Atkinson, G. D. and Sadler, J. C. Mean-cloudiness and gradient-level-wind charts over the
tropics. Vol. II, Charts, Technical Report
215, Air Weather Service (MAC), USAF, 1970.
Baveye, P. and Sposito, G. The operational significance of the continuum hypothesis in the
theory of water movement through soils and aquifers.
Water Resources Research, 20, 521-530, 1984.
Bathurst, J. C. Physically-based distributed modelling of an upland catchment using the
Système Hydrologique Européen. Journal of Hydrology,
87, 79-102, 1986a.
Bathurst, J. C. Sensitivity analysis of the Système Hydrologique Européen for an upland catchment.
Journal of Hydrology, 87, 103-123, 1986b.
Bear, J. Hydraulics of Groundwater, McGraw-Hill, New York, pp. 28-33, 1979.
Beven, K. J. Towards a new paradigm in hydrology.
International Association of Hydrological
Sciences, Water for the Future Hydrology in Perspective,
(Ed. J. C. Rodda and N. C. Matelas), Publication N° 164, pp. 393-403, 1987.
Beven, K. J. Changing ideas in hydrology the case of physically-based models,
Journal of Hydrology, 105, 157-172, 1989.
Bonell, M. Hydrological processes and implications for land management in forests and
agricultural areas of the wet tropical coast of north-east Queensland. In:
Fluvial Geomorphology of Australia (Ed. R. F. Warner), Academic Press, 41-68, 1988.
Bonell, M.; Cassells, D. S. and Gilmour, D. A. Vertical and lateral soil water movement in a
tropical rainforest catchment. In: O'Loughlin, E. M. and Bren, L. (Eds.),
National Symposium on Forest Hydrology, (Proc. Melbourne Symp. May 1982), Canberra, Institution of Engineers,
Australia/Forest Hydrology Working Group Australian Forestry Council, 30-38, 1982.
Bonell, M.; Cassells, D. S. and Gilmour, D. A. Vertical soil water movement in a tropical
rainforest catchment in northeast Queensland, Earth Surface Processes and Landforms,
8(3), 253-272, 1983a.
Bonell, M.; Cassells, D. S. and Gilmour, D. A. Spatial variations in soil hydraulic properties
under tropical rainforest in north-eastern Australia,
International Conference on Infiltration,
Development and Application, (Ed. Yu-si Fok), University of Hawaii, Water Resources Center,
January 1987, 153-165, 1987.
Bonell, M. and Gilmour, D. A. Variations in short-term rainfall intensity in relation to
synoptic climatological aspects of the humid tropical north-east Queensland coast,
Singapore Journal of Tropical Geography, 1(2), 16-30, 1980.
Bonell, M.; Gilmour, D. A. and Cassells, D. S. A preliminary survey of the hydraulic properties
of rainforest soils in tropical northeast Queensland and their implications for the runoff process.
In: the Ploey, J. (Ed.), Rainfall Simulation, Runoff and Soil Erosion,
Catena Supplement 4, 57-78.
Bonell, M.; Gilmour, D. A. and Cassells, D. S. The storm runoff response to various rainfall
systems on the wet tropical cost of northeast Queensland.
East-West Center, Environment and Policy Institute Working Paper, Honolulu, Hawaii, 96848, 48 pp. Presented in modified form
in: Australian National Rainforests-Study
Report, vol. 2, Australian Heritage Commission (P.
A. Kershaw and G. Werran, eds.), 1989 (in press).
Bonell, M.; Gilmour, D. S. and Sinclair, D. F. A statistical method for modelling the fate of
rainfall in a tropical rainforest catchment, Journal of
Hydrology, 42, 241-257, 1979.
Bonell, M.; Gilmour, D. A. and Sinclair, D. F. Soil hydraulic properties and their effect on
surface and subsurface water transfer in a tropical rainforest catchment,
Hydrological Sciences Bulletin, 26, 1-18, 1981.
Bonell, M.; Pearce, A. J. and Stewart, M. K. The identification of runoff-production
mechanisms using environmental isotopes in a tussock grassland catchment, eastern Otago, New
Zealand, Hydrological Processes, 3, 1989 (in press).
Bonell, M. and Williams, J. The two parameters of the Philip infiltration equation: their
properties and spatial and temporal heterogeneity in a Red Earth of tropical semi-arid Queensland,
Journal of Hydrology, 87, 9-31, 1986a.
Bonell, M. and Williams, J. The generation and redistribution of overland flow in a massive oxic
soil in a eucalypt woodland within the semi-arid tropics in north Australia,
Hydrological Processes, 1, 31-46, 1986b.
Bonell, M. and Williams, J. Infiltration and redistribution of overland flow and sediment on a
low relief landscape of semi-arid, tropical Queensland.
International Association of Hydrological Sciences, Forest Hydrology and Watershed Management,
Publication N° 167, pp. 199-211, 1987.
Bosch, J. M. and Hewlett, J. D. A review of catchment experiments to determine the effect
of vegetation changes on water yield and evapotranspiration.
Journal of Hydrology, 55, 3-23, 1982.
Bruijnzeel, L. A. Environmental impacts of (de)forestation in the humid tropics. A
watershed perspective MURS Programme 1986-87.
Water for Mankind in the Years 2000, Paris,
France, Seminar 25-26 April 1987, 33 pp., 1987a.
Bruijnzeel, L. A. On the hydrology of moist tropical forests: with special reference to the study
of nutrient cycling. Paper presented at British Ecological Society Symposium on
mineral nutrients in tropical forest and savanna ecosystems, Stirling, 9-11 September 1987, 45
pp. 1987b.
Bruijnzeel, L. A. Review Paper. (De)forestation and dry season flow in the tropics: A closer
look. Journal of Tropical Forest Science, 1989 (in press).
Burch, G. J.; Bath, R. K.; Moore, I. D. and O'Loughlin, E. M. Comparative hydrological
behaviour of forested and cleared catchments in south eastern Australia.
Journal of Hydrology, 90, 19-42, 1987.
Burdekin Project Committee. Resources and Potential of the Burdekin River Basin, Queensland
Appendix 2a: Land and Associated Features, Canberra, Australian Government Publishing Service, 1976.
Bureau of Meteorology. Australian Tropical Cyclone Forecasting Manual,
Australian Bureau of Meteorology, 1978.
Burt, T. P. Slopes and slope processes, Progress in Physical Geography,
9, 582-599, 1985
Capelin, M. A. and Prove, B. G.. Soil conservation problems of the humid coastal tropics of
north Queensland, Proceedings of Australian Society of Sugar Cane Technologists,
87-93, 1983.
Calder, I. R. What are the limits on forest evaporation? Comment,
Journal of Hydrology, 82, 179-184, 1985.
Cassells, D. S.; Gilmour, D. A. and Bonell, M. Watershed forest management practices in the
tropical rainforests of north-eastern Australian. In: O'Loughlin, C. L. and Pearce, A. J. (Eds.),
Proc. IUFRO Symp. on Effects of Forest Land Use on Erosion and Slope
Stability, East-West Center, Honolulu, NZ Forest Research Institute Publication, 289-298, 1984.
Cassells, D. S.; Bonell, M.; Hamilton, L. S. and Gilmour, D. A. The protective role of tropical
forests: A state of knowledge review. In: Agroforestry in the Humid Tropics its protective
and ameliorative roles to enhance productivity and sustainability.
(Eds. N. T. Vergara and N. C. Briones), Environment and Policy Institute, East-West Center, Honolulu, Hawaii,
Southeast Asian Regional Center for Graduate Study and Research in Agriculture, College,
Laguna, Philippines, pp. 31-58, 1987.
Chang, J. Hydrology in humid tropical Asia. Paper presented at the UNESCO
International Colloquium on the Development of Hydrologic and Water Management Strategies in the
Humid Tropics, Townsville, 15-22 July, 1989, Australia, 21 pp. 1989.
Commonwealth of Australia, Bureau of Meteorology.
Rainfall Statistics Australia, Metric
Edition, Australian Government Publishing Service, 1977.
Coventry, R. J. and Williams, J. Quantitative relationships between morphology and current
soil hydrology in some Alfisols in semi-arid tropical Australia,
Geoderma, 33, 191-218, 1984.
Davidson, N. E.; McBride, J. L. and McAvaney, B. J. The onset of the Australian monsoon
during Winter MONEX: Synoptic Aspects, Monthly Weather Review,
111, 496-516, 1983.
DeKeyser, F. Explanatory Notes on the Innisfail, Qld Sheet
SE/55-6, Department of National Development, Bureau of Mineral Resources, Geology and Geophysics, Commonwealth
of Australia, 1964.
Douglas, I. and Spencer, T. (Eds.). Environmental Change and Tropical
Geomorphology, London, Allen and Unwin, 1985.
Dunne, T. Field studies of hillslope processes. In: Kirkby, M. J. (Ed.),
Hillslope Hydrology, Chichester, Wiley, 227-293, 1978.
Dunne, T. Relation of field studies and modelling in the prediction of storm runoff.
Journal of Hydrology, 65, 25-48, 1983.
Edwards, K. A. The water balance of the Mbeya experimental catchments.
East African Agricultural and Forestry Journal,
43, 231-247, 1979.
Foster, S. S. D. and Smith-Carrington, A. The interpretation of tritium in the chalk unsaturated zone.
Journal of Hydrology, 46, 343-364, 1980.
Gillham, R. W. The capillary fringe and its effect on water table response.
Journal of Hydrology, 67, 307-324, 1984.
Gilmour, D. A. Catchment water balance studies on the wet tropical coast of north Queensland.
Unpublished PhD Thesis, Department of Geography, James Cook University of
North Queensland, 1975.
Gilmour, D. A. Effects of logging and clearing on water yield and quality in a high rainfall zone
of northeast Queensland. In The Hydrology of Northern Australia
(Proc. Brisbane Hydrol. Symp., June 1977), Canberra, Institution of Engineers Australia, Nat. Conf. Publ., 775,
156-160, 1977.
Gilmour, D. A.; Bonell, M. and Sinclair, D. F. An investigation of storm drainage processes in
a tropical rainforest catchment. Australian Water Processes Council Technical Paper
56. Canberra, Australian Government Publishing Service, 1980.
Gilmour, D. A.; Bonell, M. and Cassells, D. S. The effects of forestation on soil hydraulic
properties in the middle hills of Nepal: a preliminary assessment.
Mountain Research and Development, 7, 239-249, 1987.
Gilmour, D. A.; Cassells, D. S. and Bonell, M. Hydrological research in the tropical rainforests
of north Queensland: Some implications for land use management. In:
First National Symposium on Forest
Hydrology (Eds. E. M. O'Loughlin and L. J. Bren), Melbourne, May 1982,
Instit. Engrs. Australia, Nat. Conf. Public, N° 82/6, pp. 145-152, 1982.
Gifford, G. F. Infiltrometer studies in rangeland plant communities of the Northern Territory,
Aust. Range J. 1(2), 142-149, 1978.
Griffiths, J. F. Climates of Africa, Elsevier, Amsterdam, 1972.
Hamilton, L. S. Overcoming myths about soil and water impacts of tropical forest land uses. In:
Soil Erosion and Conservation (Eds. by S A. El-Saify, W. C. Moldenhauer and A. Lo). Akeny,
Soil Conservation Society of America, pp. 680-690, 1986.
Hamilton, L. S. The environmental influences of forests and forestry in enhancing food
production and food security. FAO, Rome (1987). Presented at
Expert Consultation on Forestry and Food
Production/Security, Bangalore, India, 14-20 February, 1988, 69 pp, 1988.
Hamilton, L. S. (with P. N. King). Tropical Forested Watersheds: Hydrology and Soils Response to
Major Uses or Conversions, Boulder, Westview Press, 1983.
Hardjono, H. W. Influence of a permanent vegetation cover on streamflow. In:
Proceedings Seminar on Watershed Management, Development and Hydrology,
Solo, 3-5 June 1980, pp. 280-297 (in Indonesia reported by L. A. Bruijnzeel, 1989), 1980.
Herwitz, S. R. Infiltration-excess caused by stemflow in a cyclone-prone tropical rainforest.
Earth Surface Processes and Landforms, 11, 401-412, 1986.
Hewlett, J. D. Watershed managements. In: Report for 1961 Southeastern Forest Experiment
Station, Asherville, North Carolina, US Forest Service, 1961.
Hewlett, J. D. and Bosch, J. M. The dependence of storm flows on rainfall intensity and vegetal
cover in South Africa, Journal of Hydrology,
75, 365-381, 1984.
Hewlett, J. D.; Fortson, J. C. and Cunningham, G. B. Additional tests on the effect of rainfall
intensity on stormflow and peak flow wild-land basins,
Water Resources Research, 20, 985-989, 1984.
Horton, R. E. The role of infiltration in the hydrologic cycle,
Transactions of American Geophysical Union,
14, 446-460, 1933.
Horton, R. E. Erosional development of streams and their drainage basins: hydrological approach
to quantitative morphology. Bull. Geol. Soc. Amer.,
56, 275-370, 1945.
Jackson, I. J. Relationships between raindays, mean daily intensity and monthly rainfall in the
tropics. Journal of Climatology, 6, 117-134, 1986.
Jackson, I. J. Climate, Water and Agriculture in the Tropics.
Longman, Second Edition, 1988a.
Jackson, I. J. Daily rainfall over northern Australia: Deviations from the world pattern.
Journal of Climatology, 8, 463-476, 1988b.
Kirkby, M. J. (Ed.). Hillslope Hydrology,
Chichester, Wiley, 1978.
Kirkby, M. J. and Morgan, R. P. C. Soil Erosion,
Wiley, Chichester, 1980.
Lal, R. Deforestation of tropical rainforest and hydrological problems. In:
Tropical Agricultural Hydrology Watershed Management and Land Use
(Eds. by R. Lal and E. W. Russell), Wiley, Chichester, pp. 131-140, 1981.
Law, K. F. and Cheong, C. W. Effects of land use changes on the hydrological characteristics of
Sungai Tekam Experimental Basin. Paper presented at UNESCO/MRP workshop on
Impact of Operations in Natural and Plantation Forest on Conservation of Soil and Water Resources,
23-26 June 1987, Universiti Pertanian Malaysia, 50 pp., 1987.
Littleboy, M.; Silburn, M. and Rosenthal, K. RRUMS Version 3.0 User Manual. Qld. Dept.
Primary Industries, Brisbane, 1986.
Lockwood, J. G. Probable maximum 24 hour precipitation over Malaya by statistical methods.
Meteorological Magazine, 96, 11-19, 1967.
Lourensz, R. S. Tropical Cyclones in the Australian Region, July 1909 to June 1980,
Canberra, Australian Government Publishing Service, 1981.
Lundgren, L. Comparison of surface runoff and soil loss from runoff plots in forests and small
scale agriculture in the Usambera Mts, Tanzania. Geografiska Annaler, 62, 113-148, 1980.
Lyne, V. and Hollick, M. Stochastic time-variable rainfall-runoff modelling. In:
Hydrology and Water Resources Symposium, Perth 1979,
Canberra, Institute of Engineers, 89-92, 1979.
McAlpine, J. R. and Keig, G. with Falls, R.
Climate of Papua New Guinea, Canberra,
Commonwealth Scientific and Industrial Research Organization in association with Australian
National University Press, 11-38, 1983.
McIntyre, D. S. Permeability measurements of soil crusts formed by raindrop impact.
Soil Science, 85, 185-189, 1958a.
McIntyre, D. S. Soil splash and the formation of surface crusts by raindrop impact.
Soil Science, 85, 261-266, 1958b.
McNaughton, K. G. and Jarvis, P. G. Predicting effects of vegetation changes on transpiration
and evaporation. In: Water Deficits and Plant Growth
(T. T. Kozlowski, Ed.), Academic Press, vol. 7, pp. 1-47, 1983.
Middleton, H. e. Properties of soils which influence soil erosion.
USDA Technical Bulletin, pp. 1-16, 1930.
Moore, J. D.; Mackay, S. M.; Wallbrink, P. J.; Burch, G.J. and O'Loughlin, E. M.
Hydrologic characteristics and modelling of a small forested catchment in southeastern New South
Wales, Pre-logging condition, Journal of Hydrology,
83, 307-335, 1986.
Morton, F. I. What are the limits on forest evaporation?
Journal of Hydrology, 74, 373-398, 1984.
Morton, F. I. What are the limits on forest evaporation? Reply
Journal of Hydrology, 82, 184-192, 1985.
Moss, A. J.; Walker, P. H. and Hutka, J. Raindrop-stimulated transportation in shallow water
flows: an experimental study. Sedim. Geol.,
22, 165-184, 1979.
Mott, J.; Bridge, B. J. and Arndt, W. Soil seals in tropical tall grass pastures of northern Australia,
Aus. J. Soil. Res, 17, 483-494, 1979.
Nautiyal, J. C. and Babor, P. S. Forestry in the Himalayas How to avert an environment
disaster. Interdisciplinary Science Reviews,
10, 27-41, 1985.
Nik, A. R. Impact of forest conversion on water yield in peninsular Malaysia. Paper presented at
the Workshop on Impact of Operations in Natural and Plantation Forests on Conservation of Soil
and Water Resources, held at Universiti Pertanian, Malaysia, Serdan, 23-26 June 1987, 17
pp., 1987.
Nortcliff, S. and Thornes, J. B. Seasonal variations in the hydrology of a small forested catchment
near Manaus, Amazonas, and the implications for its management. In:
Tropical Agricultural Hydrology Watershed Management and Land Use
(Eds. by R. Lal and E. W. Russell), Wiley, Chichester, pp. 37-57, 1981.
O'Brien, A. L. Rapid water table rise, Water Resources Bulletin,
American Water Resources Association, 18(4), 713-715, 1982.
O'Loughlin, E. M. A method for analyzing catchment topography for applications to hydrology
and land conservation, Water Resources Research,
22, 794-804, 1986.
Pearce, A. J.; Stewart, M. K. and Sklash, M. G. Storm runoff generation in humid
headwater catchments 1. Where does the water come from?
Water Resources Research, 22, 1263-1272, 1986.
Peck, A. J. and Williamson, D. R. Effects of forest clearing on groundwater.
Journal of Hydrology, 11, 147-66, 1987.
Philip, J. R. Theory of infiltration, Advances in Hydroscience,
5, 215-296, 1969.
Pilgrim, D. H.; Cordery, I. and Doran, D. G. Assessment of runoff characteristics in arid western
New South Wales, Australia. In: The Hydrology of Areas of Low Precipitation.
Proceedings of the Canberra Symposium, International Association of Hydrological Sciences, Publication
128, 141-150, 1979.
Ramage, C. S. Role of a tropical 'maritime continent' in the atmospheric circulation,
Monthly Weather Review, 96, 365-370, 1968.
Reynolds, W. D.; Elrick, D. E. and Topp, G. C. A re-examination of the constant head
well permeameter method for measuring saturated hydraulic conductivity above the water
table. Soil Science, 136-250-268, 1983.
Richards, T. H. The Geological History of the Frankland Island Region, Coastal North
Queensland, Unpublished Honours Thesis, Department of Geology, James Cook University of
North Queensland, 1977.
Rodhe, A. The Origin of Streamwater Traced by Oxygen-18,
Uppsala University, Department of Physical Geography, Division of Hydrology (Doctoral Thesis), Report Series A, N° 41,
260, pp. 1987.
Rubin, J. Theory of rainfall uptake by soils initially drier than their field capacity and its
application. Water Resources Research, 2, 739-749, 1966.
Sadler, J. C. The Monsoon Circulation and Cloudiness Over the GATE Area,
Department of Meteorology, University of Hawaii.
Sadler, J. C. and Harris, B. E. The Mean Tropospheric Circulation and Cloudiness Over South-east
Asia and Neighbouring Areas, Hawaii Institute of Geophysics, University of Hawaii, 1970.
Salati, E. The forest and the hydrological cycle. In:
The Geophysiology of Amazonia Vegetation and Climate Interactions
(Ed. by R. E. Dickinson), Wiley/UNU, pp. 273-296, 1987.
Salati, E. and Vose, P. B. Amazon Basin? A system in equilibrium,
Science, 225 (4658) 129-138, 1984.
Shuttleworth, W. J. Evaporation from Amazonian rainforest.
Proceedings of the Royal Society of London,
Series B, 233, 321-346, 1988.
Shuttleworth, W. J. Priorities in Climate Related Hydrological Research in Amazonia,
Institute of Hydrology, Unpublished Report, 16 pp. 1989.
Sisson, J. B. and Wieringa, P. J. Spatial variability of steady-state infiltration rates as a
stochastic process. Soil Science Society of America Journal,
46, 20-26, 1981.
Sklash, M. G. and Farvolden, R. N. The role of groundwater in storm runoff.
Journal of Hydrology, 43, 45-65, 1979.
Soil Survey Staff. Soil Taxonomy, A Basic System of Soil Classification for Making and Interpreting
Soil Surveys, US Department of Agriculture Handbook 436, 1975.
Stace, H. C. T.; Hubble, G. D.; Brewer, R.; Northcote, K. H.; Sleeman, R. Jr.; Mulcahy, M. J.
and Hallsworth, E. G. A Handbook of Australian Soils,
Glenside, Rellim Technical Publications, 1968.
Swift, L. W.; Cunningham, G. B. and Douglas, J. E. Climatology and Hydrology. In:
Forest Hydrology and Ecology at Coweeta (Eds. W. T. Swank, D. A. Crossley), Springer-Verlag, New York,
pp. 35-55, 1988.
Talsma, T. and Hallam, P. M. Hydraulic conductivity measurement of forest catchments,
Australian Journal of Soil Research, 18, 139-148, 180.
Walsh, R. P. D. Runoff processes and models in the humid tropics,
Zeitschrift fur Geomorphologie, NF Supplement-Band 36, 176-202, 1980.
Weyman, D. R. Measurements of the downslope flow of water in a soil.
Journal of Hydrology, 20, 267-288, 1973.
Williams, J. and Bonell, M. Computation of soil infiltration properties from the surface hydrology
of large field plots. International Conference on Infiltration, Development and
Application (Ed. Yu-Si Fok), University of Hawaii, Water Resources Center, pp. 272-281, 1987.
Williams, J. and Bonell, M. The influence of scale of measurement on the spatial and
temporal variability of the Philip infiltration parameters an experimental study in an
Australian savannah woodland. Journal of Hydrology,
104, 33-51, 1988.
Williams, J. and Coventry, R. J. The contrasting hydrology of red and yellow earths in a
landscape of low relief. In: The Hydrology of Areas of Low Precipitation,
Proceedings of the Canberra Symposium, International Association of Hydrological Sciences, Publication 128,
385-395, 1979.
Williams, J. and Coventry, R. J. The potential for groundwater recharge through red, yellow and
grey earth profiles in central north Queensland. In:
Groundwater Recharge. Australian Water Resources Council, Conference Series 3, 169-181.
Williams, J.; Day, K. G.; Isbell, R. F. and Reddy, S. J. Soils and Climate. In: Muchow, R. C. (Ed.),
Agro-Research for the Semi-Arid Tropics: North-West
Australia.. St. Lucia University of Queensland Press, 31-92, 1985.
World Meteorological Organization (WMO). Operational Hydrology in the Humid Tropical Regions.
Final Report, Geneva, 1983.
Youngs, E. G. Soil physical theory and heterogeneity.
Agricultural Water Management, 6, 145-159, 1983.
