Rainfall-runoff modelling:
Parsimonious spatial representation of tropical soils within dynamic, rainfall-runoff models
New and more considered application of models that simulate the rainfall-runoff behaviour of tropical catchments, would greatly add to our understanding of tropical hydrology and provide greater objectivity in the assessment of water-related management issues. Choice of model is often difficult, but we believe should have consideration for the level of detailed, quantitative information that is available for the catchment under study.
Recent work has demonstrated that the functional relationship between catchment rainfall and the river-discharge generated, often requires very few (3 to 5) model catchment parameters* to be described. The use of greater numbers of catchment parameters within more complicated model-structures* (that characterise many process-based catchment models: see Beven, 2001a), therefore, should necessitate that these parameter-sets / model-structures are tested and hence 'justified' by distributed (thus inevitably, semi-quantitative) field observations. As model complexity increases, the riverflow time-series becomes predictable from a very wide range of parameter sets, so that each sensitive* parameter becomes more and more uncertain. There is, therefore, value in limiting the complexity of models (i.e., parsimony*) to allow evaluation of the component structures or parameters. Dynamic rainfall-runoff models that utilise topographic information (e.g., TOPMODEL and TOPOG_SBM / TOPOG_DYNAMIC) can have more parsimonious model structures, though this is not always the case. While the value of their topographic indices in predicting saturated areas has been tested by numerous studies over the last decade (albeit within temperate climates), their spatial distributions (in 1 to 3 dimensions) of the other model component, the soil-rock transmissivity and/or soil-rock permeability*, has received much less attention, often because of the limitations of the field data. Yet, soil-rock permeability is often seen as the most important soil-rock property specified within physics-based models.
As a result, this partial assessment of the current status of tropical catchment modelling will focus on those issues central to the parameterisation* of soil-rock permeability within those dynamic rainfall-runoff models simplified by the use of topographic information. The key points being: (a) what is the relationship between field measurements of soil-rock permeability and the 'effective values' capable of simulating the riverflow time-series ? and (b) how complex can model parameterisations of soils become before we cannot justify them statistically ?
We believe that the results of the analyses presented, demonstrates that the spatial distribution of soil permeability derived from inversion* of catchment-scale models is very different to that derived from tests on small-samples (even allowing for the severe limitations of the model inversion process). The differences are consistent with the presence of zones of preferential flow* (i.e., natural soil pipes*, percolines* and fractures) that are inadequately characterised by tests on small-samples. A new strategy for deriving measures of soil permeability integrated over whole hillslopes (hence including larger preferential flow pathways) is illustrated. Early results suggest that the results from such techniques give soil-rock permeabilities that are more consistent with those integrated over whole catchments (and hence observable by catchment models).
Clearly, addressing the issue of how best to characterise those tropical soil parameters that regulate catchment rainfall-runoff behaviour, is an important precursor to the reliable prediction of how land-use change might alter tropical soil parameters and thence streamflow generation and nutrient transport.
Chappell, N.A., Bidin, K., Sherlock, M.D., and Lancaster, J.W. 2004. Forests-Water-People in the Humid Tropics, Cambridge University Press, Cambridge.
Climate cycle and land-use change sensitivity of tropical hydrological systems: A precursor to GCM land-use change simulations
Climate cycle and land-use change sensitivity of tropical hydrological systems: A precursor to GCM land-use change simulations.
Chappell, N.A., and Tych, W. 2002. UGAMP Newsletter, 25, 11.
Statistical modelling of rainfall and river flow in Thailand
Thailand experiences severe floods and droughts that have been shown to impact on the nation's agriculture. New technologies are being developed to study and forecast these events to form the basis for improved flood and drought alleviation policies and practices. Data-Based-Mechanistic modelling is a new approach, which can be used to study the rainfall and riverflow behaviour.
In this paper we use Dynamic Harmonic Regression (DHR) models to analyze the rainfall and the discharge time series across Thailand to find the evolution of annual and supra-annual cycles, (which reflect the character of seasonality, trends and forecasted rainfall and discharge. The aims of the study are first, to identify changes in rainfall regime and in river flow and their spatial distribution. Secondly, to identify statistical patterns in the frequency of extreme rainfall and flow periods with a view to improving predictions of medium and longer-term rainfall and river flow patterns.
The results indicate the existence of both temporal and spatial variation within the annual rainfall pattern in the 8 study catchments. For example, the seasonality of the rainfall in the South is less pronounced (i.e. more equatorial). The discharge seasonal pattern show stronger semi-annual cycles, with the weakest pattern in the South of country (station X.113), whereas the strongest discharge seasonality is observed in upper North (P.14). The overall areal rainfall trend has not changed significantly over the last 20 years. The identified dry years (1982-1983, 1991, 1997-1998) can be associated with the ENSO events. The discharge trend has also tended to drop in 1982 and 1987 ENSO years.
The DHR forecasts of rainfall and river flow data for 1998-1999 using data up to 1997 have relatively low prediction errors.
Boochabun, K. Tych, W., Chappell, N.A., and Carling, P.A. 2002. Global Continental Paleohydrology (GLOCOPH) 2002 Conference, Pune, India, 2-7 December 2002.
Dominant Flow Path (DFP) appraoch to flow and transport processes within catchments
This paper describes a 4-step approach to the identification and modelling of those 'land-units' that regulate the water, ion or particulate fluxes observed at the catchment scale. The steps are illustrated with examples of water, nitrate, and particulate fluxes within the Slapton Wood Catchment (Devon) and the Baru Catchment (Malaysian Borneo). These steps are: (1) Identify 'spatial patterns' (or variability) of flux integrated over 'large land-units' (i.e., 10 ha, or at least 1 ha in plan area) of a catchment. A nested contributory area structure can be used. (2) Identify and then focus on the locally important / dominant patches (land-units). (3) Characterise the 'temporal patterns' (dynamics) of representative land-units (in particular those with a high water, ion or particulate flux) given dynamic input variables (e.g., rainfall). An approach not constrained by physical concepts, that may be inappropriate at the 10 ha scale, may be a useful starting point (e.g., Data-Based-Mechanistic modelling). (4) Correlate flux patterns with those of 'catchment parameters', both observed at hectare-scales. Physical concepts / theory need to be addressed to understand which catchment parameters (e.g., permeability) are correlated with the flux patterns, however, this may require new parameter measurement directly at the scale of the 1-10 ha land-unit.
Chappell, N.A., 2001. International Conference and Workshop on Environmental Flows, 26-28 March 2001, Dundee, UK.
Parsimonious modelling of water and suspended-sediment flux from nested-catchments affected by selective tropical forestry
The ability to model the suspended-sediment flux (SSflux) and associated waterflow from terrain affected by selective-logging is important to the establishment of credible measures to improve the ecological sustainability of forestry practices. Recent appreciation of the impact of parameter uncertainty on the statistical credibility of complex models with little internal-state validation supports the use of more parsimonious approaches such as Data-Based-Mechanistic (DBM) modelling. The DBM approach combines physically-based understanding with model-structure identification based on transfer-functions and objective statistical inference. Within this study, these approaches have been newly applied to rainfall-SSflux response. The dynamics of the sediment system, together with the rainfall-riverflow system, were monitored at five, nested contributory areas within a 44-ha headwater region in Malaysian Borneo. The data series analysed covered a whole year at a 5-minute resolution, and were collected during a period some 5-6 years after selective timber harvesting had ceased. Physically-based and statistical interpretation of these data was possible given the wealth of contemporary and past hydro-geomorphic data collected within the same region.
The results indicated that parsimonious, three-parameter models of rainfall-riverflow and rainfall-SSflux for the whole catchment describe 80 and 90 % of the variance, respectively, and that parameter changes between scales could be explained in physically meaningful terms. Indeed, the modelling indicated some new conceptual descriptions of the riverflow- and sediment-generation systems. An extreme rainstorm having a 10-20-year return-period was present within the data-series and was shown to generate new mass movements along the forestry roads that had a differential impact on the monitored contributory areas. Critically, this spatially-discrete behaviour was captured by the modelling and may indicate the potential utility of DBM approaches for (a) predicting the differential effect of alternative forestry practices, (b) estimating uncertainty in the behaviour of ungauged areas and (c) forecasting riverflow and SSflux in terrain with temporal changes in rainfall regime and forestry impacts.
Chappell, N.A., McKenna, P., Bidin, K., Douglas, I., and Walsh, R.P.D. 1999. Phil. Trans. Roy. Soc. Lond. B., 354, 1831-1846.
The utility of multi-scale objective conditioning of a distributed hydrological model using uncertain estimates of saturated areas
For the purpose of environmental management, distributed physically based hydrological models are utilised. The inability to reliably measure the distributed physical characteristics of a catchment results in significant uncertainty in the parameterisation of physically based, distributed models. Calibration of such models is usually achieved with limited discharge data. Due to the parameteric complexity of such models, robust calibration of parameters is not achieved as these models are over-parameterised; many combinations of parameter values from many areas of the parameter space may produce simulations that fit the observed flow data reasonably well. The parameteric uncertainty produces significant predictive uncertainty, in terms of the range of discharge predictions, but also in terms of internal states of acceptable model simulations; the interaction of processes through ill-defined parameter estimates permits the reproduction of the discharge hydrograph via a range of modelled process mechanisms. In this study, uncertain estimates of the internal behaviour of catchment response are used as an additional criterion in the assessment of the acceptability of model simulations. In two applications, it is shown that such uncertain estimates may greatly reduce the parametric uncertainty associated with acceptable parameterisations, and hence the predictive uncertainty of such models. This is achieved through the improved definition of internal processes afforded by the uncertain estimates. Additionally, as it is shown that multi-objective criteria may be used to constrain parameter estimates by model inversion, investigation of multi-scale hydraulic conductivities is permitted. The scaling behaviour observed in one such investigation indicates that preferential flow pathways dominate the characteristic conductivity of the catchment response.
Franks, S.W., Beven, K.J., Chappell, N.A., and Gineste, P., 1997. In A.D. McDonald, and M. McAleer (Eds).,
Proceedings of the International Congress on Modelling and Simulation (MODSIM '97), Hobart, 8-11 December 1997, Volume 1, 335-340.
Multi-scale estimation of effective permeability within the Greenholes Beck catchment
This study addresses the problem of spatial variability and scale effects through a combination of fieldwork and modelling. A catchment has been instrumented to allow measurement of rainfall, stream discharge, piezometric levels and soil moisture content. This was supplemented with a variety of point-scale measurements of soil permeability. Because of the acknowledged problem of relating point-scale values to the catchment-scale techniques for the estimation of hillslope-scale permeability through the use of kinematic wave theory, pulse-wave and tracer experiments have been developed and applied to several hillslope transects. The catchment-scale rainfall-runoff data was first subjected to error analysis and then modelled using a combination of recession curve analysis, TOPMODEL theory and Data-Based Mechanistic modelling.
The upscaled point-scale estimates of permeability were found to give values lower than those provided by the hillslope-scale methods of analysis. This was attributed to the effects of macropore flow. The hillslope-scale estimates were in turn lower than those derived from the catchment-scale analysis. This is attributed to simplification of catchment processes within the catchment-scale model structure.
Ultimately, this work has illustrated the need to undertake multi-scale experiments and modelling in order to develop greater understanding of catchment hydrological processes.
Lancaster, J.W. 1999. Unpublished PhD thesis, Lancaster University, UK
Multi-scale permeability estimation in a tropical catchment
Physically based and spatially distributed modelling of catchment hydrology involves the estimation of block or whole-hillslope permeabilities. Invariably these estimates are derived by calibration against rainfall-runoff response. Rarely are these estimates rigorously compared with parameter measurements made at the small-scale.
This study uses a parametrically simple model, TOPMODEL, and an uncertainty framework to derive permeability at the catchment-scale. The utility of expert knowledge of the internal catchment dynamics (i.e., extent of saturated area) in constraining parameter uncertainty is demonstrated. Model-derived estimates are then compared with core-based measurements of permeability appropriately upscaled. The observed differences between the permeability estimates derived by the two methods might be attributed to the role of intermediate scale features (natural soil pipes). An alternative method of determining block permeabilities at the intermediate or hillslope scale is described. This method uses pulse-wave tests and explicitly incorporates the resultant effects of phenomena such as soil piping and kinematic wave migration. The study aims to highlight issues associated with parameterising or validating distributed models, rather than provide a definitive solution. The fact that the permeability distribution within the Borneo study catchment is comparatively simple, assists the comparisons. The field data were collected in terrain covered by equatorial rain forest. Combined field measurement and modelling programmes are rare within such environments.
Chappell, N.A., Franks, S.W., and Larenus, J. 1998. Hydrological Processes, 12, 1507-1523.
