Progress in arid-lands rainfall-runoff modelling

Abstract

Hydrological data are lacking in arid regions for a variety of reasons, making emprical methods of flood estimation unsatisfactory. Physically based models which are calibrated in the field using measurements of soil hydrological properties offer a viable alternative modelling strategy, and this article reviews the main elements required in such an approach. These include coupled submodels of 1) infiltration and runoff production; 2) overland flow routing; 3) channel routing to propagate flood waves down wadis beyond runoff source areas; and 4) transmissison losses into wadi beds. Examples are presented of flood simulations using a new physically-based model with such a structure. The role of both coarse-grained (bed material) and fine-grained (slackwater deposit) sediments in wadis in providing additional evidence of flood magnitudes is also discussed, and a methodology is suggested in which physical modelling and sedimentological analyses are combined.

A high magnitude storm and flood in a hyperarid catchment, Nahal Zin, Negev Desert, Israel

“…Desert¯ood evaluation, especially for large catchments, which is usually based on one stream¯ow gauge and a couple of rain gauges, is grossly inadequate, and often leads to erroneous conclusions. Much of the rainfall±runo modelling as generally practised today ignores the intricate meso-scale atmospheric patterns and complex terrain processes that control desert¯oods and is, therefore, largely irrelevant to their scienti®c understanding (Pilgrim et al, 1988;El-Hames and Richards, 1994). This study shows that¯ood hydrology in arid regions cannot be dealt with as a simple extension of semi-arid regions, only with less rainfall.…”

Conceptual modelling to assess the influence of hydro-climatic variability on runoff processes in data scarce semi-arid Andean catchments

“…4). Figure 5 shows the 5% and 95% uncertainty bounds compared to the observed discharge and precipitation 1973-1977Wet 1983-1986Wet/normal 1991-1994 Normal for the dry and wet periods in both catchments. The wet period calibration showed a median best fit of NSE = 0.68 and lnNSE = 0.65 for the Rio Grande, and median NSE = 0.53 and lnNSE = 0.52 for the Tascadero.…”

“…5). 1971-2001Dry 2003Dry 1973-1977Wet 1983-1986Wet/normal 1991-1994 Normal A semi-distributed configuration of the conceptual HBV light model was applied to capture the elevation gradient in precipitation in two perennial Andean headwater catchments. After testing alternative model structures applied to the baseline calibration period from 1971 to 2001, a simple model based on a snowmelt routine and two runoff generation reservoirs was adopted.…”

“…High-elevation data and information are particularly important to adequately represent snow-related processes in models in which climate, cryosphere and hydrological feedbacks are to be successfully simulated (Pellicciotti et al 2008, Ragettli et al 2014. Semi-arid regions are usually challenging environments for rainfall-runoff modelling; this is due partly to data scarcity and the significance of infrequent large precipitation events, but also to often unknown partitioning of contributions from near-surface, sub-surface and groundwater flow paths (Pilgrim et al 1988, EI-Hames and Richards 1994, Lidén and Harlin 2000, van Loon and van Lanen 2012. Therefore, relatively few hydrological modelling exercises have been carried out in semi-arid mountainous catchments, e.g.…”