Development of regionalisation procedures using a multi-model approach for flow simulation in an ungauged catchment

Flow simulation in ungauged catchments is presently regarded as one of the most challenging tasks in surface water hydrology. Many of the ungauged catchments are located in the headwaters of rivers in mountainous regions of the world having enormous potential for sustainable water resource development. However, due to inaccessibility, rugged and inhospitable terrain, and historical lack of foresight concerning the need to have these headwaters adequately gauged, their potential is not readily realizable. Many downstream sites also suffer from non-availability of site-specific data as even in countries having extensive networks of gauged stations data may not be available at sites where these are most needed. As predictive tools for water resources, water quality, natural hazard mitigation and water availability assessment are generally data-driven, the lack of adequate hydrometric records poses difficult problems for planners, engineers, managers, and stake-holders alike. In this study, a methodology is developed for flow simulation in ungauged catchments using a regionalisation and multi-model approach involving a suite of rainfall-runoff models and combination techniques. Daily observed hydrometeorological data for 12 French catchments are used for illustrating the procedures. Following a preliminary investigation of the regional homogeneity of that group of catchments, three regional flow simulation techniques are applied. Although all 12 catchments are gauged, initially each catchment is successively considered as being ungauged for the purpose of flow simulation in that catchment, their actual discharges being subsequently used for evaluating the performance of the flow estimation procedures for the catchment. The Nash-Sutcliffe efficiency index (R^2) is used for assessing and ranking the relative performances of the regionalisation-model couples to identify the most appropriate couple for the region. The final step of applying that couple to a truly ungauged (13th) catchment in the region is described. Results are presented and conclusions drawn on the efficacy of the regional-multi-model approach. Of the couples considered, the pooling method of regionalisation coupled with the conceptual soil moisture accounting and routing (SMAR) model is deemed to be the best for simulating flow in an ungauged catchment in the region.