Disaggregating the impact of hillslope changes by geomorphology-based deconvolution of outlet observation
Résumé
The use of geomorphology-based transfer functions is a strong perspective of low-calibrated and structurally explicit rainfall-runoff. In this study, the whole basin rainfall-runoff process is considered as the coupling of hillslope and stream network processes. The coupling variable -net rainfall- along the hillslope-river interface cannot be monitored in situ. A robust geomorphology-based transfer function through the river network is proposed, as well as a methodology for its inversion. Discharge at the catchment outlet is deconvoluted thanks to the geomorphology-based transfer function inversion, which allows to assess average net rainfall. The application of this net rainfall assessment method for two distinct periods, before and after a major hillslope change, for a given catchment allows to deglobalise the hydrological impact from the catchment outlet towards the hillslope foot. In semi-arid regions, erosion is a major problem. Contour ridges are a widely and intensely developed water and soil conservation system at the hillslope scale. The Central Tunisian semi-arid catchment of El Gouazine (18.1 km²) was intensely and widely changed by the sudden implementation of contour ridges during summers 1997 and 1998. Hydrological impacts have been assessed at the catchment outlet in terms of global runoff coefficient, total volumes and maximum discharge. The application of the deconvolution and deglobalisation geomorphology-based approach to pre-1997 and post-1998 runoff events, induced by similar rainfall events, show various significant changes in the hillslope-scale rainfall-net rainfall process which were not appearing in the basin-level rainfall-runoff process. The major changes are the emergence of a time delay (230 min) and of a generation threshold (16 mm) of runoff. This deglobalising approach proves to be a promising impact assessment method from downstream, as an alternative to upscaling and aggregating hillslope approaches, in a catchment where a major hillslope sudden change occurred. It is based on a robust and generic geomorphologybased transfer, which is assumed to be stationary through time since changes occurred on hillslope and not in streams. Quantitative perspectives are open to consider other types of sudden changes (e.g. wildfire), as well as progressive changes (e.g. ageing of landscape artefacts, including of contour ridges themselves). Furthermore, such an assessment of the hillslope-river network coupling variable will allow to develop a library of production functions to be coupled with the geomorphology-based transfer function, according to hillslope types.