Hydrologically-Aided Interpolation (HAI) of Precipitation in Complex Alpine Terrain

Abstract:

Hydrological modeling in mountainous regions requires unbiaised precipitation estimates at scales of a few hundreds to a few thousands square-kilometers (meso-scale). At these scales, precipitation patterns are complex and exhibit orographic enhancement, a phenomenon which is often poorly captured by scarce gage networks. Usually, the estimation of areal precipitation is performed independently of the hydrological modeling step (e.g. using precipitation reanalysis datasets or gage interpolation products). In this approach, it is not possible to easily correct precipitation biases in the case of discrepancies between observed and simulated discharges. In this study, we introduce the concept of Hydrologically-Aided Interpolation (HAI): a gage-based interpolation scheme, producing gridded daily precipitation estimates, is coupled to a semi-distributed hydrological model running at the daily time-step. The parameters of the interpolation scheme (precipitation gradients with elevation) are estimated jointly with the parameters of the hydrological model (snow scheme, soil moisture accounting scheme, and routing scheme). The whole hydrometeorological model is evaluated against cross-validation precipitation gages, point-scale snow water equivalent (SWE) measurements, and catchment-scale discharge estimates at several streamflow gaging sites in a 3,500 square-kilometer Alpine catchment in the French Southern Alps. Results show that adding hydrological constraints leads to much more robust estimates of precipitation gradients, which in turn produce improved precipitation estimates in temporal cross-validation both at point-scale and catchment-scale.