Assimilation of soil moisture into hydrological models:: the sequential method

Improving the accuracy of rainfall-runoff models, and in particular their performance in flood prediction, is a key point of continental hydrology. This paper presents a new approach to these problems by the use of soil moisture data (remote sensing or in situ measurements). A first step is to correct past modeling errors, especially errors in simulated runoff, to obtain a more accurate forecast of runoff. Other significant parameters are the interactions occurring within the soil-vegetation-atmosphere interface, which are dominating factors in the processes of the transformation of rainfall into flow at a catchment-area scale. These phenomena can be integrated in hydrological modeling by introducing soil moisture measurements and thus explicitly taking into account the hydric state of the soil. The hydrological models used are global conceptual models. The methodology that we used is a sequential assimilation procedure, which permits step by step control of the evolution of the model and limits its divergence from the available data (soil moisture and flow). The efficiency of the assimilation procedure in flood prediction is discussed, with a particular focus on the contribution of soil moisture data. The question of the time repetitivity of measurements and its influence on the performance of the modeling is also tackled.