The AIGA method: An operational method using radar rainfall for flood warning in the south of France

International audience; Flash flood forecasting is one of the most difficult tasks in operational hydrology: by definition, affected catchments have a very short time response and, very often, no data are available to calibrate the models. In Europe, as mentioned by Gaume et al. (2009), particularly Mediterranean regions are subject to such events, but other inner continental countries can be affected as well, with dramatic consequences. Thus, there is a real need for operational tools, which make it possible to better anticipate this kind of event. The classical approaches for flood forecasting generally combine conceptual rainfall–runoff models that generate upstream discharges and hydraulic models that propagate this information through the monitored river network (Moore et al., 2005; Rabuffetti & Barbero, 2005). In such systems, real-time streamflow information can be assimilated in order to improve the forecasts (Berthet et al., 2009). But this approach cannot be adapted to small ungauged catchments located outside the monitored network. The flash flood guidance (FFG) method, used routinely in the USA, aims at providing flood warnings at ungauged locations. However, as mentioned by Reed et al. (2007), this method is based on a lumped model (the Sacramento model), using parameters transferred from bigger gauged catchments, and therefore can lead to scale problems. As shown by Reed et al. (2007), Blöschl et al. (2008) and Cole & Moore (2009), simple distributed models can now be used at an operational level, and provide realistic forecasts at ungauged locations. In this context, the aim of this paper is to present a flood warning system currently used in the south of France for ungauged catchments. The case study analysed concerns a dramatic event (causing 25 casualties) which affected the region of Draguignan on 15 June 2010. This approach, called AIGA, combines radar rainfall and a simple distributed hydrological model taking into account antecedent soil moisture conditions. The next section describes the method used in the AIGA approach. Then a case study application is presented followed by a few concluding remarks concerning future needs.