A probabilistic adaptation of meteorological model outputs to hydrological forecasting

Guillaume Bontron (1) , Charles Obled (2)

1 Unité de Recherche Great Ice
2 LTHE - Laboratoire d'étude des transferts en hydrologie et environnement

Anticipating flash floods over quick responding catchment such as the Mediterranean ones requires an appropriate anticipation of future rainfalls. We try here to answer this demand through a probabilistic adaptation of meteorological model outputs, based on an analog search for past situations similar to the expected one. The precipitations collected during those past situations allow to derive a conditional distribution for the expected rainfall. After calibration of the adaptation algorithms using the NCEP/NCAR reanalyses, an operational prototype of this analog approach has been implemented at our laboratory, using the outputs of the American model GFS . This prototype considers concurrently two types of analogy: a first one relies only on the general synoptic circulation, while the second involves both the general synoptic circulation and more local variables from the humidity fields. The catastrophic rain event of September 8 and 9th 2002 in the Gard region illustrates the value added by this second type of analogy. In this particular case, it would have allowed to foresee the exceptional character of the rain amounts already on the morning of Septembre 7th.

Domaines: Hydrologie

Guillaume Bontron, Charles Obled. A probabilistic adaptation of meteorological model outputs to hydrological forecasting. La Houille Blanche - Revue internationale de l'eau, 2005, 1, pp.23 à 28. ⟨10.1051/lhb:200501002⟩. ⟨insu-00383135⟩