A streamflow assimilation system for ensemble streamflow forecast over France G. Thirel (1), E. Martin (1), J.-F. Mahfouf (1), S. Massart (2), S. Ricci (2) and F. Habets (3) (1) CNRM-GAME, Météo-France, CNRS, GMME, France, (2) CERFACS, France, (3) UMR SISYPHE, UPMC, ENSMP, CNRS, Paris, France +33 (0) ) SIM is a distributed hydrometeorological system composed of a meteorological analysis (SAFRAN), a land surface model (ISBA) and a hydrogeological model (MODCOU). SIM simulates the evolution of water and energy budget at the surface, the soil wetness and the discharge at 900 stations over France. In order to produce better Ensemble Streamflow forecasts, an assimilation system is developed at Météo-France. Its aim is to provide initial data for prognostic variables closer to observations. The data assimilation system is developed with a modular software (PALM, from the European Center for Research and Advanced Training in Scientific Computation, CERFACS, France). Analysis state (adjusted ISBA soil moisture [D- N]) Background state (ISBA soil moisture [D-N]) Model equivalent of the streamflows [D- N;D-1]) Assimilation (BLUE) Observed streamflows [D-N;D-1] Updated SIM streamflows [D-1] +ISBA soil moisture [D-1] final states SIM run R, B and H matrices SIM run Hydrological ensemble forecasts [D;D+P] ESPS runs Innovation vector D-ND+PD-1time Methodology of the assimilation system The Best Linear Unbiased Estimator equation xb is the estimated soil water content, and xa is the new analysed soil water content value. H (Jacobian matrix) is computed with small perturbations around the xb state (linear assumption). R and B are diagonal and proportional to, respectively, the observed streamflows squared errors and the a priori estimated soil water content squared errors. y0 are the observed streamflows, and H(xb) the model streamflow corresponding to the estimated soil water content. Definition of the six different experiments of assimilation Assimilation of observed streamflow over a long period (10 March 2005 / 30 September 2006) – Results for 6 different configurations, for a selection of 148 relevant stations Definition of the 6 experiments Nash criteria, rmse score, and mean of the absolute values of the BLUE adjustments for the 6 experiments. Nash criteria and rmse score are also provided for a corresponding reference simulation (i.e. without assimilation). Results of the assimilation system for the IS2 experiment, for the Doubs river at Besançon (10 March 2005 – 25 September 2005). (a) : Streamflows (m^3/s). (b) : Soil moisture (m^3/m^3) for the Doubs river basin irrigating the Besançon gauge station (averaged on the considered sub-basin meshes). Soil water features for this basin are represented in dashed lines (w_sat, w_fc and w_wilt).(c) : The crosses represent the daily adjustments applied by the BLUE to the soil moisture. A value for the adjustment of 1.05 means that 5% of the soil moisture is added by the BLUE for the given assimilation. Reference : A streamflow assimilation system for ensemble streamflow forecast over France, Thirel et al., submitted to the Journal of Hydrology, Conclusions and perspectives The 148 assimilated stations and the ISBA meshes irrigating them SIM was never re-initialized with observed streamflows when running before this assimilation system. This poster shows the methodology used to improve the simulated streamflows for 6 different configurations, and some results. Daily observed streamflows from a maximum of 186 gauge stations are assimilated over France, in order to improve the streamflow simulation, by adjusting the initial ISBA soil moisture. Simple basins and embedded basins networks are simultaneously assimilated. This study encompasses the performance of the assimilation for three different state variables. For each of them, the use or not of an improvement of the ISBA physics (the use of the exponential profile of hydraulic conductivity) is tested. The assimilation system is performed every day, over a 1-day assimilation window. -IS3 and IS4 lower performance : runoff is not the only process to improve for a better streamflow simulation -Best Nash scores for IS1, IS2 and IS5. -Lower rmse score and mean of the absolute values of adjustments of BLUE for IS2 : this is the best configuration! -Choosing an experiment with the best physics (i.e. with the exp. profile) could ensure a better persistance of the effects of a better initial state for the ensemble streamflow forecasts. -Higher skill for high floods than for low flows. This study has shown the feasibility and the good overall performance of an assimilation system for improving the streamflow simulation of a distributed hydrometeorological model over France. The strategy used can encompass embedded stations networks, and improve the simulation of high floods. The use of the IS2 experiment will be assessed in a following work for the initialization of short- and medium-range ensemble streamflow forecasts over France. A project for assimilating piezometric levels for the Seine and the Rhône rivers could be launched in the future. Indeed, these basins are more influenced by the aquifer layers than by the soil moisture, so that the current assimilation system has a low effect. ISBA fluxes Principle of the assimilation process for an assimilation done over a N-day time window, initializing an ensemble streamflow forecast long of P days