Development of a hydrological distributed model in real-time flood forecasting

Distributed conceptual modelling is an excellent tool to simulate rainfall-runoff processes during flood events. Basically, it takes into account the spatial and temporal variability included in different hydrological processes.; TETIS model is a hydrological conceptual distributed model which was reviewed and adapted to increase its predictive capacity. The proposed real-time flood forecasting scheme is named TETIS-PTR and can be described in two parts.; The first module selects a period before the initial forecasting time; this period must be long enough to produce a good hydrological simulation using real-time data. This module has been called "initial moisture adjustment". The optimal initial storage set is found using the SCE-UA procedure. If differences between observed and simulated flows still remain, some virtual rain stations must be included to improve the simulated flow.; The second part is the real-time flood forecasting model itself. Future estimation of precipitation is performed using Quantitative Precipitation Forecasting and its uncertainty. Therefore, it is feasible to build a representative statistical sample with future hyetograms. Then, applying TETIS to these hyetograms with the initial moisture adjustment previously estimated, it is easy to obtain different responses, which can go into a statistical analysis to obtain the quantiles of 5%, 50% and 95%.; The real-time forecasting model TETIS-PTR has been applied in different basins, with range areas between 70 km2 to 1857 km2 . These basins belong to Mediterranean and Continental climates. All studied cases produced satisfactory results.; As a conclusion, the initial moisture adjustment allows to obtain an initial state at the cell storages as close to reality as possible. But special treatment is required in Mediterranean regions where flash flood are common and a previous knowledge of initial state variables is required to guarantee a good real-time forecasting. The use of virtual rain stations gets better simulated flows during the period before forecasting and therefore the real-time flood forecasting is improved. The location of rain gauge stations are really important capturing the spatial behaviour of storms, where a good density of rain gauge or radar information allow to obtain reliable results during flood forecasting.