Hydrologic Modelling: an accurate description of all, extremely complex, basic mechanisms involved in the three-dimensional space of a catchment remains impossible today. Only simplified solutions seem feasible.
Appropriate modelling choices, reasonable parametrisation
In our research, an original and pragmatic approach is adopted, giving up the detailed description of various processes.
A logical approach reproducing at best the overall behaviour of a catchment under the combined action of two key forcing variables (precipitation and evaporative demand) was built, by progressively testing the effectiveness of each new contribution.
The approach is based on the following elements.
- The catchment is seen as a unique unit (lumped modelling approach).
- The approach is empirical (no a priori use of physical laws).
- Model structures are developed starting from the most simple ones and progressively adding complexity.
- The level of structural complexity of a model depends on its ability to reproduce the rainfall-runoff transformation.
- Model structures must be a general as possible (applicable over a wide range of catchments and conditions).
- The calibration-validation test, carried out over a large set of catchments, is the standard procedure for model evaluation.
- Models are evaluated in relative terms (there are no “good” models, only models that are better than others).
Sensitivity analyses of hydrological models are at the heart of several of our studies:
- sensitivity to errors in rainfall input data and rainfall spatial resolution (Andréassian et al., 2001; Oudin et al., 2006),
- sensitivity to potential evapotranspirtion (Andréassian et al., 2004; Oudin, 2004; Oudin et al., 2004; Oudin et al., 2005a,b,c; Oudin et al., 2006),
- sensitivity to spatial distribution in modelling (Baudez, 1997; Baudez et al., 1999; Andréassian et al., 2004; Bourqui, 2008),
- sensitivity to rainfall and discharge data availability for model calibration (Rojas-Serna, 2005; Rojas-Serna et al., 2006; Perrin et al., 2007; PhD thesis L. Lebecherel, 2015),
- sensitivity to calibration procedures (Oudin et al., 2006; Perrin et al., 2008),
- sensitivity to snow accumulation and melt module (Valéry, 2010 ; Riboust, 2018).
Our current research works in hydrologic modelling also aim to :
- improve calibration procedures (Coron et al., 2012),
- integrate new data sets: weather radar data, soil moisture content, satellite estimates of precipitation, etc. (PhD Thesis F. Lobligeois, 2014),
- quantify uncertainties in rainfall-runoff modelling (PhD Thesis F. Bourgin, 2014).
Applied research methods are developed to meet the needs of operational hydrology in several practical applications:
- statistical prediction of floods,
- flood and drought forecasting,
- ungauged catchment modelling,
- trend detection in hydrologic time series (evolution of the hydrologic catchment behaviour) and impacts of climate and environmental changes,
- design of structures,
- design and management of single-purpose or multi-purpose reservoirs (flood control, low flow replenishment, etc.).