Large Sample Basin Experiments for Hydrological Model Parameterization: Results of the Model Parameter Experiment–MOPEX. IAHS Publ. 307, 2006, 237–245.
Characterization of flow regimes for three French rivers
using a 6-parameter continuous simulation rainfall–streamflow model: the
importance of the parameter selection procedure
Ian G. Littlewood
Centre for Ecology and Hydrology, Wallingford, Oxfordshire OX10 8BB, UK
igl@ceh.ac.uk
Abstract An established 6-parameter rainfall–streamflow modelling package
(PC-IHACRES) is applied to three French catchments using daily data sets made
available to participants of a Model Parameter Estimation Experiment (MOPEX)
workshop held in Paris, 2004. For one catchment, Le Loupe at Villeneuve-Loubet,
the main aim is to characterize the flow regime over as wide a range of flows
as possible, using a single set of model parameters selected by a
multi-objective model parameter selection procedure. Using a more restricted, twin-objective,
parameter selection procedure, a subsidiary aim for each of the three
catchments (additionally Le Guillec at Trézilidé and Le Toulourenc at
Malaucène), is to investigate how well a model calibrated on a sub-period of
the available record performs on a different sub-period not used for model
calibration. For Le Loupe at Villeneuve-Loubet, the 6-parameter model
structure, with a single set of parameters calibrated over almost the entire
available record, characterizes the flow regime between 5 percentile and 80
percentile flows. Split-sample analyses for Le Guillec at Trézilidé and Le
Loupe at Villeneuve-Loubet, using two sub-periods of the available record in
each case, demonstrate the difficulty of capturing a time-invariant characteristic
response for those two catchments. A similar split-sample analysis suggests
that Le Toulourenc at Malaucène has an essentially time-invariant characteristic
response to rainfall. The results are discussed in the context of similar
analyses reported in the literature, and with respect to the potential of the
6-parameter model for some aspects of regionalizing model parameters against
physical catchment descriptors to enable flow simulation at ungauged sites.
Key words
flow regime characterization; rainfall–streamflow modelling;
regionalization; ungauged catchments