Premium
Parameter identification for conceptual modelling using combined behavioural knowledge
Author(s) -
Dunn Sarah M.,
Soulsby Chris,
Lilly Allan
Publication year - 2003
Publication title -
hydrological processes
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.222
H-Index - 161
eISSN - 1099-1085
pISSN - 0885-6087
DOI - 10.1002/hyp.1127
Subject(s) - identifiability , streamflow , surface runoff , estimation theory , identification (biology) , model parameter , flow (mathematics) , conceptual model , hydrology (agriculture) , environmental science , groundwater recharge , computer science , drainage basin , groundwater , mathematics , data mining , geology , algorithm , aquifer , machine learning , geotechnical engineering , ecology , botany , geometry , cartography , database , biology , geography
Improved methods for identification of conceptual parameter values are necessary if hydrological models are to be applied to catchments other than those to which they have been specifically calibrated. A technique has been devised to calculate conceptual parameter sets of a semi‐distributed hydrological model using a soil hydrological classification in addition to topographic data. The method is tested in this paper by converting multiple parameter sets, calibrated to two separate catchments, into parameter sets for a third catchment. The prediction capabilities of the parameter sets are studied for the new catchment in terms of both simulation of total streamflow and the separation of that flow into three components, corresponding to groundwater recharge, sub‐surface flow, and surface runoff. The results from an end‐member mixing analysis (EMMA) using geochemical tracers are employed to assess the flow separation. Results from the simulations demonstrate that there is quite wide variability in the success of the parameter sets at predicting streamflow for the new catchment. There is also considerable variation in the predicted stream flow separation, with only 28 out of 500 simulations giving a comparable result to the EMMA. By accepting or rejecting simulations using these results, the EMMA can be used to reduce the structural uncertainty of the model. However, it does not help to reduce constraints on acceptable parameter values for the simulations, and further research is still necessary to improve parameter identifiability. Copyright © 2003 John Wiley & Sons, Ltd.