Jump to Main Content
Can a regionalized model parameterisation be improved with a limited number of runoff measurements?
- Viviroli, Daniel, Seibert, Jan
- Journal of hydrology 2015 v.529 pp. 49-61
- basins, hydrograph, hydrologic models, ice, melting, prediction, runoff, snowmelt, spring, summer, watersheds, Switzerland
- Application of hydrological models to ungauged basins is both a highly relevant and challenging task. While research has brought forth various approaches for inferring or transferring tuneable model parameters from gauged and calibrated catchments, it has also been recently shown that a few short measurements can support predictions in an ungauged basin by constraining the acceptable range of the parameters. For the present study, we examined a combination of both parameter regionalisation and short-term runoff measurements. More precisely, we attempted to select complete parameter sets from a range of calibrated catchments using a few measurements. Then, we tested a number of ways to combine the hydrographs simulated with these parameter sets with those simulated using a well-established Nearest Neighbour scheme, in order to make use of both actually measured runoff data as well as hydrological similarity. The experimental basis for our study were 49 representative catchments in Switzerland which have been successfully calibrated and regionalised with the hydrological modelling system PREVAH. Results show that even a few short measurements during mean runoff conditions can lead to models that are more efficient than those achieved with hydrological similarity alone. The possible improvement depends largely on the regime type of the catchment examined. Also, the most suitable season to perform measurements varies: In catchments dominated by snow melt or ice melt or both, considerable improvements can be achieved with as few as two measurements during spring or summer, whereas rainfall-dominated catchments show only moderate improvements with no particular season being more suitable for the measurements. Our findings highlight the value of field measurements in mountain areas. The information gained in these regions from short measurements may act as a counterbalance to the sparse operational observation networks.