Seibert, Jan
- Department of Aquatic Sciences and Assessment, Swedish University of Agricultural Sciences
- University of Zürich
Research article2019Peer reviewedOpen access
Pool S, Viviroli D, Seibert J
Even in regions considered as densely monitored, most catchments are actually ungauged. Prediction of discharge in ungauged catchments commonly relies on parameter regionalization. While ungauged catchments lack continuous discharge time series, a limited number of observations could still be collected within short field campaigns. Here we analyze the value of such observations for improving parameter regionalization in otherwise ungauged catchments. More specifically, we propose an ensemble modeling approach, where discharge predictions from regionalization with multiple donor catchments are weighted based on the fit between predicted and observed discharge on the dates of the available observations. It was assumed that a total of 3 to 24 observations from a single hydrological year were available as an additional source of information for regionalization. This informed regionalization approach was tested with discharge observations from 10 different hydrological years in a leave-one-out cross validation scheme on 579 catchments in the United States using the HBV runoff model. Discharge observations helped to improve the regionalization in up to 94% of the study catchments in 8 out of 10 discharge sampling years. Sampling years characterized by exceptionally high peak discharge, or high annual or winter precipitation were less informative for regionalization. In the least informative years, model efficiency increased with an increasing number of observations. In contrast, in the most informative sampling year, 3 discharge observations provided as much information for regionalization as 24 discharge observations. Overall, discharge observations were most effective in informing regionalization in arid catchments, snow-dominated catchments, and winter-precipitation-dominated catchments.
ungauged basin; regionalization; spatial proximity; attribute similarity; value of data
Water Resources Research
2019, Volume: 55, number: 1, pages: 363--377 Publisher: American Geophysical Union ({AGU})
Oceanography, Hydrology, Water Resources
DOI: https://doi.org/10.1029/2018WR023855
https://res.slu.se/id/publ/98711