Evaluation of streamflow forecasting methods

TitleEvaluation of streamflow forecasting methods
Publication TypeConference Proceedings
Year of Conference1979
AuthorsLettenmaier, D. P., and Garen D. C.
Conference Name47th Annual Western Snow Conference
Series TitleProceedings of the 47th Annual Western Snow Conference
Date PublishedApril 1979
PublisherWestern Snow Conference
Conference LocationSparks, Nevada
KeywordsComparative forecasting methods, Digital models, Water supply forecasting

Choice of a seasonal snowmelt runoff forecast model represents a tradeoff between data availability and the level of model complexity which can be justified. The basin storage accounting method, based on a storage estimate derived as the difference between accumulated low altitude precipitation and winter season runoff, with a correction for snow water equivalent observations represents a fairly accurate, although simple approach to forecasting seasonal runoff. The worth of precipitation and snow course data during the winter snow accumulation season, and the worth of forecasts of runoff forecast period precipitation may be evaluated in a relatively straightforward manner for such models. The evaluation is made directly through comparison of split sample seasonal forecast accuracy as a function of the number of precipitation stations and snow courses used. Precipitation forecast accuracy worth is assessed through comparison of split sample runoff forecast errors using long term average precipitation and recorded precipitation in the runoff forecast season, representing the range from no precipitation forecast to perfect precipitation forecast accuracy.