Setting expectations for hydrologic model performance with an ensemble of simple benchmarks
Wouter Knoben
Abstract
Example of benchmark inputs and results for a snow-dominated basin, subset to a 4-year period on either side of the calculation (left) and evaluation (right) divide (dotted red line). KGE scores in legends are calculated for the evaluation period. (a) Observed streamflow, precipitation and a ‘rain plus melt’ flux (RPM) derived from precipitation and temperature. RPM is used to define the benchmarks shown in c and d. (b) Flow-only benchmarks. The straight light green line is the traditional (NSE = 0; KGE = 1-√2) mean flow benchmark. (c) Rainfall-runoff ratio benchmarks. A single rainfall-runoff ratio is derived from the data in the calculation period and used to scale annual and monthly RPM sums into flow benchmarks. (d) Simple models that represent catchment function.