Research papersTesting the generalized complementary relationship of evaporation with continental-scale long-term water-balance data

•The generalized complementary relationship of evaporation is revised.•The revised version responds better to sudden changes in moisture availability.•The revised version yields fewer over- and underestimations.

The original and revised versions of the generalized complementary relationship (GCR) of evaporation (ET) were tested with six-digit Hydrologic Unit Code (HUC6) level long-term (1981-2010) water-balance data (sample size of 334). The two versions of the GCR were calibrated with Parameter-Elevation Regressions on Independent Slopes Model (PRISM) mean annual precipitation (P) data and validated against water-balance ET (ETwb) as the difference of mean annual HUC6-averaged P and United States Geological Survey HUC6 runoff (Q) rates. The original GCR overestimates P in about 18% of the PRISM grid points covering the contiguous United States in contrast with 12% of the revised version. With HUC6-averaged data the original version has a bias of −25 mm yr−1 vs the revised version's −17 mm yr−1, and it tends to more significantly underestimate ETwb at high values than the revised one (slope of the best fit line is 0.78 vs 0.91). At the same time it slightly outperforms the revised version in terms of the linear correlation coefficient (0.94 vs 0.93) and the root-mean-square error (90 vs 92 mm yr−1).