Fig. 6: Comparison of dPL and regionalization schemes for streamflow calibration. | Nature Communications

Fig. 6: Comparison of dPL and regionalization schemes for streamflow calibration.

From: From calibration to parameter learning: Harnessing the scaling effects of big data in geoscientific modeling

Fig. 6

a Calibrating the VIC hydrologic model via a differentiable surrogate model on the CAMELS dataset over the conterminous United States, in comparison to Multiscale Parameter Regionalization (MPR). b Calibrating the HBV hydrologic model (not a surrogate) on the Beck20 global dataset, in comparison to the Beck20 regionalization scheme. We used NSE for the CAMELS case and KGE for the global case because these metrics were used by the respective papers, and the main purpose of the case studies was to compare with the existing literature. For both panels, curves on the right represent better models.

Back to article page