Fig. 6
From: Limitation of super-resolution machine learning approach to precipitation downscaling
MLPerfect model response diagnostics (maximum response (a) and number of responses (e)) when perturbed with 0.5 mm/h input at a particular grid point and made rest all grid points as zero; and in the same way iteratively executed at all grid points. MLPerfect model response diagnostics when perturbed with 1 mm/h input are shown in subplots (b; maximum response) and (f; number of responses). MLImperfect model response diagnostics when perturbed with 0.5 and 1 mm/h input are shown in subplots (c; maximum response), (g; number of responses) and (d; maximum response), (h; number of responses), respectively. For more details about model response diagnostics refer to data and methods section. Maps are drawn using the Python Cartopy package (v0.24.1).
