Fig. 1: Comparing power spectra and distributions.
From: Residual corrective diffusion modeling for km-scale atmospheric downscaling
The power spectra and distributions are compared for an interpolated ERA5 input, CorrDiff, RF, UNet, and target (WRF). These results reflect reductions over space, time and for CorrDiff across 32 different samples per each time. Left: power spectra for kinetic energy (a), 2-m temperature (b) and radar reflectivity (c). Right: distributions of 10-m windspeed, (d), 2-m temperature (e) and radar reflectivity (f). Radar reflectivity is not included in the ERA5 dataset. We show the log-PDF to highlight the differences at the tails of the distributions. Here wavenumber is the inverse of a wavelength.
