Deep learning for bias-correcting CMIP6-class Earth system models
This addresses the need for more reliable precipitation projections in climate modeling, which is crucial for ecological and socioeconomic impact assessments, representing an incremental advance over existing bias correction techniques.
The authors tackled the problem of correcting biases in Earth system model precipitation outputs, showing that a physically constrained generative adversarial network (cGAN) method simultaneously improves local frequency distributions and spatial patterns, strongly outperforming existing methods in correcting spatial intermittency of extremes.
The accurate representation of precipitation in Earth system models (ESMs) is crucial for reliable projections of the ecological and socioeconomic impacts in response to anthropogenic global warming. The complex cross-scale interactions of processes that produce precipitation are challenging to model, however, inducing potentially strong biases in ESM fields, especially regarding extremes. State-of-the-art bias correction methods only address errors in the simulated frequency distributions locally at every individual grid cell. Improving unrealistic spatial patterns of the ESM output, which would require spatial context, has not been possible so far. Here, we show that a post-processing method based on physically constrained generative adversarial networks (cGANs) can correct biases of a state-of-the-art, CMIP6-class ESM both in local frequency distributions and in the spatial patterns at once. While our method improves local frequency distributions equally well as gold-standard bias-adjustment frameworks, it strongly outperforms any existing methods in the correction of spatial patterns, especially in terms of the characteristic spatial intermittency of precipitation extremes.