Performance of generalized Gaspari–Cohn localization in complex multiscale, heterogeneous, and anisotropic regimes

Evaluate how generalized Gaspari–Cohn localization performs in complex multiscale settings that are highly heterogeneous and anisotropic, such as extreme weather events that are not represented by the test problems considered.

Background

GenGC localization introduces spatial inhomogeneity and anisotropy via spatially varying hyperparameters and showed slight improvements in some of the paper’s experiments. However, the test problems were not designed to capture the full complexity of multiscale, highly heterogeneous and anisotropic phenomena such as extreme weather events.

The authors explicitly state that it remains an open question how GenGC would fare in these more complex regimes and leave this for future work.