Artificial Intelligence (AI) weather models are explored for initial condition sensitivity studies to analyze the physicality of the relationships learned. Gradients (or sensitivities) of the target metric of interest are computed with respect to the variable fields at initial time by means of the backpropagation algorithm, which does not assume linear perturbation growth. Here, sensitivities from an AI model at 36-h lead time were compared to those produced by an adjoint of a dynamical model for an extreme weather event, cyclone Xynthia, presenting very similar structures and with the evolved perturbations leading to similar impacts. This demonstrates the ability of the AI weather model to learn physically meaningful spatio-temporal links between atmospheric processes. These findings should enable researchers to conduct initial condition studies in minutes, potentially at lead times into the non-linear regime (typically >5 days), with important applications in observing network design and the study of atmospheric dynamics.

Atmospheric rivers (ARs) are narrow corridors of intense water vapor transport, shaping precipitation, floods, and economies. Temporal clustering of ARs tripled losses compared to isolated events, yet the reasons behind this clustering remain unclear. AR orientation further modulates hydrological impacts through terrain interaction. Here we identify unique ARs over the North Pacific and Western U.S. and utilize Cox regression and composite analysis to examine how six major climate modes influence temporal clustering of unique ARs and orientation during extended boreal winter (November to March). Results show that climate modes condition temporal clustering of unique ARs. The Pacific-North American weather pattern strongly modulates the clustering over the Western U.S. from early to late winter. The quasi-biennial oscillation and Pacific decadal oscillation affect late winter clustering, while the Arctic oscillation dominates early winter. Climate modes also strongly influence AR orientation, with ENSO particularly affecting the orientation of temporally clustered ARs.