Stochastic Downscaling to Chaotic Weather Regimes using Spatially Conditioned Gaussian Random Fields with Adaptive Covariance

Abstract Downscaling aims to link the behaviour of atmosphere at fine scales properties measurable coarser scales, and has potential provide high resolution information a lower computational storage cost than numerical simulation alone. This is especially appealing for targeting convective which are edge what possible simulate operationally. Since scale weather degree independence from larger generative approach essential. We here propose statistical method downscaling moist variables using conditional Gaussian random fields, with an application wet bulb temperature (WBPT) data over UK. Our model uses adaptive covariance estimation capture variable spatial scales. further validation, historically been challenge models.