Presentation | IN41D: Connecting Data to Science and Discovery: Innovations and Infrastructure Bridging Disparate Observations to Drive Earth Science I Poster
Poster
IN41D-0385: Neural Operators for High Resolution, Topography Aware Wind Downscaling
Author(s): Ryan DeMilt, Spatial Informatics Group, LLC (First Author, Presenting Author) Nicholas LaHaye, Jet Propulsion Laboratory, California Institute of Technology
Downscaled climate projections, particularly of wind speed and direction, are highly valuable to local and regional planners. Despite many advances in the work of machine learning based climate variable downscaling, the impacts of multivariate data and topography variables have not kept pace with recent innovations in architecture. Our work combines the latest in downscaling architectures with a multivariate and topographic focus to address the complex relationships of wind with local topography.
