Marine Atmospheric Boundary Layer Profiles via Satellite-based Remote Sensing Data Fusion

Satellite-based atmospheric soundings are extremely valuable in remote regions lacking nearby radiosondes. They provide a depiction of temperature and water vapor structure to improve prediction of clouds, storms, and rainfall, and massively contribute to numerical weather prediction (NWP) via data assimilation. However, satellite profiles are too smooth and under-determined, thereby suffering from large biases, low vertical resolution, and other random errors, making them less useful for depicting important phenomena in the marine atmospheric boundary layer (MABL) that can aid weather forecasters and at-sea Naval operations. To address the need to improve at-sea Naval operations, the University of Alabama in Huntsville (UAH) and Areté propose the development of an artificial intelligence (AI) framework that performs data fusion of single-source satellite profile retrievals from multiple modalities and sensors, NWP forecasts, and satellite-derived sea surface temperature and winds to produce temperature and water vapor profiles of the MABL at 100 m vertical resolution. Phase I technical objectives include: 1) a regionalized proof-of-concept demonstration in deriving 100 m vertical resolution thermodynamic profiles within the MABL using AI, 2) evaluating single-source profile retrieval biases and overall importance in contributing to enhanced understanding of the MABL, and 3) benchmarking AI profiles against state-of-the-art profile retrievals.