Assessing the Feasibility of an NWP Satellite Data Assimilation System Entirely Based on AI Techniques

Data assimilation (DA) is facing major challenges in terms of its ability of handling the ever-increasing volume of valid, useful, and potentially impactful environmental data and the problem is expected to be exacerbated in the near future if a solution to dramatically increase efficiency is not found. A new approach to perform large-volume data fusion and assimilation, based entirely on artificial intelligence (AI) modern techniques including machine learning and computer vision techniques, is proposed in this study. This approach to DA is applied and demonstrated to real environmental data measured by NOAA-20 and MetOp-C microwave satellite-sounders to reproduce traditional numerical weather prediction DA performances from the U.S. National Oceanic and Atmospheric Administration (NOAA). We assess the impact of our AI-based analysis on forecasts by; 1) performing statistical assessments versus the European Centre for Medium-Range Weather Forecasts analyses, 2) assimilating the AI-based analyzed fields as pseudo-sounding observations in the NOAA global data assimilation system (GDAS), and 3) running forecast experiments using FV3GFS initialized with those observations. To identify the impact of our AI-based assimilations, we compare the forecast skill of several experiments where GDAS is driven with conventional and satellite radiometric observations and with conventional and AI-based pseudo-observations. The results presented are encouraging but are considered only a first initial step toward demonstrating an entirely AI-based environmental data fusion/assimilation system capable to efficiently handle large-volume data and take of the information content available.