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

U.S. Naval operations rely upon accurate, high-resolution observations of Earth’s marine atmospheric boundary layer (MABL) environmental conditions for communications, C4ISR, operational planning (e.g., flight and small boat ops, personnel insertion, visibility, etc.), and directed energy weapon systems. However, there is a lack of sufficient data across broad geographic areas of interest to resolve detailed MABL vertical structure to accurately analyze and predict tactical scale environmental conditions. Limitations of current satellite observations of MABL thermodynamic properties include refractivity, insufficient spatial coverage, resolution (horizontal and vertical), and/or temporal refresh. To innovate and minimize the negative impact of these limitations of satellite observations on the quality of forecast and environmental condition products, the development of data fusion products that combine observations of broader environmental data, especially newly available data from smallsat platforms with other remotely sensed data encompassing multiple observed mediums, is necessary. The proposed effort will improve the fidelity of tactical and operationally relevant atmospheric forecasting and contemporaneous environmental condition products by developing novel multi-stream satellite-based environmental data assimilation algorithms for the retrieval of thermodynamic properties of the marine atmospheric boundary layer for incorporation into numerical weather prediction models alongside the assimilation of traditional measurements. This interdisciplinary, multi-organizational Phase I effort will develop novel machine learning (ML) informed methods for assimilating satellite-based environmental monitoring data (i.e., Radio Occultation from geospatial satellites, with microwave and infrared satellite sensors) with traditional methods to improve retrievals of thermodynamic profiles of the lowest 1-2 km of the troposphere. ML techniques add flexibility and efficiency in learning relationships among features of datasets, while reducing the cost and time of creating novel products. The novel algorithm and data fusion product will initially be developed and validated over the tactically important and sparsely observed Arctic. The Weather Research and Forecasting (WRF) Model will be used to run retrospective forecasts to evaluate integration of the data fusion product. Environmental profiles and ducting conditions will both be considered during the data integration evaluation process. These model runs comparing single- and multi-stream data products will aid in identifying factors contributing most to enhanced understanding of the MABL resulting from the novel data fusion product constructed during these efforts. During the Option Period we will apply the method to hindcasting multiple Naval exercises and one USCG icebreaker deployment in the arctic, validating predictions with measurements, and discussing operational impact of improved forecasting skill.