D. Groeneveld, G. Archamboult, F. Kann
Advanced Remote Sensing, Inc.,
United States
Keywords: rapid atmospheric correction, closed form
Summary:
DoD global recon with smallsats flocks can be expected to generate many hundreds of daily satellite images, each requiring automated analysis. To support image analysis, these data must be atmospherically corrected to remove variable aerosol and water vapor effects – cleaner data yields better accuracy. Rapid processing and high accuracy across the correctable range of atmospheric effects are critical for this mission, but existing methods are severely limited. “Closed-form Method for Atmospheric Correction” (CMAC) now TRL 6-7 through NSF SBIR support, employs a closed form conceptual model based on observations in nature, using only scene statistics to correct images in near real-time. CMAC is readily applicable for images from any satellite, cubesat to hyperspectral. Through a next-gen effort, CMAC functionality can be expanded to calibrate any new satellite with only a few overpasses, perform QA/QC for hundreds of satellites on-the-fly, and with the addition of artificial neural network methods, perform sub-minute correction of 100-million-pixel images from all global environments. We propose CMAC upgrades for DoD support and application. Our goal is for CMAC to be sufficiently rapid, accurate, and robust for in-satellite residence in response to DoD’s call for low latency and edge computing capabilities.