Hybrid Physics–Machine Learning Retrieval of AOD550 from HYPSO Hyperspectral Imagery

Abstract:

Atmospheric aerosols are a major component of environmental risk because they influence air quality, visibility, radiative transfer, and, indirectly, human health and climate processes. For this reason, aerosol optical depth at 550 nm (AOD550) is widely used as a compact indicator of columnar aerosol loading in satellite-based environmental monitoring, climate analysis, and air pollution assessment[1]. In the broader context of processing satellite data for threat assessments, aerosol retrieval is particularly relevant because elevated aerosol load may accompany or indicate hazardous conditions such as wildfire smoke transport, dust outbreaks, industrial pollution episodes, and regional haze events. Reliable and timely AOD550 estimates are therefore valuable not only as atmospheric products, but also as analytical inputs for threat-oriented remote sensing

[