Spatial Gap-filling of GK-2A/AMI AOD products for Estimation of Particulate Matter using Machine Learning

Abstract

Particulate Matter (PM) directly or indirectly affects climate change by changing the radiative forcing of sunlight. This is known to be harmful to the human body and affects industrial activities. In order to prevent damage to the health environment, society, and economy as a whole due to the increase in PM concentration, it is important to secure regional accurate PM concentration calculation and monitoring technology for it. However, the current PM observation network consisting of ground and point observation shows many limitations in time and space. As an alternative to this, studies that obtain PM concentration using satellite observation are actively underway [1]. As the optical characteristics of PM can be measured and the polar orbit and geostationary satellites, which are environmental satellite payloads, become more diverse, it will be more promising. Nevertheless, there is a problem here, too. The optical sensor-based Aerosol Optical Depth (AOD) images have missing parts due to clouds, etc., which make it difficult to analyze the PM variation. Therefore, this study aims to spatial gap-filling of the GK-2A (Geostationary Korea Multi-purpose Satellite 2A)/AMI (Advanced Meteorological Imager) AOD images using the meteorological data and random forest model. The spatial area of study is the Korean Peninsula, where long-distance transportation PM from neighboring Asian countries such as China, Mongolia, and Russia occurs at high concentrations every year. The experiment was conducted on 8 timeslots between 00 UTC and 07 UTC during daytime among GK-2A/AMI AOD data in 2021. AMI