GEO-GEO Radiance Inter-Calibration of INSAT-3D with MSG-SEVIRI and Total Ozone Retrieval using Machine Learning

Abstract

Atmospheric ozone has a unique vertical distribution among various trace gases and plays different roles at different altitudes. The useful stratosphere ozone absorbs harmful UV radiations, while tropospheric ozone, a powerful greenhouse gas, adversely affects living beings and vegetation. The continuous monitoring of the total ozone column (TOC) is necessary to understand the various large-scale dynamics, recovery of total ozone over different regions, and accurately estimate the incoming UV flux to the earth's surface. Nowadays, ozone monitoring via satellite-based remote sensing has gained wide importance. The Indian geostationary satellite INSAT-3D is accomplishing this need for India's tropical region. The poor understanding of dominant circulations and limited ground-based observations over tropical regions, like India, ozone recovery, and trend studies remain complex and uncertain, where INSAT-3D has enormous capabilities. However, INSAT-3D has underestimated the total ozone column by more than 22 DU, compared with ozonesonde and OMI. The large TOC bias could be attributed to lower reliability of retrieval and INSAT-3D radiance biases. To mitigate these differences, the radiance biases in INSAT-3D observations are calibrated using the GEO-GEO methodology with MSG-SEVIRI for collocated pixels. It shows biases in INSAT-3D by more than 5 K. These biases of INSAT-3D are corrected to better retrieve the total ozone over the Indian region from INSAT-3D. A well-trained machine learning (ML) model is then used for TOC retrieval, which retrieved TOC effectively and matches well with SEVIRI.