Retrieval of ozone profiles from geostationary infrared sounder observations using principal component analysis

Abstract

The present work proposes a novel method to derive vertical profiles of ozone from infrared measurements on board geostationary sounders. The methodology is based on Principal Component Analysis (PCA) and has been applied over a large geographical region from 45°S to 45°N. It has been found that the first two principal components (PCs) together explain 71% of the total variance of the ozone profiles and are statistically significant and quite stable. Therefore, an algorithm for the retrieval of the first two PCs has been proposed. The magnitude of the retrieval error using PCA is significantly smaller than the variance of the dataset in the atmospheric layer from 200.0 to 5.0 hPa, indicating that the skill of the new algorithm is very high. The new algorithm was applied to the sounder observations from GOES‐13 (Geostationary Operational Environmental Satellite) and INSAT‐3D (Indian National Satellite) for the validation study. The validation of GOES‐13 retrieved profiles with the ozonesonde has shown that the algorithm is working better in the stratosphere than in the troposphere. Furthermore, the GOES‐13 retrieved ozone was also compared with the SBUV/2 (Solar Backscatter Ultraviolet) ozone. Overall the %RMSD (percentage root‐mean‐square deviation) values were within 30%. However, the values were within 10% above 20 hPa for most of the cases. The comparison of INSAT‐3D retrieved profiles with SBUV/2 and OMPS (Ozone Mapping and Profiler Suite) profiles have shown results almost similar to GOES‐13. The ozone values retrieved by using the existing algorithms and the algorithm presented in this work have been compared. It is found that the ozone retrieval has improved significantly with the algorithm presented in this work.