Improvement of Integrated Water Vapor Products from Sentinel-3 OLCI NIR Channels Using Ground-based GPS Measurements

Abstract

High-quality remote sensing water vapor monitoring plays a vital role in Earth's weather and climate observation. In this work, we developed a model to improve the accuracy of integrated water vapor (IWV) products retrieved from the Ocean and Land Color Instrument (OLCI) near-infrared (NIR) channels using ground-based GPS-derived IWV data. This algorithm uses the empirical regression equations to calibrate the official OLCI IWV products published by Sentinel-3. The reference IWV data, from June 1, 2018 to May 31, 2019, from 453 GPS sites over Australia, were employed to calculate the differential IWV data by subtracting GPS IWV data from OLCI IWV data. The OLCI IWV pixels were grouped into two categories according to the quality flag of each pixel. For each group, the relationship between the differential IWV data and the official OLCI IWV products was defined, and the model parameters were independently calculated from each season. The performance of the model was evaluated using reference IWV data from GPS and European Centre for Medium-Range Weather Forecasts (ECMWF) from June 1, 2019 to May 31, 2020 over Australia. Taking GPS IWV data as reference, the root-mean-square error (RMSE) has reduced 27.63% from 3.475 to 2.515 mm for Sentinel-3A, and 18.06% from 4.030 to 3.302 mm for Sentinel-3B. Taking ECMWF IWV data as reference, the RMSE was reduced by 25.27% from 3.490 to 2.608 mm for Sentinel-3A, and by 23.54% from 3.535 to 2.703 mm for Sentinel-3B. The improvement of OLCI IWV products was further confirmed in Mainland China, with smaller RMSE against reference IWV data from 214 GPS stations from June 1, 2019 to May 31, 2020.