Water vapour products from ERA5, MERSI-II/FY-3D, OLCI/Sentinel-3A, OLCI/Sentinel-3B, MODIS/Aqua and MODIS/Terra in Australia: A comparison against in situ GPS water vapour data

Australia is a region of high sensitivity to the El Nino-Southern Oscillation events in association with atmospheric water vapour, which is an essential atmospheric parameter in hydrological cycle, energy transport and climate monitoring. In this article, we thoroughly validated the quality of multisource precipitable water vapour (PWV) products sourced from ERA5 reanalysis, advanced Medium Resolution Spectral Imager (MERSI-II) sensor on the FY-3D satellite, Ocean and Land Colour Instrument (OLCI) sensor on the Sentinel-3A and Sentinel-3B satellites, and Moderate Resolution Imaging Spectroradiometer (MODIS) sensor on the Aqua and Terra satellites. The PWV estimates measured by 453 in situ Global Positioning System (GPS) sites from 1 June 2019 to 31 May 2020 over Australia were employed as the common reference. The validation results show that all the reanalysis-based and satellite-based PWV products had a good agreement with reference GPS-based PWV data. ERA5 reanalysis had the highest PWV accuracy with a root-mean-square error (RMSE) of 2.016 mm and a relative RMSE (RRMSE) of 12.038%, while the MODIS/Terra instrument had the lowest PWV accuracy (RMSE = 4.903 mm and RRMSE = 34.187%). In contrast to the MERSI-II/FY-3D instrument that underestimated PWV, the PWV data from ERA5, OLCI/Sentinel-3A, OLCI/Sentinel-3B, MODIS/Aqua and MODIS/Terra overestimated the water vapour values. The quality of reanalysis-based and satellite-based PWV measurements showed significant dependencies on several variables - location, PWV, solar zenith angle, season, elevation, land surface cover and latitude. It is expected that this research can help enhance the understanding of the quality of reanalysis and satellite water vapour products in Australia, that is, ERA5, OLCI/Sentinel-3A, OLCI/Sentinel-3B, MODIS/Aqua and MODIS/Terra. This work could provide an insight into improving the accuracy of reanalysis-based and satellite-based PWV data products after exploring the dependency factors that affect PWV performance as presented in our work.