Retrieval of the atmospheric water vapour content from Sentinel-5 Precursor measurements using the AMC-DOAS algorithm

The aim of this study is to retrieve the total column water vapour (TCWV) from spectral measurements of the Tropospheric Monitoring Instrument (TROPOMI) onboard the Sentinel-5 Precursor (S5P) satellite and to assess the quality of the derived TCWV data by comparisons with other, independent TCWV products. The retrieval is based on the Air Mass Corrected Differential Optical Absorption Spectroscopy (AMC-DOAS) algorithm, which is a method to retrieve TCWV from measurements in the spectral region around 700 nm. To apply the AMC-DOAS method to spectral measurements from S5P, the surface height has newly been added to the retrieval. The quality filtering is also extended by a cloud fraction filter of 0.2 and a filter for cloud height above surface of 2.0 km. Another new aspect is the development and implementation of a cloud and albedo correction scheme. An empirical swath correction also has been applied, which reduces observed swath effects in the TCWV data over water. The AMC-DOAS TCWV values from S5P are assessed to ensure the quality of the AMC-DOAS retrieval. For that purpose, the new AMC-DOAS product is compared against other, independent data sets from GOME-2 on MetOp-B, reanalysis data from the European Centre for Medium Range Forecast, Special Sensor Microwave Imager and Sounder and two different S5P water vapour products from the Max Planck Institute for Chemistry in Mainz and the Netherlands Institute for Space Research. The GOME-2 data set is processed with a former AMC-DOAS version without surface height, surface albedo and cloud correction. The daily comparison between the AMC-DOAS S5P product and the other products reveal small differences between -2.5 kg m-1 and +2.5 kg m-1 with except for SSMIS TCWV. Best agreement of the AMC-DOAS S5P TCWV is found to the other two S5P water vapour products. The spatially resolved comparison reveals complex structures. Many of the spatial patterns are expected and are related to methodological differences and partly in observational differences. In summary, the TCWV derived by the improved AMC-DOAS algorithm applied to the S5P measurements agrees well with the other data sets. The adjusted quality filtering and the cloud and albedo correction minimises the impact of the surface reflectivity and the influence of remaining clouds.