Inter-comparison of integrated water vapor from satellite instruments using reference GPS data at the Iberian Peninsula

Abstract

This paper focuses on the inter-comparison of integrated water vapor (IWV) products derived from the following satellite instruments: Global Ozone Monitoring Instrument (GOME-2), Moderate-Resolution Imaging Spectroradiometer (MODIS) on the Terra and Aqua satellites, Ozone Monitoring Instrument (OMI), Spining Enhanced Visible and InfraRed Imager (SEVIRI), Atmospheric Infrared Sounder (AIRS), and Scanning Imaging Absorption Spectrometer for Atmospheric Chartography (SCIAMACHY). IWV data from GPS in nine groundbased stations located in the Iberian Peninsula are used as reference. The study period extends from 2007 to 2012. The results show that, in general, OMI has good accuracy (pseudomedian of the relative differences between OMI and GPS IWV of (−0.7±1.1)%). However, OMI, SCIAMACHY and AIRS show higher interquartile range (IQR) (which indicates lower precision) than the rest of satellite instruments. Both MODIS satellite instruments and SEVIRI products tend to slightly underestimate reference IWV data while GOME-2 exhibits a notable overestimation (16.7±0.8%). All satellite instruments showed a tendency to reduce IWV extreme values: low IWV is overestimated while high IWV is underestimated. As for the influence of solar zenith angle (SZA), it can be observed that GOME-2 strongly overestimates the reference for high SZA values (by around 60% for SZA 60−80°). OMI shows, however, a high IQR for high SZA values. Both MODIS instruments show an increase in the pseudomedian of relative differences and IQR with SZA at daytime, with more stable values at night. Seasonal dependence is mainly due to the SZA and IWV typical values in each season. In general, in summer the tendency is to underestimate with low IQR (which happens when IWV is high and SZA is low), and in winter the trend is to overestimate with high IQR (which happens when IWV is low and SZA is high). SCIAMACHY shows a high pseudomedian in summer and autumn, and lower in winter and spring. It must be noted that GOME-2 shows a higher overestimation and OMI shows a higher IQR than other satellite instruments in winter and autumn. The influence of clouds was also studied, showing an increase of IQR as cloudiness increases in all satellites. Pseudomedian also worsens as cloudiness increases, generally.