Evaluation of radiosonde, MODIS-NIR-Clear, and AERONET precipitable water vapor using IGS ground-based GPS measurements over China

Water vapor is one of the major greenhouse gases in the atmosphere and also the key parameter affecting the hydrological cycle, aerosol properties, aerosol-cloud interactions, the energy budget, and the climate. This study analyzed the temporal and spatial distribution of precipitable water vapor (PWV) in China using MODerate resolution Imaging Spectroradiometer near-infrared (MODIS-NIR)-Clear PWV products from 2011 to 2013. We then compared the four PWV products (Global Positioning System PWV (GPS-PWV), radiosonde PWV (RS-PWV), MODIS-NIR-Clear PWV, and Aerosol Robotic Network sunphotometer PWV (AERONET-PWV)) at six typical sites in China from 2011 to 2013. The analysis of the temporal and spatial distribution showed that the PWV distribution in China has clear geographical differences, and its basic distribution characteristics gradually change from the coast in the southeast to inland in the northwest. Affected by the East Asian monsoon, the PWV over China showed clear seasonal distribution features, with highest values in the summer, followed by autumn and spring, and the lowest values in winter. Intercomparison results showed that GPS-PWV and RS-PWV had a slightly higher correlation (R-2 = 0.975) at 0000 UTC than that at 1200 UTC (R-2 = 0.967). The mean values of Bias, SD, and RMSE between GPS-PWV and RS-PWV (GPS-RS) were 0.03 nun, 2.36 mm, and 2.60 mm at 0000 UTC, and -0.23 nun, 2.76 mm, and 2.95 nun at 1200 UTC, respectively. This showed that GPS-PWV was slightly lower than RS-PWV, and this difference was more obvious during the nighttime. The MODIS-NIR-Clear PWV product showed a similar correlation coefficient (R-2 = 0.88) with GPS-PWV compared with RS-PWV. In addition, MODIS-NIR-Clear PWV was greater than GPS-PWV and RS-PWV. The MODIS-NIR-Clear PWV showed a larger deviation from GPS-PWV (MODIS-GPS Bias = 1.50 mm, RMSE = 5.76 mm) compared with RS PWV (MODIS-RS Bias = 0.75 mm, RMSE = 5.31 mm). The correlation coefficients between AERONET-PWV and the PWV from GPS, RS, and MODIS-NIR-Clear were 0.970, 0.963, and 0.923 (with RMSE of 2.53 mm, 3.67 nun, and 4.39 mm), respectively. In the Beijing area, the overall mean bias of the AERONET-PWV product with GPS-PWV, RS-PWV and MODIS-NIR-Clear PWV was -0.09 mm, -1.82 mm, and -1.54 mm, respectively, which shows that the AERONET-PWV product was lower than the other three PWV products.