Assessment and calibration of MODIS precipitable water vapor products based on GPS network over China

The ground-based GPS provides a promising way to assess and calibrate precipitable water vapor (PWV) products from the Moderate-Resolution Imaging Spectrometer (MODIS) which has high spatial resolution and near global coverage. This study utilized PWV at about 260 ground-based GPS stations from the Crustal Movement Observation Network of China (CMONOC) from 2015 to 2018 to comprehensively assess the MODIS Near Infrared (NIR) PWV products as well as the latest ECMWF reanalysis, ERA5. Results reveal a general underestimate of MODIS PWV compared to GPS over China, with mean bias of 1.32 mm (GPS-MODIS), compared to −0.28 mm for ERA5 (GPS-ERA5). Taking GPS PWV as references, the average MODIS PWV error root mean square (RMS) is about 5.35 mm, with correlation coefficient (R) of ~90% between GPS and MODIS PWV, compared to the RMS of 1.73 mm and R of ~98% for ERA5. PWV differences between MODIS and GPS also show clear seasonal and climate type dependency over China. A refined linear MODIS PWV calibration model taking into account of climate region difference was then developed and examined. The calibrated MODIS PWV at more than 800 independent GPS stations from China Meteorological Administration (CMA) in the year of 2016 show obvious improvements, with RMS reduced by about 14.7, 17.7, 20.9, 2.5 and 11.3% in tropical monsoon (TM), subtropical monsoon (SM), midlatitude monsoon (MM), humid continental (HC) and plateau (PL) climate region, respectively. Compared to the unified calibration model which does not distinguish the climate regions, the RMS values after calibration using the refined model are reduced by 10.5, 8.3, 9.1, and 11.9% in TM, SM, MM and Plateau climate (PL) region, respectively. In addition, bias values are reduced from 2.01, 3.42, 1.50 and 0.84 mm to 1.64, 0.33, 0.68 and −0.06 mm in TM, SM, MM, PL region, respectively.