| description abstract | ong-term trends in precipitable water (PW) are an important component of climate change assessments for the Tibetan Plateau (TP). PW products from Moderate Resolution Imaging Spectroradiometer (MODIS) are able to provide good spatial coverage of PW over the TP but limited in time coverage, while the meteorological stations in the TP can estimate long-term PW but unevenly distributed. To detect the decadal trend in PW over the TP, Bayesian inference theory is used to construct long-term and spatially continuous PW data for the TP based on the station and MODIS observations. The prior information on the monthly-mean PW from MODIS and the 63 stations over the TP for 2000?06 is used to get the posterior probability knowledge that is utilized to build a Bayesian estimation model. This model is then operated to estimate continuous monthly-mean PW for 1970?2011 and its performance is evaluated using the monthly MODIS PW anomalies (2007?11) and annual GPS PW anomalies (1995?2011), with RMSEs below 0.65 mm, to demonstrate that the model estimation can reproduce the PW variability over the TP in both space and time. Annual PW series show a significant increasing trend of 0.19 mm decade?1 for the TP during the 42 years. The most significant PW increase of 0.47 mm decade?1 occurs for 1986?99 and an insignificant decrease occurs for 2000?11. From the comparison of the PW data from JRA-55, ERA-40, ERA-Interim, MERRA, NCEP-2, and ISCCP, it is found that none of them are able to show the actual long-term trends and variability in PW for the TP as the Bayesian estimation. | |
