An evaluation of the precipitable water (PW) in the reanalysis of the National Center for Environmental Prediction (NCEP) over Antarctica was carried out using observations from eight coastal Antarctic GPS stations. PW time series were derived from tropospheric parameters in conjunction with meteorological observations provided by the British Antarctic Survey for the period from 1994 to 2004 with a 2h temporal resolution. The tropospheric parameters this study is based on are a product of the common GPS reprocessing from the Universities of Technology in Munic and Dresden.
The validation of NCEP PW reveals an underestimation of the seasonal signal in the PW from NCEP by 25% on the coast of East Antarctica and an overestimation of the PW by about 10% on the Antarctic Peninsula. Subdaily variations in the Antarctic PW are not correctly represented by NCEP due to the coarse spatial and temporal resolution of the global model. The agreement between GPS and NCEP PW is much better for most of the other regions on the earth than it is over Antarctica. The reason for the higher un-certainties in the Antarctic PW from NCEP is the lower availability of water vapour data in other region on the earth than it is over Antarctica. The reason for the higher un-certainties in the Antarctic PW from NCEP is the lower availability of water vapour data in the southern hemisphere especially, over Antarctica. PW values derived from GPS data could help to fill the large gape of water vapour data over Antarctica and could be assimilated in numerical weather prediction models within the near future.