Studies of atmospheric effects on Global Positioning System (GPS) signals h
ave proven the possibility of deriving the total water vapor content from e
stimates of tropospheric path delays. The accuracy of GPS derived Precipita
ble Water (PW) depends (besides other parameters) on the quality of satelli
te orbits used in the analysis. High precision orbits provided by the Inter
national GPS Service (IGS) yield PW estimates with an accuracy of about 1 m
m. While these orbits are provided with a delay of several days, weather fo
recasting requires near real-time determination of PW. Therefore operationa
l meteorological GPS analysis would have to rely on orbit predictions. We i
nvestigate the impact of introducing predicted orbit information on the acc
uracy of GPS water vapor retrievals. The presented data were acquired durin
g a 14-day field experiment carried out in the north-west region of Madrid,
Spain using GPS and a Water Vapor Radiometer (WVR). The comparison of WVR
measurements with estimated time series of PW using both 24 and 48 hour pre
dicted orbits and final precise IGS orbits shows that the accuracy of PW de
creases by a factor of about 2 from precise to predicted orbit data.