Water vapour is a key variable in atmospheric processes and plays a crucial
role in atmospheric motions on a wide range of scales in space and time. T
he water vapour content is approximately proportional to the zenith wet del
ay (ZWD) which, in turn, constitutes a crucial parameter in geodetic microw
ave space techniques (VLBI and GPS). Apart from being determined by measure
ment techniques, the ZWD can also be derived from numerical weather predict
ion (NWP) models such as the non-hydrostatic MM5 model and the hydrostatic
HIRLAM model. At the station Robledo de Chavela (Madrid) ZWD values were de
rived from VLBI, GPS, MM5, and HIRLAM for the beginning of December 1996. T
he results of the different techniques agree to the sub-centimetre level wi
th correlation values of 0.87 (GPS vs. MM5), 0.81 (GPS vs. HIRLAM), and 0.8
4 (MM5 vs. HIRLAM). The correlation VLBI vs. MM5 of 0.78 is based on a shor
t VLBI time series and should be considered preliminary. Further studies wi
th longer time series are necessary to confirm this value. The bias and RMS
difference values are all contained in the margin provided by the internal
errors.