Water vapour content of the atmosphere plays essential role in weather prediction and modeling. However it is still poorly sampled mainly due to its high temporal and special variability. In this respect the Global Positioning System data are very suitable source of information.

We use Swiss GPS network AGNES to derive Integrated Water Vapour (IWV) from Zenith Total Delay signal and to compare it with water vapour from Numerical Weather Prediction (NWP) model - Local Model (LM) of MeteoSwiss. LM is a nonhydrostatic mesoscale model with grid mesh 7km. We analyse data from model analysis and forecast over the period April - November 2001.