A comparison of precipitable water vapor estimates by an NWP simulation and GPS observations

Simulated time series of the total precipitable water (PW) vapor from a limited area numerical weather prediction model are compared to estimates derived from observations done with ground-based Global Positioning System (GPS) receivers. The model data examined are from the delayed-mode High Resolution Limited Area Model (HIRLAM) data assimilation (reanalysis) and the short-range forecasts on double nested grids. The observational data are derived from GPS measurements at 25 sites in Sweden and Finland over a 4-month period, August-November 1995. In general, the HIRLAM reanalysis system demonstrates considerable skill in reproducing the spatial and temporal evolution of the PW as depicted by the GPS estimations. Using a 0.2 degrees horizontal resolution and 31 vertical levels, the HIRLAM reanalysis generates a PW time series that has, in comparison to that of the GPS estimates, an average offset of -0.1 mm and a root-mean-square difference of 2.4 mm. The average correlation between the PW time series from the HIRLAM reanalysis and from the GPS observations is 0.94. An examination of the model forecasts shows no indication of spinup in the PW prediction. The correlation between forecasts and GPS estimates of PW deteriorates slowly with increasing integration length, but even for the 30-h-long forecasts the correlation level is as high as 0.93 and the mean offset remains small. It is also found that with longer integration length. prediction of PW tends to be more positively biased. The forecast error, in terms of the rms difference between model and GPS estimates, becomes larger for increasing PW. In addition, the model simulation tends to underestimate when PW becomes large. Comparing simulations with a resolution of about 0.2 degrees and 0.4 degrees shows no obvious PW dependence on resolution. The GPS and the model PW time series are found to be in guile good agreement with data derived from measurements by a microwave radiometer and by radiosondes. The results indicate that the GPS-derived PW data, with high temporal and spatial resolution, are very useful for meteorological applications. The main problem with the GPS data used in this study is found to be related to the design of GPS antennas.