A Global Analysis of the ZWD/PW Conversion Methods using Radiosonde Observations and Numerical Weather Models

Wednesday, 17 December 2014
Szabolcs Rozsa, Budapest University of Technology and Economics, Budapest, Hungary
Water vapor plays an important role as a basic climate variable in the thermodynamics and dynamics of the storm systems at the atmosphere and in hydrological cycles of local, regional and global scales. Moreover, the distribution of atmospheric water vapor is difficult to determine because of its rapid change in spatial and temporal scales. Atmospheric water vapor can be estimated by the zenith delay derived from ground-based GNSS data.

Ground-based GNSS receivers are a valuable source for determining total zenith delay (ZTD) and precipitable water vapor (PW) data for meteorology since they are portable, economic and provide measurements that are not affected by weather conditions. They cannot provide a humidity profile as radiosondes can, however they have the advantage of producing automated continuous data as opposed to operational radiosondes usually providing two measurements in a day. Therefore, tropospheric delay modeling methods for estimating precipitable water vapor using GNSS signals are being developed frequently. Wet and hydrostatic zenith delays can be computed by applying the mapping functions which are mathematical equations using elevation angles. The observed tropospheric delays can be used for monitoring the water vapor content of the troposphere. In several regions of the world GNSS derived products are already used on a routine basis for numerical weather prediction.

In this study, PW values obtained from radiosonde profiles and the ones derived from ground-based GNSS data are processed both with BERNESE v5.0 using Niell mapping function and GAMIT/GLOBK using empirical model GPT (Global Pressure and Temperature) are compared with the values computed from radiosonde analysis algorithm under severe storm conditions. In order to convert the ZWD to PW new, locally fitted models are derived using local radiosonde observations and ECMWF model data.