Analysis and Derivation of the Spatial and Temporal Distribution of Water Vapor from GNSS Observations

Ming Shangguan (Successful finalization March 2014)

Faculty VI - Planning Building Environment, Technical University Berlin

Water vapor in the atmosphere plays an important role in meteorological applications. The Global Navigation Satellite System (GNSS) provides accurate all-weather observations. The application of the existing GNSS infrastructure for atmosphere sounding leads to rather inexpensive and reliable measurements of the atmospheric water vapor. Observations from GNSS networks contain information about the spatial and temporal distribution of the water vapor. Therefore, the German Research Center for Geosciences Potsdam (GFZ) developed a water vapor tomography system to derive 3-Dimensional (3D) distributions of the tropospheric water vapor above Germany. The tomography makes use of the products provided by the GNSS processing center of the GFZ, where the atmosphere processing is currently limited to the Global Positioning System (GPS). Input data for the water vapor tomography are the GPS tropospheric products from about 300 ground stations. The GPS tropospheric products are Zenith Total Delay (ZTD), Integrated Water Vapor (IWV) and Slant Total Delay (STD).

The accuracy of STDs is one of the important factors for the quality of the derived water vapor tomography. However, the Earth Parameter and Orbit System Software (EPOS), which is used to estimate the GPS-STDs at GFZ, provides only limited information about the accuracy of STDs. Three months of Water Vapor Radiometer (WVR) data are used to validate the GPS-STD and estimate its accuracy. By comparing the GPS-STD observations with systematic hemisphere scans of the WVR it could be shown that inhomogeneous atmospheric structures are reliably reproduced by the STDs. The validation has shown a high accuracy of the estimated STDs.

The main objective of this thesis is to improve the water vapor tomography and to provide atmospheric water vapor products with good quality. A new tomographic algorithm based on a Kalman filter is added in the GFZ tomography system. The output is a 3D humidity field with a temporal resolution of 2.5 min and the error covariance matrix of the reconstructed states. The error covariance matrices for the observations and the covariance matrices for the uncertainty of the propagation are estimated in advance. The output has been validated with the Multiplicative Algebraic Reconstruction Technique (MART) tomography and radiosonde profiles.

Besides the accuracy of STDs, the quality of the derived tomography is depending on many factors such as the spatial coverage of the atmosphere with slant paths and the spatial distribution of their intersections. This leads to temporal and spatial

variations of the reconstruction quality. Independent observations are required to validate the generated water vapor tomography. One year of radiosonde data from the German Weather Service (DWD) have been used for the validation. The wet refractivity field of the tomography with about 50 km horizontal resolution and 500 m vertical resolution has been interpolated to the RS profiles. The validations have

been carried out point-by-point and also for the whole profile. A new technique has been developed to quantify the differences between humidity profiles. By considering the shape of the whole profile much more reliable conclusions can be drawn than by comparing only point-by-point differences. This method can be applied to improve the algorithms of GPS tomography.

Further attempts have been made to analyze the long-term IWV time series. Since the GPS data are available for more than 10 years, the GPS-IWV time series are used for climatological studies and they will become more important in future when long time series will be available. Trends have been calculated for the period 2002-2012 using the IWV from the German GPS ground-based network. Different methods (per station or per region) have been used to analyze the IWV time series. The methods will be helpful for meteorologists to analyze variations of the local or regional weather.

The investigations demonstrated that the ZTD, IWV and STD could describe the amount of water vapor and its distribution in the troposphere reliably. Especially the spatial and temporal variation of the water vapor distribution in the troposphere can be estimated with the tomographic technique. The quality of the derived 3D humidity fields has been checked with the help of radiosonde data. In general the result of the validation is good but it shows a need to improve the quality of the water vapor tomography.

With the development of GNSS (more satellites and more GNSS stations) and with improved algorithms (e.g., introduction of radiometer or radiosonde data), the tomography will in future provide a more complete view of the water vapor distribution in the atmosphere. In addition with increasing of GNSS time series, they can also be used for long-term studies. The GNSS meteorology can be widely applied in many fields, e.g., nowcasting, severe weather monitoring and data assimilation.