Current research

Numerical weather models for space geodetic applications

The tropospheric delay is still one of the limiting factors which reduces the accuracy of space geodetic techniques. Recent investigations have shown that the introduction of ray-traced delays from numerical weather models improve the results and help to remove systematics. Using the 10km mesoscale model from the Japanese Meteorogical Agency (JMA) it became feasible to start the development of a ray-tracing service for East Asia which provides tropospheric corrections in real-time but also supports post-processing requests. The Kashima Ray-Tracing Tools (KARAT) are able to handle different space geodetic data format and directly subtract delay corrections for each observation given. A significant improvement for precise point positioning (PPP) applicaations was found and the impact of ray-traced troposphere delays is currenly under investigation, too. The Kashima Ray-Tracing Service (KARATS), which allows users to upload their data and obtain (nearly) troposphere-free observations, is planned to become operational in the beginning of 2008.

Improvement of the accuracy of geophysical and instrumental models for space geodesy

With the advance of measurement accuracy and precision geophysical and instrumental effects, which have not been modeled until now, have attracted the attention of the space geodetic community. In order to reach mm accuracy it is necessary to identify the origin and characteristic of each error and thereafter model or compensate for the geophysical model/instrumental imperfectness. E.g. ionosphere models are usually based on the assumption the the Earth can be approximated by a sphere. Investigations have revealed that the neglection of the Earth's oblateness leads to small errors in the estimated ionosphere parameters. Moreover it could be shown that instrumental biases (DCBs) which are estimated together with the ionosphere parameters are affected significantly by the choice of the refernce figure. Additionally, new instrumental designs require modifications of the existing analysis strategies to take full advantage from system upgrades. Thus it is studied, how the mm-accuracy goal for the upcoming VLBI systems can be achieved. After simulating observational data using the specifications of the new systems it could be shown that it will be possible to compensate systematic and instrumental error well and that the ambitious accuracy goal of VLBI can be reached.

Development of algorithms for 3D real-time monitoring of the ionosphere

There are plenty of three-dimensional ionosphere reconstructions from GNSS observations, which have all in common that the media is sampled too sparsely to image the media on global scales. To overcome this drawback the spatial resolution (cell size) is enlarged or mathematical assumptions support the regions which are not crossed by rays. Thus, a simple and fast algorithm has been developed, which allows both, tomographic reconstruction of the electron density field with a high spatial resolution and the appliance of physical conditions which support regions with low observational coverage. Two-dimensional tests have revealed that the algorithm can compete with the accuracy of traditional tomographic inversions at much lower computation cost and that the method is also capable of estimating instrumental biases together with the ionospheric parameters. Modification of the algorithm for 3D tomographic reconstructions as well as real-time processing are currenly under investigation.

Development of "state-of-the-art" software interfaces for Very Long Baseline Interferometry (VLBI)

Since the time VLBI has been introduced to the scientific community in the late 70ies, the binary data format has never been adopted to modern database structures and does only provide interfaces to a single analysis software package. This is not only a drawback for the analyst, who wants to access the data directly, but also limits the correlation centers in their choice of information which is written to the database. Thus we have created a set of tools, written in C++, which can read/write the existing databases, independent of the machines Endianess, without binding of external libraries. Since the data is stored in the NetCDF format, which offers interfaces to nearly any programming language, the user can extract all the information without knowledge about how the data is stored. Moreover additional information can be kept in the database without interfering the data-structure. Although such information is not used by current analysis software packages it might be of importance in the future. Since all tools can be controlled from the command-line it became possible to automate VLBI experiments and to obtain geodetic target parameters in near real-time without human interaction. One field of application is expected to be the ultra-rapid determination of UT1 which VLBI. All satellite geodetic techniques are relying on predicted UT1 values, which show a decrease of accuracy by increasing prediction length. Thus, the ultra-rapid VLBI measurements are expected to overcome this drawback and the developed tools will help to realize an automated processing chain.

Parallel processing for geosciences

Driven by the recent development of micro-processors it became feasible to carry out tasks by software methods rather than utilizing devoted hardware components(ASICs). Moreover, deployment of multi-core architectures allows to speed up computational efficiency by parallel methods. Beside modern multi-core CPUs, graphics processing units (GPUs) allow the user to boost his applications by up to two orders of magnitude, which enables real-time applications or realization of time-critical applications. Also signal-processing can be carried out by software, which removes restrictions caused by the number of hard-wired bits and reduces development time. Additionally, new algorithms can be tested and rapid-prototyping can be realized by the help of such parallel systems.

Past research