Research Group of Prof. Dr. M. Griebel
Institute for Numerical Simulation
maximize

Meteorological Data Analysis

Description

These Images show 3D weather radar measurements (data provided by D. Meetschen, Dept. of Meteorology, Univ. Bonn) of a heavy rainfall event during May 3rd, 2001 in northwestern Germany. The rainfall is visualized by 20 transparent isosurfaces with low opacity and color ranging from white to blue. The radar information is displayed on top of a 50m digital elevation model of the measurement area (courtesy of SFB350) color shaded according to elevation value.

The whole system is in implemented in a multiresolution framework. This allows a fast interaction with the data (radar and terrain) by providing good approximations of the data at highly reduced triangle counts. This allows a interactive handling of the data at several frames per second. For the final images the full resolution is used which takes several seconds to render on a normal PC.

Since the radar data is given in a weird coordinate system (conical/cylindrical), direct volume rendering techniques, such as ray casting or splatting are difficult to implement efficiently. In contrast, multiple isosurface extraction is easily applicable and provides images with sharp boundaries in addition.

The goal of this project is the quantification of the rainfall at the ground based on the 3D radar information (Z-R relationship).

Examples

Click on the images to see a magnified version.

The left image shows the target area in northwestern Germany near Bonn. The right image shows the whole measurement area (circular with 100 km diameter) as viewed from directly above. In the center is the city of Bonn, the river Rhein runs from southeast (Koblenz) to northwest (Cologne).

The left image is a 70 degree angle view of the whole area. The right image is a closeup of the rainfall event as seen from a 87 degree angle facing northwest. The blue areas indicate high precipitation. The extent of this major event is roughly 100 square kilometers. The height of all data (relief and radar measurements) is exaggerated by a factor of 5.

Here the multiresolution approach for the terrain as well as for the rainfield are illustrated. In the upper row you see the adaptive triangular and tetrahedral grids for varying error thresholds (decreasing by a factor of 10 in between from left to right). In the lower row are the corresponding images.

A time series (time interval 30 minutes) of a heavy rainfall event (moving from left to right). Unfortunately, directly above the radar antenna no rainfall information is available.

References

Related Projects


Thomas Gerstner