Visualization for Density driven Groundwater Flows

This three years research project is funded by the Federal Department of Research and Technology. The visualization and graphical analysis of groundwaterdata, wich is described in this short introduction, is only one part of the whole simulation-project, wich started 1995.

Abstract

The security and protection of mankind and environmet is one of the main interests in our time. In this context the treatment of security in fundamental questions for deposites of dangerous wastes is neccessary. This project will develop a posibility to simulate groundwater flows with regard to salt-effects for big three dimensional complex hydrological areas in acceptable computing times. Thus it is possible to reach validated hydrological models for the security of waste deposites.

The objective of the visualization part of the current project is to develop an interactive graphic environment which should be able to picture the numerical data, to extract the relavant physical or hydrological effects and to support visual debugging in the development of the numerical code. For such purposes the SFB 256 at Bonn University and the Institut für Angewandte Mathematik at the University of Freiburg has developed the visualization environment GRAPE.

Within this research the working group of Prof. Kroener in Freiburg developed a visualization package to picture data from physical simulations, such as timedepentent flows etc.. The adaption of this program-package to the situation described above and further developments in order to extract relevant features is the main goal of this project.

Recent numerical methods to solve partial differential equations in scientific computing are based on a variety of advanced kinds of domain discretizations and appropriate finite dimensional function spaces for the solutions. The scope of grids under consideration includes structured and unstructured, adaptive and hierarchical, conforming and nonconforming meshes. The function spaces might be of Lagrangian or Hermitian type with higher polynomial degree and possibly discontinuous over element boundaries. Unfortunately, the rendering tools in scientific visualization are mostly restricted to special data structures which differ substantially from the data formats used in the numerical application. This forces users to map and interpolate their data, which is time consuming, storage extensive, and accompanied with interpolation errors. The aim of this project is to present an interface between numerical methods on various types of grids and general visualization routines which overcomes most of these disadvantages. It is based on a procedural approach managing a collection of arbitrary elements and a set of functions describing each element type. The numerical data is not mapped onto new data structures. Providing a small set of procedures and functions the user will furtheron work on his own data structures. These routines temporarily and locally manage the access to data of interest, like information about a single element. Compared with display routines on a specialized data structure, this general interface does not produce much cpu overhead.