This project visualized the processes and results of a GIS analysis for simulating the distribution of nitric pollutants in a watershed in accordance with the geographic (topological) and environmental conditions. The analysis was done by a team coordinated by Prof. Stephen DeGloria of the Center for the Environment at Cornell using various GIS software packages and independently developed programs. The goal of this project is to generat a flexible prototype of visualization modules for a typical GIS work using IBM Data Explorer.
Since the data came from different sources and the simulation was done by the analysis team using different packages and self-developed programs, the data formats are diverse. In addition, considering the huge data volume and the limitation of data storage in most machines, the data are stored in Cornell Theory Center's UniTree mass storage set, and then retrived via AFS whenever needed. After the data were fetched, they were converted into consistent format using several UNIX shell scripts and C programs. One of the purposes of the data conversion is to speed up the data extraction from the two-year master hydrology or nitrogen data, otherwise, it will be time consuming to deal with different formats of different data sets.
After importing the data, DX needs to conduct coordinate transformation because DX has a right-hand coordinate system and most of GIS data are in left-hand coordinat systems, some of them are even in their own particular coordinate systems. In addition, since the data came from different sources, they may be collected at different time, and in different cell resolutions. If two or more data sets are involved in one application, the data have to be interpolated and/or manually adjusted to correct the coordinate displacements.
Assume the data are all in correctly correspondent positions, the next step is to construct the prototype of two-dimensional and three-dimensional display. Several color look-up tables have been generated and in the format of DX compatible, these LUT's are for soil, and landuse displays. Hydrology and nitrogen data are colored by the colormap module of DX. However, because most of the distributions of the nitrigen and hydrology data are not linear, a special Log scale scheme was adopted to log transform the data interactively. The log algorithm is described below:
By adopting this algorithm, the values of data will be scaled to values between 0 (if NewMin is specified as 1.0) to 1. Unfortunately, the results of this log algorithm varies with user sepecified NewMin and NewMax, therefore, different data sets need different parameters to accomplish a reasonable redistribution of the data values.
The last part of the visualization procedures is to provide adjustible animation options for different scenarios. In this project, both two-dimensional and three-dimensional animations are provided.
In order to provide ultimate flexibility, parameters are built in the control panels for users to specify the values interactively. These parameters include file (data) selectors, data layer selectors, coordinate correction parameters, log algorithm parameters, display options, viewing control, etc.
The DX nets generated in this project provide the analysis team an alternative to review their work in different perspectives and to present the processes and results in a more effective and understandable way. Since those DX nets were built to provide a prototype for visualizing the GIS work, they can be used to generate various displays and animations in accordance with different purposes. They can also be easily modified for similar GIS analysis projects to adopt.
The tentative future work of this project might be to incorporate the visualization algorithms with the internet utilities, for example, build up the VRML applications in the Web pages.
A perspective view of colored DEM. Last updated on NOv. 18, 1995
The watershed area. Last updated on Nov. 18, 1995
The soil map. Last updated on Nov. 19, 1995
Landuse map for test #1. Last updated on Nov. 19, 1995
Isosurface generate from a volumetric (4-layer) nitrogen data. Last updated on Nov. 26, 1995
A slice of the organic-nitrogen volume. Last updated on Nov. 26, 1995
An animation of the runoff data (day-1 to day-50) for test #1.Last updated on Nov. 20, 1995
An animation of the watercontent changes (day-1 to day-50) for test #1.Last updated on Nov. 20, 1995
I also want to thank Chris Pelkie at the Visualization Group of Cornell's Theory Center for providing me helpful information about the IBM Data Explorer.
Also thank Prof. Stephen DeGloria for coordinating the GIS work and visualization work for the project, and thank Ms. Wenling Kuo and Mr. Dennis Swaney for providing me the original data.
2. Graf, K. Ch., Suter, M., Nuesch, D., "Perspective terrain visualization-A fusion of remote sensing, GIS, and computer graphics." in Computers & Graphics, 1994 v18 n6, p-795.
3. Wood, J. D. and Fisher, P.F., "Assesing Interpolation Accuracy in Elavation Models.", in IEEE computer graphics and applications, Mar. 1993, p-48.
4. Denzer, R., "Graphics for Environmental Decision Making.", in IEEE computer graphics and applications, Mar. 1993, p-58.
5. Theresa Rhyne et. al., "Visualizing Environmental Data at the EPA.", in IEEE computer graphics and applications, Mar. 1993, p-34.
6. IBM Data Explorer document.
Your comments are welcome. Please E-mail me your opinions.