[ Home ] [ Climate ] [ PET ] [ Spatial Analysis ] [ Visualization ] [ Fire ]

Data Visualization

We have evaluated a number of data visualization routines which allow three-dimensional representations of the landscape based on digital elevation models coupled with numerous geographic and statistical information surfaces.   Software packages evaluated ranged in price, performance, and ease of use; these included VistaPro, Persistence of Vision, Landscape Visualization System, Flight Simulator, Truflite, ARC/INFO, OZ, ERMapper, and several others.  To generate visualization products, we selected VistaPro software, which is relatively inexpensive, capable of handling single or multiple DEMs, and allows draping or texture mapping of other information layers.  VistaPro also allows relatively sophisticated control of flight, camera, and target parameters.

For our initial visualization work, we selected an 340 square mile area of the Chippewa National Forest which has been used for a number of field research, forest simulation modeling, resource assessment, and multiagency planning efforts.  Regional DEMs were clipped to the study area, and four thematic layers were generated.  These included a ‘false color' image derived from Landsat bands 5, 4, and 3, (middle infra-red, near infra-red, red) which represents a more ‘realistic' view of the landscape, a modified  multitemporal forest cover classification developed by Peter Wolter at the Natural Resource Research Institute (University of Minnesota at Duluth); a map of Normalized Vegetation Difference Index (NDVI), which is correlated with the amount of green biomass on the landscape; and a map of the Shannon-Weiner Diversity Index (H') derived from land cover data. 

To conduct this latter analysis, we developed an analytical tool to quantify H' within windows of user-specified radii.  The technique uses a ‘moving window' approach to generate a circle of radius x around each pixel, calculate H' based on the relative abundances of land cover values within that window, and assign that value to the pixel.  This process is repeated for each pixel on the map, resulting in a map of H' for that particular radius.  At small radii, this is an effective edge detection technique, while larger radii show the distribution of H' at progressively larger scales. This technique is available as a FLEX FORTRAN program which runs under ERDAS 7.5.

The thematic layers we generated for the visualization project represent progressively abstract data, from an aerial view of the landscape to a three-dimensional representation of landscape diversity.  These layers were used to make a series of eight flyovers (four high-altitude, four low altitude) which show different perspectives on LandType Association E and the surrounding landscape.  The flyovers were created on a 200 MHz Pentium computer and written to CD-ROM as AVI files.  

Flyover of classified Landsat Image, Chippewa National Forest If an image from the movie doesn't appear to the left, you can try running the AVI file from here.   Alternatively, you can download the file here. This file is about 3.3 Mb.

This page last updated 1.15.02