IBM Data Explorer (DX) has been applied to to three environmental issues. In the first case, DX was used to construct a visual model of San Diego Bay and to create a San Diego Bay VRML data repository where three-dimensional thematic maps can be composited, viewed, manipulated and analyzed in a web environment. In the second case, DX was used to assess and visualize a large meteorological database in midwestern United States. A time series of monthly temperature accumulation based on 1044 locations were mapped to analyze seasonal dynamics of heat. In the third case, we applied DX to an insect population in Michigan. Gypsy moth numbers have measured since 1985. A time series of moth trends were produced using DX to assess change in this animals' population throughout the State of Michigan.
Five municipalities share the coastline of the San Diego Bay and many more depend on it for domestic and international commerce and diverse recreational opportunities. Over thirty organizations and municipalities are involved in the Bay's management. Together, SDSC and the San Diego Bay Interagency Water Quality Panel are defining a new way in which information technology can be used to assist public policy decision-making. The central focus of this effort is the development of a visual, three-dimensional model of the Bay based on data contributed by each of the panel member organizations which have collection programs. The current state of the project can be seen at http://www.sdsc.edu/~sdbay/
This web site offers a number of images rendered from spatial and temporal data using the IBM Data Explorer (DX) software. We've also used DX to create three-dimensional data files in the Virtual Reality Modeling Language (VRML) format for user-controlled data integration and exploration.
The gypsy moth is an insect that prefers to consume the foliage of trees like oak and aspen. Imported in error into the United States after the Civil War to develop a silk industry, the moth escaped from confinement and began a costly trek across parts of North America over the past 100 years or so causing millions of dollars in economic loss to the US and Canadian Governments and to the tourism industry, the forest industry and to homeowners who dwell in the path of the gypsy moth outbreaks. We've used DX to visually track an outbreak of this pest in the State of Michigan. Further information on this project can be found at http://www.ent.msu.edu/esal/gypsy/
Weather is a key component which defines seasonal dynamics of biotic systems. For example, over the past few years the midwest corn belt has experienced significant floods and droughts which have seriously affected the potential of the region to produce the high crop yields necessary to maintain farm profitability. High variability in heat and moisture within and between years makes forecasting crop productivity challenging. High variability in weather patterns are also implicated as indicators of potential climate change. To study within and between dynamics of weather, patterns of monthly heat accumulation time and space variability was selected to analyze using DX. Further information on this project can be found at http://www.ent.msu.edu/esal/ncregion/
Two-dimensional and three-dimensional data visualization of the San Diego Bay watershed has been performed. Numerous datasets have been transformed to Universal Transverse Mercator (UTM) coordinates and visually integrated including: San Diego Bay bathymetry (underwater topography), eel grass distribution and concentrations in the Bay, San Diego Bay coastline, digital elevation models, and sediment chemistry concentrations. We used the Global Coordinate Transformation Program (GCTPc) from the EROS Data Center (ftp://edcftp.cr.usgs.gov/pub/software/gctpc/gctpc.tar.Z) to project most of these data coverages from their native coordinate system to our common UTM system.
Through an iterative experimentation process, several imaging methods have been developed. For example, color coded eelgrass density polygons were overlaid on a greyscale three-dimensional surface representation of the bathymetry. The greyscale basemap has proven useful and is a common theme in all of the visualizations.
Temporally and spatially sparse datasets detailing sediment chemistry samples were collected over a ten month period in 1992 and 1993 at up to eight stations per day. Tests for more than a hundred compounds were conducted. Our goals in creating visualizations of the sediment chemistry data were to:
The image below shows an IBM DX visualization method that achieves these goals. The image is one frame from an MPEG animation that displays one frame for each testing date. Test sites for a particular day are highlighted with a white disk. Station locations are indicated by a black dot, sample values are displayed as an opaque colored disk behind the blackdot. Non-detects are displayed as blue diamonds.
We also experimented with displaying interpolated concentration values as a semi-transparent surface grid behind the colored disks. This technique seemed useful for suggesting concentration trends, however, was deemed too controversial in general practice.
Precipitation for the appropriate year and day are shown in a one-dimensional line graph that has been superimposed on the image. Station key legend and a colorbar are also superimposed. The animation was generated using the DX sequencer and iterating through the test dates. Data were read in from DX .general files where successive fields indicated date of test, position, day of year, year, and value measured. Each of the 117 compounds tested for were rendered in a separate animation. In addition to identifying areas where high concentrations may exist, the resulting images have been useful in determining where further data should be collected.
Here is an example animation that shows what stations were sampled on what days during the data collection projection. The animation also shows each sampling date on a precipitation graph so that possible correlations between stormwater and sediment chemistry can be studied.
To experiment with providing web-based visualization capabilities for novice visualization users, we have used IBM DX to convert a number of our San Diego Bay data coverages to virtual reality modeling language (VRML) version 1.0 files. One goal of this work is to study how VRML can be used to present web-based three-dimensional thematic maps to scientists and policy makers.
The DX2VRML module from the Cornell Theory Center (CTC) (http://www.tc.cornell.edu) was used within DX to convert downsampled versions of San Diego Bay data coverages into VRML files. For the bathymetric and elevation data this required fitting a polygonal surface to the gridded depth/height values. The sediment chemistry samplings are displayed as disk icons where the disk color represents sediment concentration. A colorbar legend giving the numerical value range for the colors can also be composited into the VRML scene.
In the image below, the table on the left shows how a user can select a number of these individual VRML datasets and legends to be composited into a single VRML file for interactive viewing. The image on the right shows a collection of four datasets being viewed, manipulated, and analyzed together in the SDSC webview VRML-based web-browser [SDSC95].
Compositing a collection of VRML files into a single scene is achieved using a PERL script which builds the new VRML file on-the-fly using a series of VRML inline nodes. An inline node contains a URL address for a VRML file, in this case, a VRML file containing actual geometric data. When the VRML-based web-browser encounters an inline node, the browser retrieves the VRML file named in the URL and inserts the contents of the file in place of the inline node. This substitution capability makes construction of composited files quick and simple.
Many datasets, dataset visualizations (rendered mostly using IBM DX), and the VRML repository just described are at the following URL: http://www.sdsc.edu/~sdbay/ .
Note that a VRML-based web-browser must be installed and the .mime.types and .mailcap files must be configured to be able to view the San Diego Bay VRML files.
The gypsy moth is an insect that prefers to consume the foliage of trees like oak and aspen. Imported in error into the United States after the Civil War to develop a silk industry, the moth escaped from confinement and began a costly trek across parts of North America. Over the past 100 years or so, this pest has caused millions of dollars in economic loss to the US and Canadian Governments and to the tourism industry, the forest industry and to homeowners who dwell in the path of the gypsy moth outbreaks [Pija95].
Here is an e-mail message sent on 7/24/1996 representing the issue for homeowners-
Dr. Gage,
I live on Dobie Road in Okemos, MI and my (approximately) 200 year old oak in my front yard is, as of yesterday, covered with gypsy moths. Can you please tell me where, how, from whom to seek assistance? My tree is right next to the road, so is easily seen. I would appreciate any help you may be able to offer, before I lose my beautiful tree.
Thank you,
Susan L. Hughes
In Michigan a gypsy moth tracking survey was initiated in 1985 to help government officials and educators determine the locations of gypsy moth populations in the state so they could plan programs of biological regulation, pesticide application and education about gypsy moth and its management. To measure gypsy moth populations, 3,000 traps are set in July each year at the same place to measure the abundance of the moths at each location [Gage90].
A system was developed to map populations of the moths based on these records. One way to determine how the populations are cycling in Michigan is to use visualization tools like Data Explorer to examine trends in abundance.
The resulting images were written to tiff files an example of which is shown below.
A composite mapset from all the images (1985-1995) was developed using imstoryboard [Nade91].
The abundance of the gypsy moth changes from year to year and varies widely for place to place within a year, depending on the forest type, the population history of the moth and the climate. Maps such as those shown in the composite image below are very helpful to plan for future gypsy moth invasions. 
Weather is a key component which defines seasonal dynamics of biological systems. Over the past few years, the midwest corn belt has experienced significant floods and droughts which have seriously affected the potential of the region to produce high crop yields necessary to maintain farm profitability [CAST92]. High variability in heat and moisture within and between years makes forecasting crop productivity challenging. High variability in weather patterns are also implicated as indicators of potential climate change. To study within and between dynamics of weather, we selected patterns of monthly heat accumulation time and space variability to analyze using DX.
The database consists of a summary of daily temperature and moisture measurements made in midwestern states including MI, OH, MN, IO, ND, SD, KS, NE, MO, WI, IL, and IN [Gage91]. Data are summarized at the centroid of each of the 1044 counties in each of the 12 states. The daily data are summarized for 1972-1991. These meteorological data were organized by the USDA CSREES Regional Committee NC94 to allow the scientific community to conduct research on climate dynamics associated with midwest agriculture [Gage94].
To complement this activity, work at the San Diego Supercomputer Center has involved developing the technology and tools to assist in the analysis and summary of this large data set. Monthly heat accumulation from average daily temperature greater than 50F (10C) were computed for each of the 240 months in the 20 year period of record.
Variables in the datafiles include State, County, Crop District, Day of Year (Julian day), daily maximum temperature, daily minimum temperature, and precipitation.
Data Explorer was used to visualize each month in the 20 year dataset. The processing was done as follows:
Header Record for the 240 monthly files
file = w001.img grid = 389 x 528 format = ascii interleaving = record majority = row field = field0 structure = scalar type = float dependency = positions positions = regular, regular, 0, 1, 0, 1 end
The DX process model shown above illustrates the DX modules used to compute each of the tiff images. imstoryboard was used to organize the tiff images into a the composite image. An example single image and the final composited image are shown below.
The methods and procedures developed for the Coordinated Comprehensive San Diego Bay Monitoring Program are applicable to other geographic areas as well. This project is a hopeful glimpse of how information technology will constructively change policy maker's ability to make better decisions about how we maintain the quality of the world we live in.
Models are being constructed in a number of locations around North America to predict gypsy moth population dynamics. Tools like Data Explorer will prove invaluable for visualizing model results and such tools will ultimately aid in decision making.
Further work in heat visualization will include linking meteorological variables to crop productivity and physical landscape characteristics. Data Explorer provided a valuable method to visualize a large dataset containing geographical attributes. Data Explorer was able to handle the 240 grid files (398 x 528) and produce a image set consisting of 240 tiff image files.
[CAST92], Council for Agricultural Science and Technology, "Preparing U.S. Agriculture for Global Climate Change", Task Force Report, Council for Agricultural Science and Technology, 1992, Ames, Iowa.
[ESRI94], Environmental Systems Research Institute, "Understanding GIS : the ARC/INFO method : self-study workbook, version 7 for UNIX and OpenVMS", Environmental Systems Research Institute, 1994.
[Gage90], S.H. Gage, G.A. Simmons, and T. Wirth, "Predicting regional gypsy moth population trend and risk using pheromone trap catch and geographic information systems", Environ. Entomol. 1990, vol. 19, pp. 370-377.
[Gage91], S.H. Gage, "Spatial and temporal crop production in the north central region", NC94 Technical Report, Michigan State University, East Lansing, 1991, 23pp.
[Gage94], NC94 Technical Committee, "Climate and agricultural landscape productivity analysis and assessment in the north central region", NC94 Technical Report, Michigan State University, East Lansing, 1994, 20pp.
[Nade91] D.R. Nadeau, T.T. Elvins, and M.J. Bailey, "Image Handling in a Multi-vendor Environment," In Proceedings of the IEEE Visualization '91 Conference IEEE Computer Society Press, Los Alamitos, Calif., 1991, pp. 276-283.
[Pija95], B.C. Pijanowski, S.H. Gage, and D.G. McCullough, "Policy Issues as they relate to the impacts of the gypsy moth in Michigan", Policy Options for the Michigan Legislature: 1994. Inst for Public Policy and Social Research. Michigan State University Press, East Lansing, MI. 240 pp.
[SDSC95] San Diego Supercomputer Center, "Source code release of webview", http://www.sdsc.edu/SDSC/Research/Visualization /vrml/tools/webview/webview.html
T. Todd Elvins is a staff member at San Diego Supercomputer Center and a Ph.D. candidate in the department of Electrical and Computer Engineering at the University of California, San Diego. His research interests include volume visualization, scientific databases, and parallel and distributed image computing. Author's Present Address: San Diego Supercomputer Center, University of California, San Diego, MC 0505, La Jolla, CA 92093-0505. USA. e-mail: todd@sdsc.edu, homepage: http://www.sdsc.edu/~todd/
Stuart Gage is a professor at Michigan State University and is the director of the Entomology Spatial Analysis laboratory. He has expertise in applied ecology and animal population dynamics and emphasizes spatial and temporal dynamics of organisms at multiple spatial scales. e-mail: 23027shg@msu.edu, homepage: http://www.ent.msu.edu/esal/
John Helly is a senior staff scientist at San Diego Supercomputer Center. He leads SDSC's computational ecology and environmental systems group. e-mail: hellyj@sdsc.edu