Remote Sensing and Environmental Data Visualization in GLOBE Program:
Automated and On-Demand Visualization Using IBM Data Explorer
Shahram Shiri and John M. Cavallo
Abstract
Availability of meteorological and environmental data from government and private sources has offered the public a tremendous opportunity to learn about the environment. However, this data is essentially inaccessible without visualization.
The Global Learning and Observations to Benefit the Environment (GLOBE) program, an international program initiated in the United States, makes this data available to educational institutions and involves students in the environmental sciences. GLOBE students all over the world take environmental measurements daily at their schools and submit their data via the Internet to a GLOBE collection facility. Visualization of the student measurements as well as satellite and remote sensing data are created using IBM Data Explorer software. For this purpose, a series of Data Explorer networks, developed using a visual programming interface, operate in batch mode for the daily production of thousands of images that are made available three times a day via Web servers. After the images are produced by Data Explorer, they are submitted to a post processing system which adds annotations, icons, and other visual elements for better representation in Web pages. As an alternative to the automated production of the images, a series of Web CGI scripts allow the user to select customized views, the desired level of detail, and the data product which are then generated by Data Explorer.
Introduction
In recent years, the availability of remote sensing data throughout various government and private organizations, have provided the potential for many individuals to visualize and map the data in order to better understand the global environmental changes. The advent of world wide web technologies has provided a mechanism for collecting and distributing data over vast geographic regions. The Global Learning and Observations to Benefit the Environment (GLOBE) program uses this data to support a worldwide network of students, teachers, and scientists working together to study and understand the global environment. GLOBE students make environmental measurements near their schools and report their data over the Internet. Scientists use GLOBE student data in their research and provide feedback to the students to enrich their science education. Each day, images created from the GLOBE student data sets and forecast remote sensing data sets are posted on the World Wide Web, allowing students and visitors to the GLOBE web site to visualize the environmental observations.
The GLOBE data visualization server based on a SGI Challenge produces over 5700 images daily for display on the Web. For the production of these images, remote sensing reference data and student measurements are preprocessed and imported to a series of IBM/Data Explorer networks which produce a variety of cartographically projected images. The DX images are postprocessed for Web pages by addition of annotations and color bars and color corrections for Netscape browser.
Data Preprocessing
GLOBE visualization server provides images of environmental data worldwide. Some of the data are student measurements while others are reference data from the Environmental Modeling Center (EMC) or from satellites in Earth orbit. The EMC reference data which are produced by EMC in GRIB format, are converted into HDF (Hierarchal Data Format) and imported into the DX networks. In addition, the text file list of student reporting stations are preprocessed in tabular format readable by general importer of DX. Other reference data such as AVHRR/NDVI are processed in 8 Kilometer resolution and then converted to HDF format.
DX in GLOBE Development
In an evaluation and assessment of available visualization software packages, IBM/Data Explorer was selected over other packages for the production of several GLOBE products. Some of these products are the automated production of processed images, the on-demand web accessible visual images, the automated movie production of data products, and the glossy (poster) image generation. Depending on the product, preprocessing and post-processing of data is performed to enhance the visual products. In addition, we have used DX2VRML module in Data Explorer to produce global models of environmental datasets and port them as VRML geometry for use with Virtual Reality systems. NDVI.199109.wrl shows a VRML geometry of global vegetation index. The following figure shows rainfall from student measurements and AVHRR/NDVI vegetation index, at 1 Kilometer resolution, textured over DMA elevation data, at 30 arc second resolution, for Washington, DC and Chesapeake Bay region.
I. Automated IBM/DX Networks
Using DX Visual Programming Editor a series of Data Explorer networks are created to cartographically project incoming datasets and generate multiple levels of zoom to global regions. These networks in a production environment run as batch jobs where ingest of data and image production run in automated mode. This figure
shows a sample image of reference dataset. The following figure shows a sample image of student measured minimum temperature.
Currently, DX networks process 9 reference datasets and 14 student datasets. A total of over 5700 images are produced in a daily basis. The GLOBE visualization server provide access to over 6300 images for the entire world through a series of Web accessible pages.
II. On-Demand IBM/DX Networks
In addition to automated DX networks, in a prototype attempt, series of on-demand DX networks produce higher level zooms with variety of data products accessible through Web pages. These network can be activated either by filling Forms Web pages which activate DX networks in script mode or running the DX networks in server mode waiting for an update in the configuration file. In addition to above, JAVA applets could register a location in an image and pass that information to DX networks for zooming to that particular location. The following two figures represent the maximum temperature measurements globally and in comparison with forecast data represented as isosurfaces for August 15, 1996.
III. Automated DX Movie Networks
A series of networks where generated to produce movies of student data and/or reference data over a period of time. In a series of movies created for CNN, DX generated the necessary images and Wavefront/Composer was used to compose the final movie segment with titles, color bars, and annotations in script mode. NDVI movie (small and large) show global vegetation index for 12 months of 1991 overlaied on ETOP05 elevation map.
IV. Public Relations Image DX Networks
In addition to movies and on-line production of images, DX was used to produce high-resolution images up to 4000 x 4000 pixels used for high resolution printing for public relation purposes. Again, similar to other categories of GLOBE development, images produced in DX were post processed with titles and color bars. The left image shows the EMC/AVN cloud cover and the right image shows AVHRR/NDVI vegetation index.
Data PostProcessing
The post processing step adds the color scale, explanatory text, and icons to the image to make it more visually interesting and informative. The color scale shows the relationship between the color and value. The text gives the location of the view, the date of the data, the data set name and a brief description of the data. The icons represent the GLOBE program and the data set shown.
The color scales are pre-generated using a DX network that directly uses the color table used in the visualization. Post-script is used to generate the values on the scale, outlines, and tick marks. The icons are supplied by the web page designer.
At run time, Perl is used to control the creation of auxiliary image files, and to composite the elements into a GIF image that is ready for the Web. An image of the explanatory text is generated at run time for each visualization by the Perl script creating PostScript commands which are then rendered to a file. Using PostScript provides accurate control over the font and the location of the text, giving a finished look to the image. The Perl scripts also control a series of NetPBM programs that place each of the elements on the image.
Conclusion
For the past year, IBM Data Explorer was integrated into several categories of GLOBE development. The major concerns in the GLOBE visualization system has been automatic image production and on-demand image generation. We have shown that DX can be utilized effectively in both categories. DX provides the flexibility of allowing networks to be created interactively with Visual Programming Editor and allowing to run them in batch mode in a production environment. For on-demand applications, we modified the automated DX networks to read configuration files which were created from CGI scripts. The configuration files depict the location and the amount of zoom desired for each image requested.
Acknowledgment
Support for this work was provided by NASA Scientific Visualization Studio under Dr. Horace Mitchell and Dr. Richard White in collaboration with Mesoscale Atmospheric Processes Branch under Dr. Fritz Hassler. We wish to extend special thanks to Lloyd A. Treinish from the IBM Data Explorer group for assisting every step of way by providing us with example DX networks and debugging our programs. In addition, Ray Twiddy generated beautiful public relations posters and produced numerous movies for CNN, and Cindy Starr who took time to edit this paper. Also, Jeff de La Beaujardiere, the GLOBE Visualization Server Web Master and Chris O'Handely assisted us on making the pictures available on the Web and generated color maps for the produced datasets.