Key problem areas within the earth, space and environmental sciences today often involve the integration of large-scale remote sensing and in situ observations with supercomputer simulations. The models are typically empirical in nature, using such observational data as input. Of course, support of large-scale data processing and analysis of remote sensing data and more traditional modelling is still required. Hence, visualization solutions must be able to address the inconsis-tencies and irregularities associated with observational data, and their (visual) integration with (gridded, well structured) output from computer models as well as scale to potentially very large data sets. From the Data Explorer perspective, the ability to annotate qualitative and quantitative visualizations, support of irregularities in the observational data (e.g., miss-ing data, use of statistics, etc.), geographic coordinates, probing to map the image back to the data handling multiple data types simultaneously, and easy generation of animation sequences are critical. These programs illustrate examples of the class of data visualizations that can be generated by IBM Visualization Data Explorer for correlative data analysis in the earth and space sciences. The provided form of the data in most of these demonstrations (a rectilinear geographic grid) is ill-suited for the study of phenomena that occur continuously over a nom-inally spherical surface (i.e., it tears the data). Hence, cartographic techniques are introduced to suitably deform the data via a toolkit developed for Data Explorer. Traditionally, if you are going to do such a transformation, you define a new carte-sian grid in the cartographic projection coordinate system, and then interpolate from the original rectilinear grid to the new one prior to any realization. Given the curvilinear nature of the transformation, non-linear interpolation techniques are typically required to make the transformation of acceptable quality. In addition to typically being computationally ex-pen-sive, such interpolation may make it difficult to preserve the data fidelity. Data Explorer dispenses with such steps since it directly supports operations on deformed grids. In this case, the grid structure itself is transformed without affect-ing the data! Many of these programs show a different example of such techniques with disparate data. For more information about this approach and some specific examples, you should refer to Visualizations Illuminate Disparate Data Sets in the Earth Sciences, Computers in Physics, 8, n. 6, pp. 664-671, November/December 1994. All of the examples and applications described herein are available publically, via anonymous ftp at the Cornell University Theory Center, ftp.tc.cornell.edu. These tools are also accessible from the World-Wide-Web, starting at the Data Explorer home page via the URL http://www.almaden.ibm.com/dx and on the Data Explorer BonusPak CD-ROM. You NEED to have these tools and data to reproduce the examples discuss herein. The tutorial networks and images corresponding to Figures 1 through 25 are in applications/cartography/tutorial on the CD-ROM, the data are in applications/cartography/data and the macros are in applications/cartography/macros. The example applications plus a number of others are are in applications/environment, applications/meteorology and applications/space.
To begin using Data Explorer for geographic-based data from typical earth, space or environmental science problems, start Data Explorer with either the DXMACROS environment variable or the -macros option pointing to where you have placed the cartographic tools. You also need to ensure that the DXDATA environment variable or the -data option points to where you have the data associated with these tools. The dx_demo script provided with the tools is a c-shell script that shows one way of doing this. You should also familiarize yourself with the basic operations of Data Explorer by trying some of the simple examples discussed in the Tutorial Video Tape and the User's Guide. All of the figures used are screen images captured while Data Explorer was executing. The first set shows how applications can be built using the Visual Program Editor and a set of tools that support cartographic warping and other techniques for both two-dimensional and three-dimensional data. The second category shows actual applications built with Data Explorer that only expose Control Panels and Image windows. If you do not want to try this (or any of these exercises manually), simply select the appropriate network (e.g., figure1), from the Open File dialog box, and hit Execute-Once. There is a similar network for each of these examples. All of these example networks have been saved in software rendering mode for a 24-bit display. If you are operating Data Explorer from an 8-bit X-server, you will receive a warning message to that effect. What this means is that the displayed images will use 8 bits of color, which have been dithered from the original 24-bit resolution. If you are operating Data Explorer on a workstation with hardware support for three-dimensional graphics, you may wish to take advantage of that configuration. If so, then in the Image window under Options, simply select the pull-down menu for Rendering Options, and click the button labelled hardware. If you now hit Execute-Once, the image will rerender with graphics hardware.