Home Projects Multiscale Databases and Resolution

Multiscale Databases and Resolution

• Multi-Resolution Raster Data for The National Map
• Fractal and Variogram Analysis of Scale and Resolution Effects in Geospatial Data

Additional information and project materials

Objective: Research and develop a Web implementation for fast and accurate reprojection and resampling of raster data for The National Map.

History: Originally funded by CTM as a 3 year project now in the second year.

Accomplishments: We have implemented the USGS mapimg software package on a variety of platforms, including Windows, linux, and Solaris. This software package is designed specifically for global projections of raster data, and solves many of the problems inherent in commercial offerings including areal and categorical distortion of raster cells and wrap-around effects. The software includes new categorical resampling methods that allow significantly better preservation of categories when downsampling, and includes a resampler for data of counts, such as population numbers. We have developed an object-oriented version of the General Cartographic Transformation Package (GCTP). This software was the subject of a presentation to the GIScience community at the International Society for Photogrammetry and Remote Sensing Commission IV Symposium on Geospatial Databases for Sustainable Development, Goa, India (September 2007).

Results and Outputs: We previously have presented and published several papers in refereed journals concerning this effort, which is a continuation of work on projections that began under Prospectus funding several years ago. In the past year, we have delivered open-file reports, user manuals, and currently (2007) are preparing another paper for submission to a refereed journal. We have open source software with test datasets available for mapimg ( http://carto-research.er.usgs.gov/projection/index.html ) and will soon have the object-oriented version of GCTP available to the public from this site.

Current Status (2007): We have made excellent progress with global data and fundamental development of projection and resampling methods and software.

Planned Future Work: We are proceeding with this project by adapting the mapimg software and resampling algorithms to datasets for The National Map, developing the capability to apply the software in a Web environment through a Web browser, and incorporating the ability to manipulate regional geospatial data (as is required by a state partner using a specific Universal Transverse Mercator (UTM) zone). We are now examining Open Geospatial Consortium (OGC) compliant software for rapid projection across the Web using grid computing technology.

Anticipated Timeline and Resource Requirements: We anticipate that 2 additional years will be required to accomplish method development and adequate testing of the planned future work described.

Top

Objective: Fractals and variograms are established methods to determine effects of scale and resolution in geospatial phenomena and processes. We propose to use these methods to examine the impacts of scale and resolution on data integration and generalization for The National Map and the National Spatial Data Infrastructure (NSDI).

Background and Approach: Building on our earlier work with scale and resolution from the Automated Data Integration and Generalization projects, and applying fractal and variogram analyses to specific datasets, we hope to establish analytical methods to achieve data integration and generalization limitations and parameters. We have conducted preliminary analysis of fractal methods to interpolate terrain elevation from a set of seed values, or ground-control points. We also have worked with lidar data and high resolution DEM's for detection of geologic structures such as meteorite impact craters. Dr. Shuo-sheng Wu, a post doctoral student, joined us in August 2006 to conduct research on variogram analysis of scale effects. This project is to bring these various research threads together to analyze specific effects of scale and resolution in data for The National Map. We will examine elevation, land cover, orthographic images, and hydrography data at various scales and resolutions to establish fractal dimensions of these datasets and their interrelations. We will develop variograms to determine spatial autocorrelation effects and optimum scale parameters for the combination of these datasets. With the results of the fractal and variogram analyses, we hope to establish the optimum scales at which these datasets can be integrated and fused into a single geospatial product.

Discussion: We see this as a pilot project to begin an analytical approach to accomplishing data integration and generalization for The National Map. This project is specifically focused on the raster datasets, elevation, land cover, and images and includes vector-based hydrography because of its close association with terrain elevation. With our previous work on resolution effects on watershed analysis, data integration, and generalization, and the planned work by Dr. Wu on variogram analysis, we think it is essential to use previous work to establish analytical methods to handle multiple resolution datasets in The National Map and determine capabilities for accomplishing a seamlessly integrated product in positional and thematic integration. This proposed project will be connected with the continuing projects on data integration and generalization, providing analytical tools and methods to accomplish the objectives of those projects.

Top