Mapping the World By Damon Judd The Defense Mapping Agency (DMA) of the United States Department of Defense is primarily responsible for providing maps of various parts of the world in support of America's armed forces. The Vector Product Format (VPF) is defined by a Department of Defense Military Standard (MIL-STD-2407, July 1993), and represents a new data model for digital vector products to be built, maintained, archived, and distributed by DMA. Recently, NATO adopted VPF as a standard for the international exchange of digital spatial data. What is VPF? As part of a contract to develop the Digital Chart of the World data product for DMA, Environmental Systems Research Institute Inc. (ESRI) was also asked to develop a new data format to support a variety of mapping, charting, and geodesy applications. VPF is the result. It is a direct use, machine-independent database format for vector data. VPF is designed to allow applications software to read data directly from the storage medium (CD-ROM or magnetic disk) without the need for an intermediate conversion step. This design goal requires the incorporation of tables and indices in the VPF database that support direct access of locational and thematic information. Because of this unique spatial indexing requirement, VPF databases are capable of supporting real-time mapping and charting applications, and provide for rapid updates using GPS equipment. VPF is a standard format, structure, and organization for large geographic databases that are based on a georelational data model. The VPF data model has five levels. The lowest level of a VPF database consists of cartographic primitives (edges, faces, nodes, and text). Primitives that are related to thematic information, such as surface hydrology or transportation, are called features. Primitives and features together form coverages which are stored in libraries using a pre-defined tiling scheme. A set of libraries for a specific geographic area of coverage make up a VPF database. Converting Paper Maps Many of the existing maps at DMA's mapping center were compiled using manual cartographic methods over the last several decades. There are several source map scales of varying quality and with varying levels of content, accuracy, and detail. DMA has negotiated contracts with 10 data conversion consultants to systematically convert the paper maps to digital map databases in the VPF data structure. By the year 2000, DMA plans to convert over 11,000 individual map sheets to digital files. The conversion process is a complicated series of translation steps with rigid quality control requirements. Because ESRI developed the VPF data structure, the ARC/INFO software contains the commands necessary to perform the final conversion to VPF. However, several processing steps are required before the final VPF conversion can be completed. The overall production workflow consists of the following general steps: 1. Prepare source documents for scanning. The source documents generally include film negatives of the individual color map separates that are used to produce the lithographs. It is usually preferable to create mylar positives of the linework for each individual map separate before scanning. Also, some manual preparation of the source materials can greatly reduce the amount of graphic editing required after the scanning. 2. Scan the source materials to produce raster files of the graphics primitives. Perform a raster to vector conversion to generate vector graphic primitives of the raw source data. 3. Clean up (edit) the vector primitives, digitize any graphics primitives that may have been missed during the scanning step, and begin assigning attributes to create features from the primitives. Some of this work can be automated, but much of it is manual and labor-intensive. 4. Perform coordinate transformations to convert features from the native map source projection (e.g. UTM) to decimal degrees in the WGS 84 projection and edge-match to adjacent map sheets. 5. Insert appended, edge-matched, projected coverages into libraries. 6. Perform the VPF translation. The VPF translation step (VPFEXPORT) takes the contents of the libraries and combines the spatial data, attribute data, and some additional metadata (conversion control files, value description tables, and geographic reference tables) to build the final VPF database. It also builds the spatial index during this step. Quality Control and Review VPF is a very complex data structure and DMA performs a thorough review of any VPF database that is delivered to them before accepting it. For these reasons, stringent quality control (QC) is a fundamental requirement for the map conversion process (Note: See the Down to Earth column in the November 1995 issue of EOM for a discussion of this topic). Because errors may potentially be introduced at each conversion step, quality control measures are performed at every possible opportunity. For example, the scanned raster files are carefully compared against the original source materials. The vectorized primitives are plotted and checked against the original paper maps. Features that are assigned attributes are plotted with different colors and shading patterns to verify correct attribute assignments. Cross-tile topology - essentially the proper mathematical description of a line-string or area that crosses a tile boundary - is carefully checked to ensure that the edge-matching and append process did not introduce any errors. Some tools are generally available for performing certain aspects of the quality review process, but many tools must be customized depending on the specific conversion procedures used. ESRI has developed a set of tools and procedures for performing the conversion process and for automating some of the quality control steps using ARC/INFO and AML. Other consultants are developing their own procedures based on the production methods they are using. At Rust Environment & Infrastructure, production processes and a quality assurance program for VPF data conversion are being developed and refined for the production of Digital Nautical Chart (DNC) and Vector Smart Map (VMAP) libraries. The production philosophy is aimed at reducing errors in the final product by implementing a comprehensive quality assurance plan. In addition to the automated QC procedures mentioned earlier, project supervisors monitor a selected subset of graphic and attribute features that are being processed on a daily basis. Any identified errors are analyzed for the cause of the error, then a corrective action, additional training, a clarification of the specific requirement, or a change in the production process is implemented to prevent that type of error from occurring again later on. This procedure helps to identify any potential shortcomings in the project work flow and prevents additional errors from being propagated by the production processes. Improving the Tools In addition to the strict quality assurance procedures that are continuously implemented, there are also programmatic QC checks of the data being performed at various stages of the production work flow. Quality control validation software for VPF data that Rust is developing includes components to: check file structure integrity; verify that topologic relationships are accurate and complete (links exist for right and left edges and for right and left faces); verify that attribute links exist for each feature record and that feature links exist for each attribute record; checks for duplicate records (i.e. checks that no 2 primitives occupy the same space); checks for valid attribute values including valid null or unknown values; and checks that the complete spatial extent of all relevant coverages in a library are accounted for including the universal area. These tools and others are continually updated and revised as maps for different parts of the world are processed, with each area presenting a unique set of requirements and challenges to overcome. By sometime early in the 21st century, DMA will have the capability to map any part of the world in real-time using VPF data converted from paper maps. The educational and scientific communities will be only a short step behind in their adoption of VPF and in the development of new methods and tools to support civilian applications of this new data structure. That is truly something to look forward to. About the Author: Damon Judd is manager of GIS and computer applications for RUST Environment and Infrastructure in Broomfield, Colo. He may be reached at 303-469-6660 (phone) or 303-469-6665 (fax). Back |