Towards Unifying Vector and Raster Data Models for Hybrid Spatial Regions Philip Dougherty Overview Introduction Problem Objective Literature & Technology Review Data Table Format Overview of Application Workflows Use Cases Results, Discussion, & Conclusion Spatial Hybrids Defined Hybrid spatial regions may be perceived as both objects and parts of a continuous field. Object properties are best modeled with vector geometry. For Example: Area and Perimeter Internal variation of fields is better represented with the cell-based structure of a raster. These regions may often be associated with a spatial event, process, or phenomenon. Spatial Hybrid Examples Precipitation As Spatial Object: Perimeter of classified regions within the field of precipitation values As Spatial Field: Spatial tessellation of all precipitation values Wildfire As Spatial Object: Perimeter of the wildfire event As Spatial Field: Internal temperature variation of the wildfire event and/or burn severity Hydrological Units As Spatial Object: Perimeter of a given level in a spatial hierarchy As Spatial Field: Internal variation of this level as represented by a lower level of that same hierarchy Problem Within the standard ESRI Geodatabase, hybrid spatial regions are implemented using several database objects. Separate, unrelated database objects store: 1.Vector Geometry and Attributes 2.Raster Data Structure Issues Include: 1.Lack of Entity Integrity 2.Limited Support for Multi-Resolution and Multi-Temporal Modeling Problem, continued Entity integrity ensures that all real-world entities are uniquely identified in the database. Non-GIS Example: All customers in the customer database are uniquely identified by a customer number. All information regarding customers should be accessible through the customer table and/or its related tables. Without this, customer data cannot be confidently queried and retrieved. Problem, continued Because spatial hybrids require dual representation, we cannot achieve entity integrity with the existing data model. Given the database schema, how do we uniquely identify a wildfire event? How do we retrieve all the data for a wildfire event? Objective Integrate the vector and raster representations of hybrid spatial regions into a single database table or properly related tables. In Order To: Maintain Entity Integrity Provide a Platform for Multi-Resolution and Multi-Temporal Mapping and Analysis Simplify Data Retrieval and Query Literature Review Support: Concept and definition of spatial hybrids Multi-resolution and multi-temporal mapping and analysis as key element in modeling spatially dynamic regions Concerns: Preserving entity integrity Formalizing a framework for identifying hybrids and modeling them at different scales and/or time periods Research did not show a major concern with the technological component of implementing a formal framework. Literature Review, continued Conceptual Object-Oriented Representation of Environmental Phenomena: Is Everything Best Represented as an Object? Dual nature of spatial regions is not adequately handled with traditional geographic data models. Stratified Map Spaces: A Formal Basis for Multi-resolution Spatial Databases Map spaces allow for translation between maps representing the same extent at different levels of detail. Bridging Vector and Raster Representation in GIS Definitions of hybrid representation Representing Complex Geographic Phenomena in GIS Suggests wildfire use case database DESIGN, APPLICATION, DEVELOPMENT, & ADMINISTRATION Entity integrity in relational databases The MurMur project: Modeling and querying multi-representation spatio-temporal databases Representations of Geographic Space: Toward a Conceptual Synthesis Annotated References Literature Review, continued Technological Modelling geographic data with multiple representations Model for spatial data with multiple representations that uses Oracle spatial data types Extending geographical representation to include fields of spatial objects Each row in a data table is a field location and each column is an associated spatial object. A framework to enhance semantic flexibility for analysis of distributed phenomena A cluster of grid cells comprise a zone, the geometry (boundary) of which is stored in a table as a run-length encodes (RLE). Shapefiles are generated on the fly based on these RLE. Annotated References Technological Concept & Technologies Technological Concept: Persistence A software object is moved from computer memory to persistent storage Technologies: Binary Large Object (BLOB) ESRI ArcObjects Microsoft.Net Framework Technological Concept & Technologies A binary large object is a collection of binary data stored as a single attribute value in a database table. Both enterprise and MS Access databases support binary large objects. ESRI technology is supported by the BLOB format. Specifically: The standard ESRI Geodatabase already makes use of BLOBs. ArcObjects provides for the creation of BLOBs. Hybrid Format Defined There are two storage / representation methods for hybrid tables: Explicit Vector geometry and raster array are explicitly stored as BLOBs in hybrid table(s). The source feature class and raster dataset are no longer needed with this format. Implicit Feature and raster layers are stored as BLOBs in hybrid table(s). The source feature class and raster dataset must remain in the database but are linked through hybrid table(s). Hybrid Format Defined, continued Implementing Hybrid Table Values of field with data type BLOB can be set with a MemoryBlobStream object. Creating a MemoryBlobStream Object Two approaches for ArcObjects that implement the IPersistStream and/or IXMLSerialize interface. User Interfaces User InterfaceApplicationDescription Dialog Derive Hybrid Representation ArcCatalogCreates hybrid from a single input raster dataset. Dialog Extract Hybrid Representation ArcCatalogExtracts hybrid by polygon from a raster Dataset. Dialog Hybrid from Related Vector & Raster Data ArcCatalogCreates hybrid from associated vector and raster data. Dialog Display Hybrid Data ArcMapAllows hybrid data to be added to map display. Tool Identify Hybrid DataArcMapUsed to identify hybrid entities. Workflows Derive Hybrid Extract Hybrid Relate Hybrid Hybrid Data Display Use Cases 1. Thirty-Year Annual Precipitation near Yellowstone National Park 2. Southern California Wildfires 3. Trinity River Basin 4. One-to-Many: Hydrological Units Use Case #1 - Yellowstone ArcCatalog Use Case #1 - Yellowstone ArcMap Use Case #1 - Yellowstone ArcMap Use Case #2 - Wildfires ArcCatalog Use Case #2 - Wildfires ArcMap Use Case #2 - Wildfires ArcMap Use Case #3 Trinity River Basin ArcCatalog Use Case #3 Trinity River Basin ArcMap Use Case #3 Trinity River Basin Object-only display and ID Use Case #3 Trinity River Basin ArcMap Use Case #4 One-to-Many This relationship allows for multi-temporal / multi- resolution modeling. Results and Discussion Advantages Hybrid data model maintains entity integrity for spatial regions. Provides a framework for multi-temporal / multi-resolution mapping. Improves the efficiency of data retrieval and query by integrating or linking several datasets. Limitations Performance liability with very large raster datasets in the explicit storage method. Multi-band raster datasets are only supported with the implicit storage method. Conclusion Given the advantages and limitations stated previously, all aspects of the objective were met. Refinement and testing would allow the application to be implemented in an organizational environment. Future Work Implement the logic to allow for data editing and updates. Changes to one of the representations would be automatically propagated to its opposite.