Integration of GIS with Dynamic Modeling Dali Wang and Nick Buchanan Department of Computer Science University of Tennessee Knoxville, TN 37996 {dwang,nbuchana}@cs.utk.edu “Computational Science for Natural Resource Managers” Workshop University of Tennessee, Knoxville April 19-22, 2006 Presentation Outline: Two Types of Integration with GIS Integrating ArcGIS with Ecological Modeling A Cross-Platform Ecological Modeling Developing ModelViewer for Ecological Modeling ModelViewer usage demo Some Info on ESRI Virtual Campus Integration with ArcGIS Data Integration Raw Data / Meta Data User’s own data format Write own code / ArcGIS Engine APIs ESRI format (proprietary) ModelViewer (Customized application) ArcMap or Desktop Analysis Tools Standard environment Development/customized environment Integration with ArcGIS Model Integration Operating System Windows/Solaris/Linux Language Java/C/C++/Fortran Location Local/Remote ArcGIS Engine Java APIs is our first choice OS independent Easy GUI design and configuration Network programming Integrating ArcGIS with Ecological Modeling Overview software architecture for an ecosystem modeling package Across Landscape Fish Model: A case need cross platform simulation Ecosystem Modeling Package ATLSS (Across Trophic Level System Simulation) ATLSS is a multiscale ecological multimodel designed to assess the effects on key biota of alternative water management plans for the regulation of water flow across the Everglades landscape. Component –based ecological multi-modeling package Landscape library with some essential GIS functionality High-performance oriented software design Overview Software Architecture Across Landscape Fish Model ALFISH is an Intermediate Trophic Level Functional Groups Model to produce projections of fish biomass – food resources for upper trophic level consumers (wading bird) across South Florida. Challenges Input Data (including water data 35*12*6 layers, each layer has 110,000 cells) Computation (36 hours on a Solaris Computer, need parallel computing) Representation (Users (managers) prefer Windows (ArcGIS) user interface and analysis tools) Simulation procedure Back-end computer Communication services Input maps 1 ModelViewer PALFISH model JobGenerator Input files 7 FileTransferServer 8 2 5 output files Engine JobLaunchBar JobMonitor 4 10 Initialization EmailClient MapGenerator 9 GIS-enable user interface Front-end computer 3 Finalization output files JobGeneratorProxy 6 FileTransferClient Communication services The model simulation you scheduled on xxx.tiem.utk.edu has completed running. Start time: Sat Sep 3 10:22:13 EDT 2005 8 processors used End time: Sat Sep 3 12:36:53 EDT 2005 Model output files have been transferred to a directory named "Mode_output" in your TIEM home area. Some Findings Multiplatform computing is possible (ArcGIS on windows, dynamic modeling on Unix/Linux) Using ArcGIS Engine, we can develop customized applications to use ArcObject functionality on variety of computational platforms (not limited to Windows) ArcGIS Engine provides potential for parallel computing using ArcObjects Part 2: ModelViewer and Virtual Campus Goals of Developing ModelViewer To act as a “front end” to preexisting models as well as sets of communication services associated with the models (launching models on multiple processors, etc.). Allow users to execute code to convert data into an easily viewable format with maximum transparency. Give users operating under Unix-based platforms the ability to view ESRI GIS data formats. Give users operating under Windows high performance multimodeling capabilities. ModelViewer Structure User interface developed in Java, which can be run on any platform (Windows, Linux, etc.). GIS components added to interface using ESRI’s ArcGIS Engine API. ArcGIS Engine shares the same foundation as ArcDesktop (ArcObjects) and thus much of the functionality. Based upon ESRI example code. ModelViewer Use Model instantiation Simulation started based upon a configuration file. Results written to predetermined location. Data visualization Model output converted into raster layers in a map file. The map file can be then viewed and manipulated using an ArcDesktop-like suite of tools. The user has the ability to save output in format (*.mxd) readable by any ESRI-capable application. ModelViewer Demo using DEER DEER is one of a family of landscape-based, spatially explicit species index models developed to assess the impact of management scenarios on habitat conditions of different species in South Florida. Not a complicated model, only require sequential computing Give a first-hand feeling of using ModelViewer Simulation procedure Communication services Input maps Deer model JobGenerator 1 3 JobMonitor ModelViewer 2 JobLaunchBar 5 MapGenerator GIS-enabled User Interface Initialization 4 Progress bar 7 Input files 6 Engine Landscape Finalization output files ESRI Virtual Campus http://campus.esri.com/ Creating and Integrating Data for Natural Resource Applications Four modules: Data Acquisition Data Creation Data Integration and Error Assessment Preliminary Data Analysis Data Acquisition Various web sources of geographic data Data Preprocessing Assigning coordinate systems, metadata, etc. Vector vs. Raster: computation, storage, context File formats http://www.esri.com/software/arcgis/extensions/datainte roperability/about/supported-formats.pdf Numerous conversion tools available Data Integration / Error Assessment Working with: Multiple coordinate systems Multiple data resolutions Raster and vector formats together Error Source Identification Detection using satellite imagery Conclusion Always helps to ensure reasonable understanding between parties involved. Common ground is the cornerstone of an efficient project. The better the specification, the better the result. CS related reference: A GIS-enabled Distributed Simulation Framework for High Performance Ecosystem Modeling, ESRI International User Conference, June 1-5, 2006, (accepted) A Parallel Structured Ecological Model for High End Shared Memory Computers, First International Workshop on OpenMP, 2005, Lecture Notes in Computer Science (in press) A Grid Service for Natural Resource Managers, IEEE Internet Computing, pp20-26, Jan/Feb 2005 On Parallelization of a Spatially-Explicit Structured Ecological Model for Integrated Ecosystem Modeling, International Journal on High Performance Computer Applications (in press) Toward Ecosystem Modeling on Computing Grids, IEEE Computing in Science and Engineering, Vol. 13, No. 1, pp55-76, 2005 Parallel Fish Landscape Model for Ecosystem Modeling on a Computing Grid, Simulation: Transactions of The Society of Simulation and Modeling International (accepted) The Design and Implementation of Parallel Fish Model for South Florida, Proceedings of the 37th Hawaii International Conference on System Sciences, 2003 A parallel Implementation of ALFISH: Compartmentalization Effects on Fish Dynamics in the Florida Everglades, Simulation Practice and Theory, Vol. 13, No. 1, pp55-76, 2005