An Object-Oriented Integrating Framework for Multi-discipline Ecosystem ModelingEPA Grant Number: R827959
Title: An Object-Oriented Integrating Framework for Multi-discipline Ecosystem Modeling
Investigators: Stotts, David
Institution: University of North Carolina at Chapel Hill
EPA Project Officer: Packard, Benjamin H
Project Period: October 1, 1999 through September 30, 2002 (Extended to July 20, 2005)
Project Amount: $863,049
RFA: Computing Technology for Ecosystem Modeling (1999) RFA Text | Recipients Lists
Research Category: Environmental Statistics
We propose to construct an object-oriented architecture and integrating framework for multi-discipline ecosystem models (MDEM). The framework will comprise 3 layers: an abstraction layer for problem definition and multi-discipline model instance specification; an OO model components layer and OO architecture for interoperation of components; and a data layer. At the definition level, the problem to be solved will be described with abstract specifications indicating which model components are needed (atmosphere, soil, hydrology, estuary, etc.) and which data sets are to be executed, as well as what the investigation goals are. The components layer will contain the individual science models wrapped in classes that will resolve spatial mismatches and temporal mismatches among them; a major problem in linking disparate multi-media models. We will construct compilers and other support tools to generate from the problem definition the objects needed at run-time to properly sequence the possibly thousands of model or component executions needed over the indicated data sets. The data layer will contain geographic information to drive the model components, mostly organized in a GIS systems currently.
We will construct the integrating framework using OO design patterns, which have been developed as collections of classes and their interconnections best suited for constructing large systems that are dynamic, maintainable, and extensible. Patterns do this by heavy use of delegation (allowing object replacement at run-time) and light use of the more static concept of inheritance. Use of modern design patterns will allow the framework to be evolved by many members of the modeling community through open source methods. We will specify the interfaces for the individual science modules, as well as for the framework itself, in an interface description language (IDL) that is independent of any specific programming language. This IDL will most likely be CORBA, an international standard for OO systems. The framework will also be documented as UML specifications.
Our physical test and prototyping problem will be taken from the EPA's Models-3 program, initially, which was a modular based system for air quality management. From our initial prototype, we will expand to specify and link cross-discipline components being investigated by other members of the UNC faculty, namely, water quality and estuary modeling. The Carolina Environmental Modeling Community is a multi-disciplinary team of researchers investigating the management, emittance, movement, and effects of nutrients into a complex natural ecosystem. We will leverage this work to the greatest extent.