A SOFTWARE-DEVELOPER'S PERSPECTIVE ON INTEGRATING SPATIALLY EXPLICIT MODELS

Spatially explicit models tend to be both data-intensive and computation-intensive. For this reason such models are generally and necessarily implemented with the complex computer software required to carry out millions of processing iterations over voluminous geographic rasters, sets of agents, components of physical systems, numerous time steps, and multiple Monte Carlo trials. Thus the problems of integrating or coupling distinct, spatially explicit models include integrating the models in software and can be neither fully understood nor solved without due consideration for the constraints and possibilities of model integration in software. The methods, materials, tools, and concerns of the coupling and integration of spatially explicit models, particularly of the integration of land-change models with physical environmental models, will be surveyed and discussed from the perspective of a software developer. The discussion will distinguish coupling from integration and will cover modeling languages and frameworks, operating environments, inter-system and inter-process communication, points of contact among models, software-development costs and constraints, design trade-offs, and computational efficiency.