GIS, Environmental Model Integration and

Environmental process models describe the exchange of energy and materials through space and time and assess their impact on the environment. Many environmental process models are closed monolithic systems that cannot be easily modified. When dealing with increasingly complex environmental problems, multiple models are often involved in a single project, but only parts of each model are relevant. An ideal approach is to integrate only the relevant parts into a customized model or an interoperable model.

The interoperable modeling approach may fundamentally change the way environmental models are developed and used, as well as the way they integrate with geographic information systems (GIS) and other tools. This development is supported by both the research community and government agencies and is foreseen as an inevitable direction for future generations of environmental process models.

Interoperation allows data and functions to be used freely across systems. At the core is the semantic interoperability, which allows for meanings of data and functions to be shared between users. Computing techniques are already available to support interoperable models, such as component-based modeling. Components are independently developed, ready-to-use software units. They can be assembled to form functional systems. The challenges lie in the conceptual issues, including the principles to guide the delineation of components, the meta-information needed to describe a component, and the semantic compatibility between components.

The first issue is to identify the primitive components, the building blocks of an interoperable model. Because the existing process models embody accumulated knowledge and time-tested coding, it is sensible to respect these developments and identify the common elements contained in them in order to delineate the primitive components. The semantic reference system can serve as a conceptual framework to guide the delineation. This system consists of (a) a semantic datum that contains the most basic terms and their meanings, (b) a semantic reference frame that organizes the semantic datum, and (c) translation functions that relate a concept in an application to the semantic reference frame and annotate the concept by the semantic datum.

Methods such as formal concept analysis can help translate the concepts (e.g., equations, processes) in heterogeneous environmental models to the discipline datum, identify the common elements embedded in the local concepts, and subsequently identify the primitive components.

Several types of meta-information are needed to describe a primitive component. One type is spatial and temporal scale, such as spatial and temporal continuity (i.e., discrete vs. continuous), spatial and temporal extent, and spatial and temporal resolution. Another type of information is the scale of processes, represented as entity hierarchy, process hierarchy, and the relationships within each and between them. A third type of information addresses parameters and their values for environment conditions, initial conditions, and boundary conditions. The fourth type is concerned with model development, such as information about model creation, validation, and referencing.

The third issue, semantic compatibility between components, addresses whether environmental process components can be assembled together. The meta-information associated with each component is critical for this step. A number of approaches can be used to evaluate the semantic similarity between components. These include trees and conceptual graphs and associated methods to evaluate compatibilities.

...