Ling Bian
Department of Geography
SUNY, Buffalo 14261-0023
lbian@geog.buffalo.edu

Environmental models, especially process models, simulate the natural process of energy and material exchange over space and time (Maidment, 1993; Kemp, 1993). Connected from a spatial perspective, the integration of environmental modeling with GIS has b een a necessary step in the development of both disciplines. The progress made in the past decade has greatly benefited GIS as an emerging science and environmental modeling as an established discipline. Despite the apparent success, many issues of integr ation remain as obstacles for further, in-depth integration between GIS and environmental modeling. Many of these issues are fundamental, especially at the dawn of Open-GIS era. This paper attempts to address several of these issues under a greater framew ork of geographic information interoperability. As an important user of GIS data and geoprocessing functions, environmental modeling is an inseparable element of the interoperability.

Integrating GIS and Environmental Models

When environmental modeling embraced GIS, it took great advantages of spatial data and spatial analytical tools offered by GIS. With the benefits there came a series of integration issues because GIS and environmental modeling have different scientific f oci and each took unique development route in different evolution time frame. Research in the past has dealt with integration issues at several levels, from simple data translation to more integrated coupling. Various conceptual integration models have be en proposed and implemented (Abel et al. 1994; Chou and Ding, 1992; Nyerges, 1993, Bian et al., 1996). However primitive or sophisticated, these integration efforts focus on sharing data without addressing the incompatibility of the basis of GIS and the m odels.

The incompatibility between the two systems is beyond the issue of data format. Because the development of environmental modeling peaked prior to the development of GIS, environmental model development had to cope with the lack of effective spatial means  by taking simplified assumptions of spatial variation. In the GIS era, many such models inevitably under-used the rich content and capability of GIS because the limitations in using spatial data. However, a more critical issue lies in the simplified spat ial assumptions the models use. The spatial assumptions are intertwined with other assumptions such as temporal assumptions. Conceptually, it is possible to create a more flexible basis to adjust the incompatibility between the model assumption and the sp atial data. Practically, overcoming this incompatibility is not an easy undertaking partially due to the structure of most process models.

Environmental Models

Similar to monolithic GIS packages, many environmental models are closed, stand-alone systems. They require fixed input/output data format with virtually no capability to interface with other systems. Many of these models are still in batch mode without modularized structure. Modifications to a model are often handled as patches added to the main body of code developed decades ago. Such a tightly wrapped structure may have been the primary cause that limited the integration to the level of "coupling", in stead of a full integration.

In contrast to the weak spatial component, process models are well developed in simulating physical processes and handling temporal variation. Process models focus on clearly identifiable entities or phenomena. The processes taken by or posed upon the en tities and phenomena are represented as mathematical functions. The state of the entities and phenomena, as initial conditions and especially the results of the processes, is the ultimate interest of the modeling. The dynamic nature of the processes is im plemented by the state change of the entities and phenomena over a series of explicit time steps. Spatial locations of the entities and phenomena, however, may not play an active role in the dynamic process.

Current GIS Systems

Unlike process models, geographic locations are explicitly represented in GISs and they form the conceptual and structural basis of GIS data system. All other properties of the entities and phenomena are attached to the locations. However, such a system is largely static with virtually no capability to handle temporal changes. This spatial-temporal incompatibility between GIS and process models requires a more revolutionary change in the basis of both systems.

Further from the issue of spatial-temporal incompatibility and closer to the core of process modeling is the dynamic nature of the processes. In addition to change in time, entities move (wildlife) and so do many phenomena (precipitation, at a proper sca le). Neither current GISs nor the process models can explicitly represent the spatial dynamics of the processes. While process models have a weak spatial component, current GISs are too rigid to accommodate frequent location changes.

Object-Orientation

In searching for a more effective model to represent the dynamic world, Peuquet and Duan (1995) proposed a system that is based on time. Changes to the location or other properties are attached to the framework of a time-line. Among many proposed systems , the approach that is repeatedly proposed as a better solution for modeling the dynamic world is the object-oriented design, Tang et al. (1996) proposed a system based on geographic features, in which the semantic feature objects form the basis of the sy stem. Geographic locations are properties of the geometric objects that are encapsulated by the semantic features. A similar design philosophy was presented by Takeyama and Couclelis (1997) although applied to a cellular automata system.

Raper and Livingstone (1995) outlined another design for modeling natural processes. The design bases the representation of real world on form, process, and material objects. Geographic location and time are treated as properties of the objects. Both fea ture-based or time-based designs allow easier handling of spatial and temporal dynamics of the entities or phenomena. These designs, and especially the ones that focus on features in dynamic progress is particularly appalling to modeling processes. This d esign is consistent with the vision of Open-GIS (Buehler and McKee, 1996).

Object-orientation is perhaps the most effective framework that can house both GIS and process models in one compatible system. With this framework, the entities and phenomena of interest to process models form the essential objects. The objects are rela ted through associations. Geographic location and time are the properties of the objects. The resultant easy update of location and time allows an effective simulation of spatial and temporal dynamics. The processes can be explicitly implemented as events  that lead to state change of an object. Issues such as incompatibility in data resolution, spatial-temporal handling, and dynamic simulation can be adjusted with flexibility within this framework.

From an implementation perspective, object-orientation supports re-use of object libraries, effective spatial-temporal query, easy interfacing with visualization, and flexible customization. These technical advantages support the realization of component ware, a concept and practice that is foreseen as the future form of Open-GIS, as well as for future environmental modeling.

References

Abel, D.J., and Kilby, P.J. (1994) The systems integration problem. International Journal of Geographical Information Systems, 8(1):1-12.

Bian, L., Sun, H., Blodgett, C.F., Egbert, S.L., Li, W., Ran, L. & Koussis, A.D. (1996) An integrated interface system to couple the SWAT model and ARC/INFO. Proceedings, the Third International Conference on Integrating GIS and Environmental Modeling, Sa nta Fe, New Mexico (January 21-25, 1996) World Wide Web and CD ROM publications.

Buehler K. and McKee L. (1996) The OpenGIS Guide. http://www.opengis.org, The Open GIS Consortium, Inc.

Chou, H.-C., and Ding, Y. (1992) Methodology of integrating spatial analysis/modeling and GIS. Proceedings, 5th International Symposium on Spatial Data handling, Charleston, South Carolina, 514-523.

Kemp, K. (1993) Environmental Modeling with GIS: A Strategy for Dealing with Spatial Continuity. Technical Report 93-3, National Center for Geographic Information and Analysis, Santa Barbara, USA.

Maidment, D. R. (1993) GIS and hydrologic modeling. in Environmental Modeling with GIS, Goodchild, M. F., B. O. Parks, and L. T. Steyaert (ed.) Oxford University Press, New York.

Nyerges, T. (1993) Understanding the scope of GIS: its relationship to environmental modeling. in Environmental Modeling with GIS, Goodchild, M. F., B. O. Parks, and L. T. Steyaert (ed.) Oxford University Press, New York.

Peuquet, D.J. and Duan, N. (1995) An event-based spatialtemporal data model (ESTDM) for temporal analysis of geographical data. International Journal of Geographical Information Science, 9: 7-21.

Raper, J., and Livingstone, D. (1995) Development of a geomorphological spatial model using object-oriented design. International Journal of Geographical Information Systems, 9(4): 359-383.

Takeyama, M. and Couclelis, H. (1997) Map dynamics: integrating cellular automata and GIS through geo-algebra. International Journal of Geographical Information Science, 11: 73-91.

Tang, Y., T.M. Adams, and E.L. Usery (1996) A spatial data model design for feature-based geographical information systems. International Journal of Geographical Information Science, 10(5): 643-659.