Due to the complexity of issues in spatial analysis, it is actually unrealistic to ask for a GIS to include in its functionalities all aspects of spatial analysis. GISs, after all, are just data processing systems which, in my view, should just stay as a data processing system for spatial data storage, retrieval, and display. We should not ask a GIS to lead our analysis. On the contrary, GISs should be used as a support to facilitate spatial analysis and decision-making. We should separate the two things but provide an efficient and user-friendly environment for their integrative utilization.
To achieve such a goal, we need to build GISs as a truly open system with which we can customize for specific spatial analysis in an efficient and effective manner. Closed systems are doing a disservice to the accomplishment of such an objective and will perpetuate the current state of the use dictated by most of the commercial products, i.e. merely a device for data processing and display with limited spatial analytical capability. Though some spatial analysis modules have been incorporated into some GIS products, and macro languages have been provided for customization, they can hardly be considered as open systems in the strict sense.
The open-system design should pay particular attention to the entertainment of the following major movements in spatial analysis:
1. Spatial Dynamics. The analysis of dynamics
in spatial structures and processes have unique requirement of data structure,
I/O, and data-model integration.The
concept of temporal GIS has been
around for quite some time, and yet a truly spatio-temporal GIS is still
at large.
2. Evolutionary Computation. Spatial analysis
in recent years has experienced an upsurge in the use of two fast evolving
paradigms – Evolutionary and neural
computation – for complex systems
analysis. The design of GIS suitable for the requirement of models
such as neural networks, genetic algorithms, and
evolutionary programming is of
importance.
3. Artificial Intelligence. The availability
of intelligent SDSS is of practical value to researchers and practicing
professionals. The design of GISs for an effective
support of AI oriented investigations
is necessary.
4. Uncertainty. Though the issues of uncertainty
in GISs have been investigated over the years, a product that can truly
convey uncertainty in GIS operations and
spatial analysis is still non-existent.
This is a totally unacceptable state of the art provided by commercial
products.
As a concluding general observation, if one takes a careful examination of GIS related researches/publications, it is apparent that quite a large number of them can be accomplished without a GIS. They are using a modern means to achieve an old task with very little contribution to the advancement of spatial analysis. If this situation perpetuates, GIS will run out of gas in the near future and GIS research may boil down to nothing but the use of commercial software to doctor up our analysis, or at most to make data management and display more efficient. To make GIS research sustainable, and to further develop the discipline, we have to look for an answer in spatial analysis. After all, we want to solve spatial problems with GISs. Therefore, issues involving the design and integration of GISs with spatial problems such as those discussed above will give us a more promising future. Of c!ourse, doing all these in the internet should also be in the agenda.