Reports and Papers





Fedra, K. (1995)
Decision support for natural resources management:
Models, GIS and expert systems.
AI Applications, 9/3 (1995) pp 3-19.




Data Bases and Models

The usefulness of organized data collections, and various forms of data base management software is quite generally recognized. And modelers and certainly model users are quite aware of the fact that input data preparation is often the main effort in applied modeling. So the integration of data bases and models, that allows users to automatically retrieve and load input data for complex environmental models is a natural step.

Model input data come in a number of different forms; they may be model control parameters such as a time step, global parameters such as the decay rate of a chemical, or dynamic data such as time series of boundary conditions, and also spatially distributed data sets, resulting in two and three dimensional matrices of initial and time series of matrices of boundary conditions.

These data sets usually differ considerably in their frequency of change: while certain scenario assumptions, say, the pumping rates in a groundwater model, will most likely be changed frequently if not for every model run, other data sets are much more static, for example, aquifer characteristics such as porosity, or geometrical data such as the depth of the aquifer. Many of these data are required by models in more or less cryptic and difficult to directly understand (let alone provide by the non-expert user) forms. Thus, an additional task for a data base management system coupled with a simulation model is in the translation of a user-friendly representation of data sets into whatever formats the code requires. Expert systems as yet another of the methods to be integrated provide some of the necessary tools.

In order to configure efficiently a new data set for a model run users have to specify their problem in terminology that fits their purposes. For example, in a regulatory framework, it is more likely that the name of a hazardous chemical is known rather than its physico-chemical parameters. These, however, can be conveniently retrieved (in the units the model expects) from a chemical data base once the chemical has been identified.

Other elements such as the basic geometry of the problem (a groundwater contamination case, for example) are automatically loaded from the appropriate data base from the choice of location, at the level of the GIS.

In addition to this internal coupling of data and models, the integration of outside sources of data in any operational system is of great importance. For any real world problem, data will come from numerous sources, in different formats and with different quality. Their integration into one unifying information system requires a number of tools to extract and filter, reformat and convert, inter- and extrapolate, adapt and often interpret the original data.


© Copyright 1995-2016 by:   ESS   Environmental Software and Services GmbH AUSTRIA | print page