Presenter/Author Information

Wilfried Winiwarter
Gerald Schimak

Keywords

emission, air pollutant, expert system, geographic information system, modeling limitation

Start Date

1-7-2002 12:00 AM

Description

Advances in environmental sciences increase the difficulty in creating products that can be readily used by non-experts. In this paper we demonstrate how a carefully designed environmental system is able to provide solutions. While complex transformations are kept in the system background, predefined expert choices give a user great flexibility in selecting what is required for the specific applications. In the given case study, emissions of atmospheric pollutants in the province of Upper Austria can not only be calculated for a multitude of options, but a scenario tool also allows for deep modifications of the models, if required. The example demonstrates how environmental experts and computer science experts have to work hand-in-hand to translate model treatment of environmental processes into a user-friendly environment. The overall system is also characterized in its theoretical framework. Seen as a deterministic model, an increasing amount of data input and better understanding of the underlying processes will allow a more realistic simulation of the reality. Without a full validation, however, just increasing the model size will not improve performance. Instead, an expert system will be built, which is overdetermined with respect to input quality, but contains the best available expert judgement on the processes described. Such an expert system should not be expected to possess any predictive qualities outside of the range of the test data. Consequently data reduction has to take place both in terms of model complexity and in terms of the required input. This kind of data reduction requires implicit expert knowledge, model reductions are the essence of any explicit process description.

Share

COinS
 
Jul 1st, 12:00 AM

The Emission Inventory System for Upper Austria: The Role of Environmental Information Systems in Translating Complex Information to a User s Requirements

Advances in environmental sciences increase the difficulty in creating products that can be readily used by non-experts. In this paper we demonstrate how a carefully designed environmental system is able to provide solutions. While complex transformations are kept in the system background, predefined expert choices give a user great flexibility in selecting what is required for the specific applications. In the given case study, emissions of atmospheric pollutants in the province of Upper Austria can not only be calculated for a multitude of options, but a scenario tool also allows for deep modifications of the models, if required. The example demonstrates how environmental experts and computer science experts have to work hand-in-hand to translate model treatment of environmental processes into a user-friendly environment. The overall system is also characterized in its theoretical framework. Seen as a deterministic model, an increasing amount of data input and better understanding of the underlying processes will allow a more realistic simulation of the reality. Without a full validation, however, just increasing the model size will not improve performance. Instead, an expert system will be built, which is overdetermined with respect to input quality, but contains the best available expert judgement on the processes described. Such an expert system should not be expected to possess any predictive qualities outside of the range of the test data. Consequently data reduction has to take place both in terms of model complexity and in terms of the required input. This kind of data reduction requires implicit expert knowledge, model reductions are the essence of any explicit process description.