A knowledge-based approach to regional acidification modelling

Area of Science:

• Environmental Science
• Geochemistry
• Hydrology

Background:

• Acidic deposition poses a significant threat to aquatic ecosystems.
• Accurate assessment of lake alkalinity is vital for understanding ecosystem health.
• Existing models for predicting water chemistry can be complex and data-intensive.

Purpose of the Study:

• To develop and validate a knowledge-based expert system for predicting lake alkalinity.
• To assess the impact of acidic deposition on aquatic environments.
• To integrate diverse scientific data and expert knowledge into a predictive model.

Main Methods:

• Assembled data from over 2000 stations on atmospheric transport, soil, lake, and wetland processes.
• Utilized expert knowledge and rules to select appropriate acidification models for watershed data.
• Grouped data by tertiary watersheds based on water chemistry.

Main Results:

• The expert system achieved median relative errors of 11.3-17.9% in computed alkalinity.
• Regression analysis showed strong agreement between computed and observed alkalinity (coefficients 0.82-0.99).
• Model performance was validated using paleolimnological and independent datasets.

Conclusions:

• The knowledge-based expert system is a viable approach for assessing acidic deposition impacts.
• Expert rules were acceptable to specialists and produced reasonable agreement with observations.
• The system demonstrates sensitivity and robustness in predicting lake alkalinity.