RP15 Assessment at Larger Scales (Watersheds, Landscape, Regional)
Thursday, 17 November 2005: 8:00 AM - 6:30 PM in Exhibit Hall

(KAP-1117-732469) A GIS-based Weight of Evidence Approach for Diagnosing Aquatic Ecosystem Impairment.

Kapo, K.¹, Burton, Jr., G. ¹, ¹ Institute for Environmental Quality, Wright State University, Dayton, OH, USA

ABSTRACT- A major goal in biological assessments of aquatic systems is determining whether impairment exists, and if so, to diagnose the dominant stressors. Current diagnostic approaches are hindered by a number of issues, such as the level of effort required to combine and interpret multiple lines-of-evidence, the nonstandard, qualitative and subjective nature of the analyses, and unknown levels of confidence (i.e., certainty) in their accuracy. Assessments should determine or quantify a diagnosis in a way that can be easily communicated and applied by those in charge of ecosystem management. GIS-based watershed-level assessment using Bayesian weights of evidence and weighted logistic regression provides a method to quantify and compare the effects of environmental stressors on lotic ecosystem health, and create predictive models of general or species-specific biological impairment across numerous spatial scales based on limited existing sample data. This technique is a data-driven, probability-based approach developed for and currently used in minerals exploration. Extrapolation of this methodology to a case-study assessment of the Great and Little Miami watersheds (OH) yielded baseline results consistent with previous assessments and additionally provided a quantitative determination of physical and chemical watershed stressor contributions, interactions, and probability of biological impairment at a spatial resolution of 0.5 km ² over the watershed study region. Channel and substrate habitat condition, and dissolved solids associated with soil run-off were determined to be major stressors within this study region, and chemicals associated with wastewater effluent and agricultural pesticides are important stressors at a land-use specific level. Stressor contributions, probability of impairment, and uncertainties of modeled predictions are directly quantifiable and displayed graphically. Integration of the model components into the ArcHydro data model allows spatio-temporal aspects such as high and low flow events to be incorporated, and various methods can assess the accuracy and versatility of this technique as an ecosystem management tool.

Key words: diagnostic, watershed, assessment, GIS