Environmental Data Integration Using Expert Systems Steven Carter 1* , Jessica Miller 1 , Peter Davies 1 , Frederick Bouckaert 2 , and Michael Wilson 2 . 1 University of Tasmania, Hobart, Tasmania, Australia. 2 Murray Darling Basin Authority, Canberra, ACT, Australia. * Corresponding author. Email: steven.carter@utas.edu.au ABSTRACT: The Murray Darling Basin Authority’s Sustainable Rivers Audit (SRA) program has developed a novel method for gathering and making sense of environmental and ecological data. Data gathered at a large number of sites according to agreed protocols are normalised to produce “metrics”, which compare observed values to estimated values under reference conditions appropriate to each variable. Metrics from different locations are spatially aggregated as required, grouped into ecological themes, and integrated by expert systems to produce condition indicators. The condition indicators for each theme are then integrated into an overall condition index, used to describe the state of river ecosystem health. This is used to communicate audit findings to regulatory authorities, and the public. This paper describes the SRA data processing method, focusing on the role of expert systems. Key points are how the development of expert systems is facilitated by the use of metrics; by the way in which environmental specialists define what each integration exercise must do; and by the choice of membership functions to simplify defuzzification using the centroid method. The paper concludes by outlining work to apply the SRA principles to develop an sustainability audit method for Lady Elliot Island, in the Southern Barrier Reef. KEYWORDS: expert systems, metrics, environmental indicators, sustainability audit. 1 Introduction The Murray Darling Basin Authority’s Sustainable Rivers Audit (SRA) program monitors environmental and ecological conditions across the Murray-Darling Basin river system, in a comprehensive and systematic manner. Observations are recorded at many locations throughout the 23 river valleys in the Basin, using standardised field sampling and data collection protocols. The SRA first reported in 2008, following data collection during 2004–07 [1]. The second report was issued in late 2012 following data collection during 2008–10, and further development of the data analysis method [2]. About 25% of the locations visited in the first audit period were also visited in the second audit, enabling direct detection of condition change between the audits; and the remainder of the locations in the second audit were new. Section 2 of this paper describes how SRA data are normalised to produce “metrics”, aggregated to the desired spatial scale. The metrics are grouped into ecological themes, and integrated by expert systems into indicators, which in turn are further integrated into an overall condition index for each theme. The use of metrics facilitates comparison of the river conditions across the Basin, and the detection of condition changes between audits. Indicators and indices provide an at-a-glance assessment of river conditions in the Basin. Section 3 explains how the metric integration exercises are defined by specialists, in a way that facilitates the development of expert systems to carry out the required work. Section 4 provides details of the expert systems. Section 5 outlines current work to apply the SRA principles to develop a sustainability audit method for Lady Elliot Island, in the Great Barrier Reef. The broad applicability of the SRA principles is underscored by the different spatial scales, ecosystems, and data-gathering methods for a river and a coral cay. 2 Metrics and Reporting Themes In the SRA program, a metric is the ratio of an observed value to its estimated or expected value under a reference condition. An example is the ratio of the number of fish species observed at a site, to the number expected at that site, if it was in its reference condition.