Auditing the precision and accuracy of environmental impact predictions in Australia

Our understanding of natural ecosystems can be measured by our ability to predict their responses to external disturbances. Predictions made during environmental impact assessment (EIA) for major development projects are hypotheses about such responses, which can be tested with data collected in environmental monitoring programmes. The systematic comparison of predicted and actual impacts has been termed environmental impact audit. Ecosystem disturbances associated with major resource developments, though of lesser magnitude than those associated with natural cataclysms, are generally of far greater magnitude than those which can be applied experimentally. Environmental audit can hence provide critical tests of theory in a number of natural sciences. It is also needed to improve the scientific content of EIA. Audits of 4 and 29 EISs respectively have been carried out previously in the UK and USA, but this is the first national scale audit for any country. It is also the first attempt to select, from the many vague statements in EISs, only those predictions that are scientifically testable, and to determine and analyse their quantitative accuracies. Its principal results are as follows. The average accuracy of quantified, critical, testable predictions in environmental impact statements in Australia to date is 44%±5% s.e. Predictions where actual impacts proved more than expected were on average significantly (p<0.05) less accurate (33%±9%) than those where they proved as or less severe (53%±6%).