The Sensitivity of AusRivAS to Variations of Input Values, Low Natural Diversity, and Temporal Variation
Final Report for AusRivAS Toolbox Project TB1
Dr Leon Barmuta, Dr Louise Emmerson and Mr Petr Otahal, University of Tasmania
Department of the Environment and Heritage, 2003
ISBN 0 642 54967 2 ISSN 1447-1280
- The Sensitivity of AusRivAS to Variations of Input Values, Low Natural Diversity, and Temporal Variation(PDF - 2,228 KB)
- Confidence Limits (Zip - 1,771 KB)
- Low Diversity (XLS - 38 KB)
About the report
Rationale for the project
Toolbox Project TB1 aims to quantify the effect of "errors" in the data input into AusRivAS on the precision of outputs from this model. Although "errors" are inevitable in any data collection exercise not all of them are large enough to be considered serious errors. The rationale for this project is to determine whether data collection errors matter to the final AusRivAS output. In the context of this project, the term "error" refers not only to mistakes that are made when collecting input data, but also to the natural variability of the input values when they are measured, including variability that results from inter-annual variation in the status of a river's fauna. The goal of this project is to determine the effect of these "errors" on AusRivAS output, and to suggest methods for reporting their effects for end-users.
The current problem with AusRivAS outputs
At present AusRivAS produces a single point estimate of river health based on macroinvertebrate community structure for each site that is tested. A site is tested either at one point in time or using information pooled over two sampling times (each of which is in a prescribed time of year, for example, wet season and dry season). Thus no information about the precision of these estimates can be provided to end-users by AusRivAS in its present form. For example, if a site has a score of 0.80 on the O/E index output from AusRivAS, end-users cannot determine whether this represents a value of 0.80 ± 0.05 or 0.8 ± 0.25. Because the range of possible values for this O/E index is from 0 (representing a site in very poor condition) to about 1.2 (representing a site in very good or slightly enriched condition), the precision of the output index values has considerable impact on assessments of river health made using AusRivAS.
This project had four aims, and cross-references to the relevant chapters of this report are provided after each aim:
- Quantify the sensitivity of final AusRivAS models to variations and errors in the input data using existing sources of data appropriate to AusRivAS.
- Develop a procedure to compute a confidence interval (confidence limits) for AusRivAS outputs.
- Determine the sensitivity of AusRivAS to sites with low natural diversity.
- Determine the impact of temporal variability on the outputs of AusRivAS.