`Assessing the biological quality of fresh waters: RIVPACS and other techniques', edited by John F. Wright, David W. Sutcliffe and Mike T. Furse.
`Assessing the biological quality of fresh waters: RIVPACS and other techniques', edited by John F. Wright, David W. Sutcliffe and Mike T. Furse. Published by the Freshwater Biological Association, Ambleside, June 2000. ISBN 0 900386 62 2. 400 pages. Price £40 softback, £60 hardback (including p. & p.).

 

RIVPACS - Sources of Error in Prediction

Errors in estimates of the expected fauna are potentially due to:

  • Having an inadequate set of reference sites; too few or poor coverage of some stream types. 
  • Although all of good or high quality, the reference sites are not all of the same quality. In fact they will vary considerable in quality, however biological quality is defined. 
  • The macroinvertebrate data for the references sites will be subject to the usual effects of sample variation. 

 
  • Not involving all relevant environmental variables (e.g. macrophyte or habitat diversity). The best environmental variables for predicting fauna may include some, such as chemistry or current flow that are a cause of the stress or pollution which the RIVPACS model is trying to assess. It is important to try to exclude such variables. Essentially a predictive model is needed which is “fit for purpose”.  
  • Not making optimum predictive use of variables. 
  • Errors in measuring the environmental variables for new test sites from which predictions of their expected fauna are made.
  • These are all potential sources of error in any model, not just RIVPACS type models.

.