The statistical basis of sampling strategies for rivers: an example using river habitat survey

1. This paper describes the statistical basis of River Habitat Survey. 2. The basic requirement of any survey based on a sample of all possible river sites is that the sample should provide unbiased estimates for the defined population of rivers and streams. Methods for ensuring a lack of bias and habitat characteristics include the definition of an appropriate sampling frame and the use of stratification to reduce the variability of the sampled population. 3. Alternative sampling strategies are reviewed and the reasons given for choosing a strategy based on random samples from a systematic stratification of 10×10 km Ordnance Survey National Grid squares. 4. The shape and size of the sample units are discussed in relation to the kinds of information required from the survey; the kinds of variables and attributes recorded in the survey are discussed in relation to the available methods of analysis and interpretation, and future development of the survey is discussed in relation to the likely changes in requirements for information about rivers and their habitats. © 1998 John Wiley & Sons, Ltd.