Lost in space,the sequel: spatial sampling issues with 1‐D PHABSIM

The Physical Habitat Simulation System (PHABSIM) is a popular method for evaluating the habitat value of a stream in terms of a statistic, weighted usable area (WUA). Usually, PHABSIM is used with one‐dimensional (1‐D) hydraulic models, and curves of WUA over discharge are calculated at transects and combined to produce a composite curve. Curves of WUA over discharge are often presented as evidence in proceedings to determine instream flow requirements, so the reliability of these curves is important. Representing a reach of stream with transects introduces ordinary statistical questions: is the sample of transects unbiased, and is it large enough to produce usefully precise estimates? Unfortunately, these questions are seldom considered in PHABSIM studies, even though most PHABSIM studies characterize streams with fewer than 15 transects. Moreover, transect locations usually are selected deliberately, so estimates of WUA will usually be biased. This paper extends an earlier analysis of the uncertainty in composite WUA curves with a much larger set of transect curves, with more analysis of uncertainty in the shape of the WUA curves, and with a different method for simulating errors in estimates of WUA at the transects. The results show that even with larger than usual numbers of transect curves, the precision of composite WUA curves is likely to be poor, especially if there are errors in the transect WUA curves. I also offer suggestions regarding sampling, and for estimating the number of transects that might be needed to achieve a given level of precision in WUA estimates. Copyright © 2009 John Wiley & Sons, Ltd.