Selecting and testing an instrument for surveying stream shade

Abstract

We evaluated the suitability of several different instruments for surveying stream shade, selected one as most suitable for our purposes, and tested its accuracy. Five different operators used the instrument to estimate shade as angular canopy density (ACD), canopy density above 60°,, and canopy density above 80° in two plots—one in a mixed-age coniferous stand and one in a mixed-age deciduous stand. We compared operator estimates (ocular method) with measurements from fisheye photographs (computer-fisheye method). In a random coefficients regression model, the effect of “plot” on regression slopes and intercepts was not significant at α = 0.05. The regression line for ACD by the ocular method versus the computerfisheye method had a slope of 0.87 and an intercept of 0.02. The slope was significantly different from 1 at α = 0.05, indicating a tendency for human operators to underestimate ACD. Estimates of mean ACD on the two plots by individual operators were 2–11 percentage points lower, respectively, than mean ACD calculated from fisheye photos and the effect of operator was highly significant (ρ < 0.0001). Operators who received 45 minutes of training performed better than did an operator who received 15 minutes of training. Results suggest that operator variability is a large potential source of error in ocular estimates and that an investment of at least 1 hour of formal training may be worthwhile. The errors associated with any ocular-type canopy density measuring instrument should be documented before it is used to make statistical inferences.