ABSTRACT: Solar exposure profoundly affects stream processes and species composition. Despite this, prominent stream monitoring protocols focus on canopy closure (obstruction of the sky as a whole) rather than on measures of solar exposure or shading. We identify a candidate set of solar exposure metrics that can be derived from hemispheric images. These metrics enable a more mechanistic evaluation of solar exposure than can be achieved with canopy closure metrics. Data collected from 31 stream reaches in eastern Oregon enable us to quantify and compare metrics of solar exposure from hemispheric images and a metric of canopy closure with a concave densiometer. Repeatability of hemispheric metrics is generally as good as or better than the densiometer closure metric, and variation in the analysis of hemispheric images attributable to differences between analysts is negligibly small. Metrics from the hemispheric images and the densiometer are typically strongly correlated, at the scale of an individual observation and for 150 m stream reaches, but not always in a linear fashion. We quantify the character of the uncertainty in the relationship between the densiometer and the hemispheric metrics. Hemispheric imagery produces repeatable metrics representing an important ecological attribute; thus those researching the effects of solar exposure on stream ecosystems should consider the use of hemispheric imagery. (KEY TERMS: solar exposure; instrumentation; meteorology/climatology; densiometer; hemispheric imagery; stream assessment.)

Solar input drives stream heating, primary production, periphyton species composition, fish life history strategies, and numerous other stream parameters (Gregory, 1980; Cummins et al., 1984; Beschta et al., 1987; Feminella et al., 1989; Tait et al., 1994; Shaw and Bible, 1996; Rutherford et al., 1999; Grether et al., 2001). Despite its significance, prominent stream monitoring protocols (e.g., Fitzpatrick et al., 1998; Peck et al., 2000) focus on canopy closure (the proportion of the sky which is obscured when viewed from a point) (see Jennings et al., 1999) rather than on more direct measures of solar exposure (the amount of solar energy received per unit area per unit time) or shading (the proportion of solar energy that is blocked by vegetation and topography). One study (Platts and Nelson, 1989) illustrated the distinction in showing that measures of shade were better predictors of salmonid biomass than measures of canopy closure in streams in the intermountain West. This result suggests that the widespread focus on canopy closure may err in quantifying solar exposure and its effect on stream characteristics. Thus, methods to quantify solar exposure should be defined and evaluated for use in stream monitoring and assessment programs.

Davies-Colley and Payne (1998) compared nine tools for measuring stream shade. They concluded that "using fisheye photography potentially yields maximum information, but requires much offsite image processing. This method may become more popular when digital cameras are fitted with fisheye optics" (Davies-Colley and Payne, 1998:258). Since their article was prepared, not only have digital cameras become fitted with fisheye optics, but a new generation of software has become available that greatly simplifies the analysis of hemispheric images.

Good reviews of the use of hemispheric imagery for plant and forest ecology are available (Chazdon and Field, 1987; Rich, 1990; Roxburgh and Kelly, 1995). Several authors discuss the merits of direct and indirect measurement methods for measuring solar exposure (e.g., Rich, 1990; Davies-Colley and Payne, 1998; Jennings et al., 1999). Direct measures of solar radiation are valuable, but they represent only the period of time when the measurements are taken; for stream assessments they would need to be compared against comparable and simultaneous measures at a nearby unobstructed location. In contrast, indirect estimates developed from single hemispheric images are valid for as long a period as the image is a valid representation of the quantity and location of features that obscure the sun. In general, the correlation between hemispheric metrics (an indirect method) and direct measures of light is good, particularly under more open canopies and when atmospheric conditions are properly evaluated (Whitmore et al., 1993; Easter and Spies, 1994; Roxburgh and Kelly, 1995; Comeau et al., 1998; Jennings et al., 1999; Machado and Reich, 1999; Ferment et al., 2001).

Given hemispheric imagery's proven track record and its potential to provide information on stream solar exposure, our goal is to evaluate its feasibility and characteristics as a tool for stream monitoring and assessment. These steps are key elements to the second and third of four steps suggested for ecological indicator evaluation (Jackson et al., 2000; Fisher et al., 2001).

We examine four issues in our analyses. First, we identify a set of candidate indicators of solar radiation that may be valuable for stream assessments. Second, we compare canopy closure estimates using a densiometer (an inexpensive and widely used device) (Fitzpatrick et al., 1998; Peck et al., 2000), which has been useful in explaining the status of instream resources (e.g., Herlihy et al., 1998; Hill et al., 1998; Bryce et al., 1999; Pan et al., 1999) and measures of candidate indicators derived from hemispheric images. Third, we characterize sources of error in the analysis of hemispheric imagery. We examine analyst error because numerous previous researchers (Rich, 1990; Whitmore et al., 1993; Jennings et al., 1999; Robison and McCarthy, 1999; Englund et al., 2000; Engelbrecht and Herz, 2001; Hale and Edwards, 2002) note this potential source of error. We also examine sampling error, because its magnitude is the key to making quantitative design decisions about the use of any monitoring tool. Fourth, we examine the relationship between both analyst and sampling error on the one hand and canopy closure on the other hand.