A COMPARISON OF OPTICAL AND DIRECT-METHODS FOR ESTIMATING FOLIAGE SURFACE-AREA INDEX IN FORESTS

Measurement of foliage surface area index (foliage SAV) is prominent in studies of terrestrial ecosystems because it is an important determinant of water, carbon, and energy exchange at the stand, landscape, and global scales, yet is very time consuming and labor intensive to measure directly. The objectives of this study were to 1) compare the resolution and accuracy of the Ceptometer, LAI-2000, and DEMON which measure a vegetation area index (VAI; includes branches, stems, cones, etc. in addition to leaves) optically using light interception, 2) examine the utility of logarithmic versus linear averaging and a correction factor proposed by Gower and Norman (1991), and 3) determine if there is a 'universal' regression relating optical and direct estimates of SAI across the range of foliage areas for most North American forests (and if one exists, how it relates to individual site regressions). For our analysis, data collected in open canopy natural ponderosa pine (Pinus ponderosa Dougl. ex Laws) forests and closed canopy red pine (Pinus resinosa Ait.) plantations were combined with data from previous studies which used one or more of the instruments. We found all instruments generally underestimated SAI when compared with direct estimates. Logarithmic averages of light transmittance reduced this problem, especially for conifer forests with foliage clumped at the shoot and canopy levels. The DEMON, using logarithmically averaged data, provided the most accurate optical estimates of SAI. We did not find the clumping correction factor suggested by Gower and Norman (1991) to be useful and suggest an alternative correction technique that is based on hemisurface area index (HSAI) and includes a shoot shape factor along with a clumping factor. Across a broad range in foliage SAI, all instruments provided optical estimates of SAI that were strongly correlated to direct estimates but the optical estimates were biased. The LAI-2000 (R(2) = 0.93) and the DEMON logarithmic (R(2) = 0.93) had the best fits. Because of the small size of the data sets evaluated, we strongly recommend the collection of larger data sets across a wider range of foliage SAIs to better test the strength of 'universal' regressions relating optical estimates of SAI to direct estimates.