[Comparison of optical methods for estimating the opening of the canopy and leaf area in leafy forests]

Based on inversion of gap fraction data (Poisson model of foliage distribution), three optical methods using the Demon, the Plant Canopy Analyzer LAI-2000 (PCA) and hemispherical photographs, have been compared to estimate canopy openness (CO) and leaf area index (LAI) in a mature, neutrophil, oak-beech-hornbeam forest on mull in eastern France. Mean CO over the whole hemisphere was similar for PCA (7.9%) and hemispherical photographs (8.0%). The needle method, a vertical point quadrat method, applied to the litter after leaf fall has served as a reference to LAI (4.7). The Demon provided the estimate (4.9) closest to the reference value. The PCA and hemispherical photographs underestimate mean LAI by 30% (3.3) and 19% (3.8), respectively, if used without correction. Based on fish-eye sensors, LAI estimates can be improved if 3 annuli (4.2) or 2 annuli (4.5) are used in place of 5 with the PCA, or by means of logarithmic averaging of gap fractions over azimuth at an appropriate angular resolution (180 degrees: 4.6, or 120 degrees: 5.2) with hemispherical photographs. Not taking into account azimuthal variation in gap fraction distribution generates a more important error than the error induced by light scattering near horizon. In order to improve LAI estimates, an original iterative procedure is presented, which allows the simultaneous calculation of LAI over a broad range of angular azimuthal resolutions.