Methodology comparison for slope correction in canopy leaf area index estimation using hemispherical photography

The slope effect and correction methods for estimation of canopy gap fraction, leaf area index (LAI), mean leaf angle and clumping index using hemispherical photography, were investigated. The evaluation was carried out in tropical cloud forest and plantations in South-East Kenya in order to consider a range of canopy architecture and slopes up to 65%. The aim was to compare two acquisition techniques and various correction procedures. All estimates assume uniform slope, canopy parallel to ground and homogeneous canopy structure at the photo site level.(1)Photographs oriented to local zenith (levelled acquisition). Calculation and removal of sky parts of the hemisphere obstructed by topography. Azimuthal inversion of gap fraction without prior averaging, deriving local LAI estimates (quasi-random model). (i) Fixed path lengths over azimuths. Zenith reference axis. LAI referred to horizontal and corrected for topographic shading. (ii) Variable path lengths over azimuths. Normal to slope reference axis. LAI adjusted to horizontal by dividing by the slope cosine.(2)Photographs oriented parallel to slope (tilted acquisition). Fixed path lengths over azimuths. Normal to slope reference axis. LAI adjusted to horizontal by dividing by the slope cosine. Azimuthal inversion of gap fraction without prior averaging, deriving local LAI estimates (quasi-random model).Gap fractions present a stronger upslope/downslope asymmetry if retrieved from levelled acquisition. As a result, gap dispersion index and clumping index proved to be significantly higher for levelled acquisition (P<0.001). LAI estimates adjusted to horizontal are not significantly different, whether retrieved from levelled or tilted acquisitions, up to 30% slopes. From levelled acquisition, fixed and variable path length do not yield significantly different LAI estimates along the whole slope gradient. From tilted acquisition, LAI values were systematically higher than from levelled acquisitions, the stronger the slope, the higher the difference. Mean leaf angles do not differ significantly (P>0.05) for fixed vs. variable path lengths along the slope gradient up to 30%. For more severe slopes, variable path lengths yield lower mean leaf angle values. The interpretation of results from tilted acquisition remains uncertain. As a preliminary study, no preference is suggested for the levelled or tilted acquisition technique. Further investigation is needed and indirect optical derived estimates should be checked against direct reference measures, which are almost entirely lacking for mountainous areas.