Automatic detection of forest canopy gaps and estimation of leaf area index (LAI) using the digital fish-eye camera's images

Abstract

We propose an automatic and nearly versatile method for detecting forest canopy gaps using the digital fish-eye camera images. By analyzing data obtained by a field research during the entire growing-season in a cool-temperate deciduous broad-leaf forest in Takayama, Japan, we found the following criteria mostly applicable except for canopies with autumn-color : 1. The pixels with the digital number of the blue band less than 150 were at the vegetation (leaves and stems) . 2. The pixels with the digital number of the blue band more than 200 were at the canopy gaps. 3. The pixels with the digital number of the blue band between 150 and 200 were at the canopy gaps if their digital numbers of red were less than those of blue. Otherwise they were at the vegetation. Estimated canopy gap fraction by these criteria corresponded well with those estimated by the conventional and subjective approach. We compared gap fraction by these criteria with gap fraction by the other approaches and a transmittance by LAI2000. These comparisons indicated necessity of consideration about sensitivity and exposure of the digital cameras. The values of the leaf area index estimated from those gap distributions were much less than the estimation by the canopy light transmittance approach but the pattern of their seasonal changes (except for autumncolor) were close to each other.