Modeling forest canopy closure using vegetation index

Abstract

The use of remote sensing to estimate forest structure has been largely tested but few researches have related such information about canopy closure in tropical forests in steep slope. This work aims to analyze the potential employ of vegetation index and sunlight radiation to generate a predictive model of canopy closure in the Ribeira Valley, south of the São Paulo State – Brazil. The canopy closure data were obtained from Spherical Densiometer in 52 sample points. The sun radiation (AIF) was obtained using TOPODATA image and literature equations. Thus, canopy closure and AIF information were related to NDVI, EVI and LAI, obtained from LANDSAT-TM and ALOS-AVNIR images. The results showed: a) field canopy closure facing to the North, East and West presented a tendency to have higher canopy closure then points facing to the south; b) the field canopy closure ranged from 0.58 to 0.97 c) and the annual illumination factor (AIF) ranging from 0.28 to 0.66 ; d) all three indexes showed lower determination coefficients when compared with image bands alone; e) two spatial resolution images were tested using 30 m (TM) and 10 m (ALOS); the lower pixel size did not result in better canopy closure estimation; f) the use of topographic correction on the TM images did not resulted in better model explanation of canopy closure, comparing it with models that use AIF; g) the blue ALOS band and TM7 Landsat band models explained, about 27% and 30% of the variation in observed canopy closure, respectively.