Forest Leaf Area Index Estimated from Tonal and Spatial Indicators Based on IKONOS_2 Imagery

Abstract

The linkage of four spectral vegetation indices (VI) and 12 texture measures (TEX) derived from an IKONOS-2 image with leaf area index (LAI) measured in the urban area of Nanjing city, China was analyzed. Models for retrieving LAI based on VI or/and TEX were established for planted low broad-leaf forest (PLB), planted mature forest (PMF), natural broad-leaf forest (NBF), and all measured plots (ALLv), respectively. The VIs were calculated from multi-spectral bands at four radiometric correction levels, and the textures were computed from the panchromatic band of the image. The results showed that LAIs are positively related to vegetation indices (r = 0.419 – 0.737), and high correlation coefficients were found between LAI and textures. For regularly planted forests, LAI was strongly correlated with both VI and TEX, and the ‘best’ models could be established only using texture measures. The determination coefficients (R 2 ) were 0.787 and 0.626 for PLB and PMF, respectively. In contrast, VI is a better predictor of LAI than texture measures for NBF. When VIs and TEXs work together, the R 2 of the ‘best’ model was 0.764 for ALLv plots. This study indicated that these two measurements should be effectively integrated for reliably retrieving forest LAI from highresolution remote sensing data.