Automated procedures for estimating LAI of Australian woodland ecosystems using digital imagery, Matlab programming and LAI / MODIS LAI relationship

Leaf area index (LAI) is one of the most important variables required for modelling growth and water use of forests. The implementation an approach to estimate LAI from digital pictures (LAID) has recently been advanced in Australia using digital image capture and gap analysis, which employs a novel methodology (Macfarlane et al 2006). This technique uses upward-looking wide-angle digital photographs to capture canopy LAID and analyses these images using gap fraction analysis at a single zenith angle (0 – 57), using commercial image processing software. After implementing this technique in Australian evergreen Eucalyptus woodland, we have improved the picture analysis method from a time consuming manual technique to an automated procedure. Furthermore, in this paper, we compare MODIS LAI values with digital image LAI values for a range of woodlands in Australia. We used Matlab 7.4 (The Mathworks, Inc), to batch process numerous upwardlooking digital images (at least 50 per site) to estimate LAI (LAIM) from different woodland sites within New South Wales, Australia. The blue band (450 – 495 nm) of each image was extracted and explored to identify a threshold between foliage and sky. In the procedure, the selection of a suitable luminance value from the blue band histograms can be fully automated for numerous images or manually generated for each image. After assigning a suitable blue layer threshold, the image is transformed into a binary image for gap analysis. The gap analysis is performed by automatically dividing each binary image into nine sub-images. From each sub-image, the program counts the total of pixels corresponding to sky (S) and leaves (L). A big gap is considered when the ratio S/L ?