Estimating the leaf area index of North Central Wisconsin forests using the landsat thematic mapper

Abstract

Leaf area index (LAI) is an extremely important structural characteristic of vegetation because it is directly related to the exchange of energy, CO 2 and mass from plant canopies at a variety of scales. Research investigating the relationship between forest LAI and satellite data for hardwood and mixed conifer-hardwood forests is lacking, however. The objective of this study was to explore the utility of Landsat-5 Thematic Mapper (TM) data for accurately estimating the LAI of conifer, hardwood, and mixed conifer-hardwood forests in north central Wisconsin. Individual bands and vegetation indices (VIs) calculated from satellite measures of exoatmospheric reflectance were related to the litterfall-estimated LAI of 24 stands. The results showed that individual bands or VIs containing at least one infrared (IR) band (either near- or mid-infrared) or a strong IR component divided data into at least two groups, with each group requiring a different regression line. The primary division was between conifer-dominated and hardwood-doin inated stands. Of the individual bands and VI.s considered, seven were strongly correlated to the LAI of conifer stands ( r 2 =0.69–0.73). For the hardwoods, the best individual band or VI was Green/mid-IR#1 ( r 2 =0.35), although an additional individual band and two VIs did much better using re subset of lower LAI stands ( r 2=0.60–0.75). For individual bands and VIs not requiring a conifer-hardwood distinction, the sixth Tasseled Cap component was most closely related to LAI ( r 2 =0.60). Multiple-variable models (using LAI as the dependent variable) were found to offer substantial improvement over single-variable models, especially for hardwood stands. We recommend for further consideration a four-variable model for the conifers, and one four-variable and two eight-variable models for the hardwoods.