Variance in Bidirectional Reflectance over Discontinuous Plant Canopies

An exploratory study of the variance of the bidirec(Deschamps et al., 1994) instruments. Also, directional measurements have been made in a variety of laboratory tional reflectance over discontinuous plant canopies indicates that the patterns in variance can be related to the or field settings. However, soon there will be considerable directional imagery available, particularly from the properties of the plant canopies. The spatial variance of the bidirectional reflectance calculated from ASAS imMISR (Diner et al., 1989) and MODIS (Ardanuy et al., 1991) instruments. Thus, the opportunity for recovery of ages shows peak values at the hotspot and near nadir. This behavior can be explained by the geometric effect of surface parameters from directional measurements will increase dramatically. discontinuous tree crowns and the regularization effect. Validation of Jupp and Woodcock’s two-component geoWith the notion of recovery of surface parameters based on directional reflectance comes the idea that metric optical (GO) model (1992) shows that it captures the basic features of the spatial variance of the bidirecBRDFs are the first-order directional properties of a landscape. Often critical to inversion processes and their tional reflectance over discontinuous plant canopies. Their two-component GO model is modified to account for the success are the second-order directional properties. Local spatial variance and large, anisotropic change in respatial interactions of four scene components. Validation shows that the modified GO model improves predictions. corded radiance with changing Sun position and sensor view angle are characteristic of images of woodlands and This exploratory study will benefit future use of directional imagery to recover surface parameters by helping forest areas. In these ecosystems, the canopy is discontinuous, or “gappy,” and the local spatial and viewing characterize the distributional properties of directional imagery. Elsevier Science Inc., 1999 angle variation in images is created by a number of ecologically significant factors. These include interactions between the discrete nature of the canopy, the high natural spatial variation in canopy, the high natural spatial INTRODUCTION variation in canopy structure, and the visible shadowing There has been much work in the last two decades on effects as the Sun and view positions vary. bidirection reflectance distribution functions (BRDF), The BRDF of forests and woodlands is a statistical both in terms of measurement and modeling [see the refunction which operates at the scale of an average patch view by Strahler (1997)]. One of the primary benefits of of cover rather than at the scale of crowns. Studies of an improved understanding of the properties of BRDFs the forests and woodlands using aerial photography and should be the ability to improve recovery of surface high spatial resolution scanner data show very high variproperties from directional imagery. Until recently only ance at this detailed scale. In addition, the directional relatively small amounts of directional imagery have been effects of the Sun and observer positions interact signifiavailable, primarily from aircraft borne sensors such as cantly to create an angular anisotropic variation. At the ASAS (Irons et al., 1991) and the airborne POLDER fine scale, surface variations cause the local intensity variation, or image texture. This variation or texture depends on both illumination and viewing direction, which was * Department of Geography and Center for Remote Sensing, Boston University, Boston defined as BTF (bidirectional texture function) in Dana † CSIRO Earth Observation Centre, Canberra, ACT, Australia et al. (1996). Here this has been called the BRVF or biAddress correspondence to Wenge Ni, Raytheon STX Corp., 4400 directional reflectance variance function, corresponding Forbes Blvde., Lanham, MD 2076. E-mail: [email protected] Received 22 May 1998; revised 23 November 1998. to BRDF (bidirectional reflectance distribution func-