Abstract
High resolution images using a range of different technologies and sensed from air and space platforms are becoming increasingly available. Spatial statistics may be used to analyze the texture of such images to infer many types of information – including plant canopy structure. This paper summarizes a line of work which has sought to interpret the spatial variation of directional images of vegetation canopies and vegetation‐covered surfaces using a common framework of geo‐optical models. The early work on statistical analysis of hemispherical photographs by Nilson is described as well as Li and Strahler's variance model for crown size estimation. Then, the texture based work by Jupp and Woodcock is described. This involved analysis of aerial photos and airborne scanner data in open forests. Finally, some recently presented work using airborne ASAS images in dense conifer forest by the present authors is described. Previously unpublished work on the inversion of a simple spatial variance model, combined with considerations of BRDF, points to how interpreting the spatial variance of directional remote images through geo‐optical models can contribute the retrieval of canopy parameters such as tree sizes and canopy gaps through the integration off all of these stands. This paper aims to benefit future use of directional imagery to recover surface parameters by bringing these various pieces of work together. They collectively show how characterizing the distributional properties of directional imagery and interpreting them through geo‐optical models can provide canopy information and suggest how interpretation of various data sources ‐ including from canopy Lidars – can also benefit from the outcomes of model‐based spatial and directional variance research.
Notes
Presented at the International Forum on BRDF, American Geophysical Union Fall Conference, December, 1997, San Francisco, California.
Address for correspondence: Wenge Ni, Department of Geography, University of Maryland, 2181 LeFrak Hall, College Park, MD20742–8225.
