SATELLITE REMOTE-SENSING FOR FORESTRY PLANNING - A REVIEW

This review attempts, from a critical perspective, to put 25 years of satellite remote sensing research into the context of forestry plannin g. It is limited to temperate and boreal regions and to remote sensing in the visible and near infra-red spectra. It is structured as follow s: firstly, the forestry planning problem is defined and information r equirements are identified; secondly, the attempts to use satellite re mote sensing as an information source for forestry planning are review ed, divided into the topics inventory and monitoring; and, finally, th e methods for extracting forest information from satellite images are re-examined, and their inherent limitations and methodological weaknes ses are discussed. It is concluded that satellite images seldom contai n enough information to support the decision process in applied forest ry. Although regional level applications may be useful, few successful and reliable applications for local forest inventory, planning or dam age monitoring have evolved. Stratification for multi-stage sampling a nd monitoring of clear-cuts are areas in which satellite images have b een shown to be feasible. However, it is doubtful whether the costs in volved call be justified fur such uses alone. A further conclusion is that many studies have adopted simplistic views of the information nee ds in the forestry planning process, These studies do not relate the a nalysis to management decisions, but instead assume that plain and app roximate mapping of the forest has a great value. Present developments in the discipline involve complex modelling of reflectance, taking in to consideration internal shading, topography and other features, yet without dramatic improvements compared with earlier studies. Furthermo re, satellite remote sensing and digital image analysis are no longer technologically spectacular, facts which may affect future research in the discipline. It is, perhaps, time to draw the conclusion that curr ent satellite sensors are not in general suitable for forestry plannin g, since (a) they contain little relevant information, and (b) for for est management planning purposes there are often more efficient ways o f collecting the information required.