Tropical dry ecosystems modelling and monitoring from space

Most dry ecosystems of the intertropical belt are subjected to anthropic pressures and climatic hazard. Many are strongly disturbed. In many cases, understanding their functioning, dynamics, evolution, and finally their general state of health, requires repetitive and overall surveys. In this context remote sensing can become a very powerful tool provided we can reliably relate ecosystem characteristics and radiometric information from space. Here, our main objective is the extraction of major ecosystem characteristics, in order to monitor tree mortality and beyond to manage reviewable resources. We developed a procedure to invert the DART (Discrete Anisotropic Radiative Transfer) model. This is well adapted to the environment, because of its capacity of simulating radiative transfer in heterogeneous 3-D scenes that comprise trees, shrubs and soil. It simulates spectral bi-directional reflectance images, for any viewing and illumination configuration. We tested the validity of this approach with a set of ten Landsat MSS (Multispectral Scanner System) images acquired during the 1972-1990 period, over a test site (84 per 43 km) located in east Burkina Faso (West Africa). Available ecosystem characteristics and rainfall data were used. This paper presents the development of the inversion methodology for retrieving ecosystem characteristics directly from multispectral and multidate remotely sensed data. Preliminary results clearly stress the potential of remote sensing systems for conducting sound ecological surveys and the interest of such a model inversion method as a management tool in sparse vegetation countries where firewood, vital fuel for human societies, becomes lacking. (C) 1998 Elsevier Science B.V. All rights reserved.