The declining statet of the tropical forests and the adverse effects of its depletion on the evironment have stimulated a number of projects to assess and monitor their state, extent and distribution. It is now widely recognised that remote sensing technology is indispensable for achievement of this goal. Complete and repetitive coverage of the forests with high resolution satellite data is not feasible due to high volume of data, their cost and time constraints. Therefore, a multistage approach has been adopted to achieve reasonable accuracy at affordable cost. Such an approach is presently being used for the implementation of FAO's Forest Resources Assessment 1990 (FRA) Project where the present state of the forests and the changes they have undergone since 1980 are assessed for about 120 samples (the size of one Landsat Scene) distributed over the tropical belt according to statistical procedures. For FRA, regional vegetations maps, also partly based on remote sensing data, have been used for stratification and the selection of primary sampling units (Landsat Scenes). However, low resolution remote sensing data such as NOAA AVHRR have recently successfully been tested for reconnaissance forest mapping at 1:1 000 000 scale and in future could complement or even replace vegetation maps for stratification. The multitemporal coverage required to assess the extent and the speed of the deforestation is often a problem due to cloud-cover prevailing in tropical forest areas. The recent launch of ERS-1 with its cloud penetration capacity offers a solution for regular observation of the sample areas. The level of vegetation classification that can be achieved with ERS-1 data, however, has still to be assessed. The analysis of remote sensing data can be enhanced in terms of accuracy and classification level that can be achieved if digital image processing techniques are being used. These include geographic and radiometric correction of the data, application-specific enhancement of the imagery to enable maximum differentiation of the required forest types for subsequent visual interpretation or - alternatively - the computer-aided classification of the remote sensing data. Results of image classification are presented in the form of thematic maps as well as statistical tables and could also be directly used as inputs to geographic interfaces to GIS and in some cases, remote sensing and GIS software packages are integrated and installed on the same hardware configuration. If appropriate printing/plotting periphery is available, GIS can also speed up and facilitate to a great extent the production of thematic maps. Presently, in most of the tropical countries, the equipment and skills required for regular monitoring of their forest resources are not available. Strengthening of indigenous remote sensing and GIS capacities in developing countries is ...
