This first sample round should be considered as a large pilot study in which all aspects of the continuous global survey have been thoroughly tested and rendered operational. The survey as such has only just been started. The results produced so far meet statistical expectations and, more importantly, are rich in information content which, in describing land cover dynamics, helps one to understand crucial cause-effect mechanisms. These results demonstrate the theoretical correctness of the initial assumptions and have shown that they can withstand the test of implementation at pan-tropical level.
With the experience gained during this first round, both on theoretical and operational issues, the planning and implementation of subsequent survey rounds would appear to constitute a clear and wide path to follow, with very promising and achievable results. The legacy of this first round can be summarized as follows:
In addition to improving the reliability of results and consequently lowering the reporting level from regional and ecological to sub-regional/ecological, further survey rounds will permit the estimation of the changes in the rates of change, i.e., on increase or decrease of deforestation rates. This information will be of paramount interest to national and international policy makers since it would allow a quantitative assessment of the impact of land conservation and development policies, and to highlight the areas exposed to a higher risk of resource depletion.
The proposed continuation of the survey is described in FAO Project Document GCP/INT/…/G5221, “Programme for Continuous Assessment of Tropical Forest Resources using High Resolution Satellite Data”, which includes all specifications necessary for conducting three additional survey rounds. The proposal has been submitted to potential donors for funding.
The results presented here refer to and are based on only the Phase II of the FAO Forest Resources Assessment 1990 Project. Notwithstanding the fact that this was only the first round of a survey designed to be continued and to provide progressively more detailed and reliable results, further potential uses of the present results are numerous. Some of these potential and promising directions for further investigation can be outlined as follows:
The next most important step, essential to meet fully the purpose of this survey, is the integration of RS data with FORIS databases, as discussed above, to complete the two-phase assessment design. This integration will be of mutual benefit, resulting in improved deforestation models, national FORIS statistics and more efficient planning of subsequent survey rounds.
The consistent use of a standard and rather detailed land cover classification system makes the spatial output of the RS survey (maps of each SU's land cover state and change) a suitable reference for the calibration of pan-tropical classification efforts, based on coarse resolution satellite data.
Spatial modelling of forest dynamics. The digital integration of the sampling unit change maps, with their rich information on class transitions, and other GIS layers such as administrative boundaries (and associated socio-economic parameters), road networks, settlements, geo-morphology, etc., will facilitate the development of spatial change models.
