Monitoring world’s resources has rapidly become a permanent necessity. Information on the dynamics affecting bio-physical resources is demanded at decreasing intervals and increasing thematic depth.
In a time frame beyond the 2000 appointment, there are a number of ideas and suggestions on how to improve and extend the scope of the FRA remote sensing survey. Survey that must be seen and designed as a permanent source of information and not constrained to specific events, such as the onset of the new millennium. Several indications and recommendations for improvement came from the last Kotka meeting, others are a direct consequence of the knowledge base created by the FORIS component and the first FRA survey, others more came with technological advances and with increasing definition of thematic requirements. These improved features embody a survey concept currently under development called World Forest Survey (WFS).
These improved survey features can be classified into two main groups: (a) technical/methodological improvements, and (b) extended scope of the survey as summarized below:
2.1 Technical/methodological improvements
Make use of current knowledge as auxiliary variables for the stratification and more efficient sample allocation. In recent years the tropical regions have been covered by a number of large-scale studies, carried out by FAO and others, that considerably improved our knowledge on forest cover and on the broad spatial distribution of deforestation risk. Since deforestation, or forest change, is certainly the key parameter to be measured, a stratification on deforestation risk, as currently applied by TREES Project, is possible and highly recommended, notwithstanding an adequate coverage of the entire population. This will likely improve the precision of forest change estimates, which normally present a comparatively high sampling error.
Preliminary studies on sampling intensities based on the FRA 1990 results indicate that, along with more efficient stratification criteria, a slightly higher intensity would also be required. The number of sampling units over tropical regions has been thus estimated at some 150, as compared to the 117 currently studied.
2.1.2 Improved digital processing and cartographic output
In the analysis of future sample sites the interdependent interpretation will remain the core element of the monitoring methodology, in order to ensure the temporal consistency, but the use of digital data will be predominant. Unlike the analogue approach followed so far, a digital approach will enable the production of time series spatially and radiometrically co-registered, which will facilitate the interpretation. A digital approach will produce digital cartographic output without additional GIS work. Procedures have been developed for the interpretation on screen of multi-date digital images that allow for the fast viewing through the time series (CORINE Update)4, which is essential to the interdependent approach. Other, more sophisticated processing systems allow, through the analysis of radiometric correlation between multi-date data sets, the automated detection of potential land cover changes (change vector analysis5).
In all cases, however, is up to the human interpreter to decide if variations within the time series are true land cover changes or simple changes in appearance due to seasonal variations, atmospheric disturbance, etc.
Another important consideration vis-à-vis technological features is that the essential expertise required for the interpretation will remain on the object of the study, i.e. land cover, vegetation, land use, and not on the tool used to estimate it.
2.2.1 From pan-tropical to global
In order to produce a complete and coherent picture of the status and changes of world’s forests the survey should be extended outside the tropics to cover the entire world. This extension of coverage to include temperate and boreal regions was recommended by Kotka III6 experts on the basis of the results obtained from the FRA 1990 pan-tropical survey. In fact, the information on forest change from these regions is often contradictory, as in the case of Russia, or so heterogeneous to prevent global syntheses with adequate thematic depth.
The expansion of the survey to the entire world presents some new methodological issues ranging from the land cover classification in temperate and boreal regions, which cannot be equal to that applied for the tropics, to the sampling design and stratification criteria. These issues have been addressed in a number of pilot studies and technical discussions among experts from the US Forest Service, the Joint research Center and FAO. The conclusions from these consultations are summarized in a proposal presented in 1998 to the North America Forest Commission7.
A total number of some 300 sampling units, half over tropical and half over temperate and boreal regions, is preliminarily considered adequate for a world-wide survey.
2.2.2 Field sampling
WFS should address increasingly complex information requirements on the dynamics interesting world’s forests. This challenging task would require, in addition to land cover change measurements, information on forestry parameters as well as on social, economic and environmental aspects.
Although the remote sensing component will remain the backbone of the survey, the collection of these important parameters will impose a terrain approach inclusive of field sampling for forestry and environmental data, socio-economic investigations, etc. The survey would be designed in a multi-phase structure where global maps such as forest cover and deforestation risk (the latter limited to tropical regions) would represent the first phase, the high resolution sample would represent the second phase and the field sample would represent the third, and subsequent, phases.
The field sampling will be the most demanding part of the survey, presenting complex logistic, administrative and statistical problems and requiring extensive financial resources.
In fact, the terrain phase will require a considerable R&D component where series of pilot studies will be carried out to test alternative statistical and operational approaches in varying ecological and socio-economic conditions.
2.2.3 Link to national inventories and to international initiatives
It is evident that such an ambitious programme can only succeed if it benefits from the partnership and terrain knowledge of national institutions. The direct link with national inventory and planning institutions will be of mutual benefit. National institutions will benefit from methodology development and internationally accepted standards, while the WFS will benefit from the location-specific data, terrain knowledge and logistic support particularly during the field phase.
In line with FAO underlying objectives in the forestry sector, which focus on supporting member countries in the sustainable management of forests, the survey must be designed to produce technical solutions that are applicable at national level and that are suitable to address national planning issues.
As per other international and regional initiatives, all possible efforts will be made to harmonize and integrate with on-going large-scale studies to the extent that this will be beneficial to meet survey’s objectives and in member countries’ interest. Such integration will be achieved through the harmonization of technical solutions aimed at valorizing the work already done and avoiding redundancy and repetition.
FAO intends to support member countries to improve the management of forests. Remote sensing is one technique that is essential for achieving this. However, it must be acknowledged that remote sensing alone cannot provide all the information needed. Remote sensing is irreplaceable for (a) visualizing geographic patterns, i.e. mapping, and measuring changes over time, and (b) to facilitate efficient and controlled field sampling. At FRA, the need for extensive field sampling is currently emphasized, to complement and enrich the remote sensing surveys of the world’s forests.
4 “Technical and Methodological Guide for Updating Corine Land Cover DataBase”, V. Perdigão, A. Annoni, SAI, JRC.
5“Landsat Pathfinder Data Sets for Landscape Change Analysis”, J. L. Dwyer, K. L. Sayler, G. J. Zylstra, EROS Data Center, 1996.
6“Expert Consultation on Global Forest Resources Assessment 2000” (Kotka III) held in Kotka, Finland in 1996
7“Global Forest Resources Assessment 2000 for North America”: A Proposal to the North America Forest Commission for the Remote Sensing Component”, Raymond L. Czaplewski, USDA Forest Service, Rudi Drigo,FAO, June 16 1998.
