As mentioned previously, WISDOM is designed to optimize the use of available resources and information. In this section, the main practical aspects and issues involved in the undertaking of a WISDOM analysis are outlined and cartographic and socioeconomic databases that can be used for this purpose are indicated.
To conduct a WISDOM case study, the analyst needs to clearly define the objectives of the exercise and the minimum administrative unit of analysis, as these will determine the level of detail and intensity of the data collection effort (RWEDP, 2000).
To decide what kind of data is required, the type of wood supply and the users and sectors to be covered need to be specified. For example, completely different trade circuits exist in rural areas from those found in urban areas. Examining household and industrial woodfuel uses involves integrating information coming from different sources and which contemplates different variables. Woodfuel extraction methods, as well as their productivity and environmental impacts differ greatly between agricultural and forest areas, and between arid and humid regions.
The large social and environmental heterogeneity of woodfuel consumption patterns and the difficulty in conducting consistent and regularly updated field surveys over whole countries or large areas frequently makes wood energy use and supply data scattered and fragmented. For these purposes, completing the WISDOM demand module requires integrating data from local surveys and studies, national or regional censuses, and even international databases (see Appendix 1).
As WISDOM uses statistical information disaggregated by sub-national units, it is necessary to have complete data sets associated to these units in order to “spatialize” the information over the maps. Gaps in the data may be filled in three ways: 1) by the use of proxy variables to “spatialize” discontinuous values (e.g. using rural population as a proxy for woodfuel users); 2) by extrapolating information available at the project level to the entire study region (i.e. to extrapolate fuelwood consumption per capita). This procedure may be valid in cases where data does not need to be highly accurate or where woodfuel consumption estimates cover at least representative regions or situations within the study area; and 3) by filling specific or critical data gaps with new data coming from field surveys. This option could be very expensive, so the surveys need to be carefully designed, trying to minimize the cost and effort needed to complete them for a given level of accuracy. Different woodfuel survey methodologies can be used for these purposes; see for example RWEDP (2000) and Arias and Riegelhaupt (2002).
Global or continental socioeconomic datasets are becoming increasingly available, a fact that can help solve at least part of the potential data gaps at country or sub-national level. One example of such a database is the Global Population Database produced by the LandScan Global Population Project of Oak Ridge National Laboratory, which presents data at a 30’’ x 30’’ (arc second) resolution (see Appendix 1).
The main sources of information for developing this module are the national forest inventories and detailed ecological surveys on biomass productivities according to land-use classes. As with the demand module, the information from these sources needs to be disaggregated at the smaller administrative unit of analysis.
Detailed LU/LC inventories are increasingly available both at national and international level. Satellite remote sensing data and analytical techniques have improved considerably since the early days of crude, and often overestimated, classification results. At present, a great variety of products is available with specific spatial and radiometric resolutions that may fit any possible requirement. Analytical methods have improved to include a solid blend of highly sophisticated, but user-friendly, tools, human interpretation and quality control. The LU/LC information produced tends to be designed to respond more and more to user requirements, rather than the self-referencing maps of the past. An interesting example that may contribute considerably to WISDOM analyses in Africa is the recent land cover mapping carried out in the framework of the FAO AFRICOVER Project covering East and Central African countries (Burundi, DR Congo, Egypt, Eritrea, Kenya, Rwanda, Somalia, Sudan, Tanzania, Uganda)15. Promising features of AFRICOVER products are the wall-to-wall coverage of all countries at a good level of detail (scale 1:100 000 - 1:200 000) and the Land Cover Classification System (LCCS) adopted, which well represents the wide variety of low-density vegetation types characteristic of African landscapes, and offers a good basis for the estimation of woody biomass stocking (Drigo, 2001).
Concerning land cover change, national deforestation rates, based on existing information, are periodically published for all of the world’s countries as part of the FAO FRA Programme (FAO, 2001a; FAO 2001b). However, these single country figures are too generic and contribute little to the WISDOM analysis. Only a few tropical countries undertake regular monitoring studies from which sub-national change patterns can be derived (India, Brazil limited to Legal Amazon), none of them in Africa. Richer information on land cover changes was produced for the tropical belt, by region and main ecological zones, by the Remote Sensing Survey of the Forest Resources Assessment (FRA RSS) carried out during the 1990 Assessment, and continued in the 2000 Assessment (FAO, 1996; Drigo, 1996; FAO, 2001a). This study produced highly consistent information on the land cover change processes and trends for the periods 1980-1990 and 1990-2000 through the analysis of satellite time series over a 10% statistical sample of tropical land. Important information from this study is the biomass flux diagram, which provides useful indication on the loss or gain of woody biomass associated with each change in land cover (FAO, 1996).
Additional evidence on the change in land cover occurring in the humid tropical regions over the period 1990-1997 has recently been produced by the TREES II Project of the European Joint Research Center on the basis of a statistical sample of high resolution satellite images covering the dense forest formations of the humid tropics (Achard et al, 2002). Neither of these two studies produced country-level results, but each of the sampling units analyzed in these surveys16, which vary in size between 1 and 3.2 million hectares, may provide some interesting insight on the local patterns of change. It is therefore strongly recommended, for the benefit of many potential users, that the spatial and statistical results of the sampling units be made web-accessible.
Since it is difficult to find accurate country-level data on land-use change processes and their distribution within a given country, there is often no alternative but to use all available pieces of information and to extrapolate them to the entire area of interest. In the case of missing reliable national change estimates, it is therefore advisable to consult the sampling units of the two studies mentioned above, if falling within the country of interest. In such a case, the land cover change index for the WISDOM supply module could be developed on the basis of (i) the detailed results of the sampling units, (ii) the regional and ecological change patterns produced by the same studies, and (iii) other references, such as national deforestation rates.
Information on biomass stocking and productivity of natural forests and plantations may be derived from the integration of land cover information with conventional forest inventory data (volume and yield)17. For instance, the LCCS applied in several African countries, which is based on classifiers independently describing three vegetation layers (trees, shrubs and herbaceous), may be combined with local volume and yield estimates to produce biomass density maps. Stocking and productivity estimates for non-forest formations such as scrublands, homestead gardens, windbreaks, roadside trees, farmland trees, etc., which may represent important woodfuel sources for the rural population, are less common (RWEDP, 1997; Millington, 1994; Kaale, 1990; Ryan & Openshaw, 1991). Usually, this aspect needs to be covered by inference and extrapolation of detailed studies conducted at the project or micro-regional level.
The values of biomass productivity can also be adjusted in terms of its physical or legal accessibility. Physical accessibility may be defined in a Geographical Information System through the use of slope information from terrain models (e.g. GTOPO 30; Digital Chart of the World-derived products), using a buffer analysis based on road networks and settlements distribution, and other parameters. Legal accessibility will discriminate those areas, such as national parks, where wood extraction is forbidden. These areas may be derived from national or international maps, such as the IUCN map of protected areas.
The costs for conducting a WISDOM analysis will vary greatly depending on: 1) the human resources and equipment available at the start of the study; and 2) the availability and access to databases, surveys, censuses, and geo-referenced maps.
With an already operational GIS unit, and full access to the relevant socio-economic and environmental information needed for the study, a GIS expert and a WISDOM analyst may cover the whole analysis in one or two months, almost independently from the area involved. Access to most databases is reasonably cheap and two or three persons may suffice to do the data gathering and analysis. High-level consultants are not needed as the methodology outlined for carrying out a WISDOM analysis is relatively simple.
On the other hand, if a totally new GIS unit needs to be installed and made operative and access to the basic data is rather conflictive, the costs would multiply several times. In this case, substantially longer periods of time would be needed. The acquisition of GIS equipment would be a major component of the costs. In contrast, the human training costs would not be very high. For example, a basic GIS software training program focussed on the elements needed to conduct a WISDOM analysis should not take more than two weeks.
15 Africover data is accessible at http://www.africover.org/
16 The FRA RSS analysed 117
sampling units, each of them covering an entire Landsat scene (185 x 185 Km).
The TREES High Resolution Study analysed 93 sampling units, 39 of which covering
full Landsat scenes and 54 covering quarter scenes (approximately 100 x 100 Km).
17 For an
example of this methodology applied in Sub-Saharan Africa please refer to
Millington (1994).