Previous PageTable Of ContentsNext Page

PART 1: Methodology

The methodological approach followed in the study is based on the following key characteristics of wood energy systems3:

Geographic specificity. The patterns of woodfuel production and consumption, and their associated social, economic and environmental impacts, are site specific [Mahapatra and Mitchell, 1999; RWEDP, 1997; Masera, Drigo and Trossero, 2003]. Broad generalizations about the woodfuel situation and impacts across regions, or even within the same country, have often resulted in misleading conclusions, poor planning and ineffective implementation.

Heterogeneity of woodfuel supply sources. Forests are not the sole sources of woody biomass used for energy. Other natural landscapes such as shrub lands, and other land uses such as farmlands, orchards and agricultural plantations, agro-forestry, tree lines, hedges, trees outside forest, etc. contribute substantially in terms of fuelwood and, to a lesser extent, as a raw material for charcoal production.

Users’ adaptability. Demand and supply patterns influence each other and tend to adapt to varying resource availability. This means that quantitative estimates of the impacts that a given demand pattern has on the environment are very uncertain and should be avoided [Leach and Mearns, 1988; Arnold et al., 2003].

Woodfuel Integrated Supply/Demand Overview Mapping (WISDOM)

In order to cope with the characteristics mentioned above, the FOPP-WE has developed and implemented the Woodfuel Integrated Supply/Demand Overview Mapping (WISDOM)4 methodology, a spatially-explicit planning tool for highlighting and determining woodfuel priority areas or woodfuel hot spots (Masera, Drigo and Trossero, 2003). To date, the WISDOM approach has been implemented in Mexico (Masera et al, 2004), Slovenia (Drigo, 2004) and Senegal (Drigo, 2004b) as a tool to support national-level wood energy planning.

WISDOM, especially when applied at regional level, does not replace a detailed national biomass demand/supply balance analysis for operational planning but rather it is oriented to support a higher level of planning, i.e. strategic planning and policy formulation, through the integration and analysis of existing demand and supply related information and indicators.

More than absolute and quantitative data, WISDOM is meant to provide relative/qualitative values such as risk zoning or vulnerability ranking, thus highlighting, with the highest possible spatial detail, the areas deserving urgent attention and, if needed, additional data collection. In other words, WISDOM should serve as an assessing and strategic planning tool to identify priority places for action.

A detailed description of the WISDOM approach can be found in Masera, Drigo and Trossero, (2003).

The use of WISDOM involves five main steps:

The diagram in Figure 1 provides an overview of WISDOM main steps.

Figure 1: WISDOM steps

The flowchart of the estimation process is shown schematically in Figure 2.

Figure 2: Flowchart of main analytical steps

In order to visualize the various steps of the process, Figures 3 to 12 show the cartographic data layers that were used and produced in a small area of Tanzania, along Lake Victoria.

Specific aspects of the data used and processing carried out in the Demand, Supply and Integration modules are discussed in the following sections.

Example of data layers

The following maps are shown as example of the sequence of spatial data layers produced and involved in the analysis of woodfuel consumption and production potential.

Figure 3: Layout of the sample area5

The two following maps represent the input (original LCCS data) and the main output of the supply module (biomass stocking), which was created through the allocation of biomass density values to each of the 2947 individual LCCS classes according to individual tree, shrub and herbaceous layer present in the classes, and to the ecological zone.

Figure 4: Example of original LCCS data

Figure 5: Example of Woody biomass stocking.

Figure 6: Example of population distribution, 30 arc-second data set.

These maps show population distribution in approximately 1 km2 cells, then categorized as rural or urban.

Rural population data was further categorized as rural “settlements” and rural “sparse” using 2000 inhabitants/km2 as a threshold.

Figure 7: Example of Rural population within 5 arc-minute cells.

These maps provided number of people in approximately 9 x 9 km cells through the aggregation of 10 x 10 30 arc-second data.

Three independent maps were provided: one reporting urban population, one rural “sparse” and one rural “settlements”.

(in this region no rural settlements were identified)

Figure 8: Example of urban population within 5 arc-minute cells.

Figure 9: Example of woodfuel consumption by cell

This map was created using population data and average per capita consumption by rural, settlement and urban dwellers estimated for each country.

Figure 10: Example of woody biomass stock within 5 arc-minute cells.

Map created through the aggregation of the biomass stock of the original LCCS maps

Figure 11: Example of woody biomass increment within 5 arc-minute cells.

The increment was estimated as a fraction of stocking and reduction of the proportion of wood used for other non-energy use.

Figure 12: Example of cell-level supply/demand balance

This map was created subtracting the consumption to the average sustainable annual productivity of each cell. This map indicates the capacity of local wood resources to satisfy local demand and it is therefore meaningful for the poorest consumers depending on local supplies –though less so for marketed woodfuels.

Selection of spatial base

The spatial base, which is defined by the smallest territorial unit for which demand and supply parameter are estimated, it is the result of a compromise between the available data and the wanted level of analysis. In this case the key variables such as population for the demand module and land cover data for the supply module, presented a spatial resolution that was higher than the aimed level of analysis:

• population distribution data was available in raster format at 30 arc-second cell size, which represents individual units of 0.92 x 0.92 km in size (at the equator).

• population distribution data at 5 arc-minute resolution (9.2 x 9.2 km on the equator) derived from aggregation of 10x10 30 arc-second data.

• land cover data produced for all countries by the Africover Project using LCCS and available in vector format, presented a very high spatial resolution comparable to map scale between of 1: 100 000 and 1:200 000.

The 30 arc-second resolution, although potentially consistent with land cover data, appeared far too fine for the purpose of the study and for achieving a meaningful supply/demand relation.

The 5 arc-minute cells cover a territory in which supply/demand balance analysis is meaningful, especially for the fraction of woodfuel consumers that depend on local resources. More importantly, this format represents the spatial base of the FAO Food Insecurity Vulnerability Mapping System (FIVIMS). This means that keeping this format for WISDOM analysis and wood energy mapping guarantee a direct link and contribution to the FIVIMS thematic layer and ensures that WISDOM contributes to the analysis of food insecurity and poverty mapping.

Sub-national administrative data was also available, although the size of the units varied considerably from country to country. The sub-national unit level was also used as a secondary level of aggregation in the supply-demand balance analysis, but only for the aggregation of 5 arc-minute cell data.

Demand Module

The scope of the Demand module was to distribute the consumption of woodfuels at the defined minimum spatial level of analysis (5 arc-minute grid cells).

The statistical and spatial data available for the development of the demand module is listed below:

Woodfuel consumption data

• Estimates of total national consumption of fuelwood and charcoal at year 2000 from various sources and with occasional subdivision by rural/urban and household/non-household consumption (i-WESTAT data).

• Per capita fuelwood and charcoal consumption data by sector and by area from consumption surveys conducted (before 2000) at sub-national and local levels. Most of these surveys were carried out in the 1980s and only few references are reasonably recent (GFPOS data and other national references).

Population data

• National statistics of rural, urban and total population estimated at year 2000 (UN population statistics).

• Distribution of 2000 population by 30 arc-second cells classified as rural and urban (FIVIMS).

• Distribution of (sparse) rural population for 2000 by 5 arc-minute cells derived from the aggregation of 30 arc-second rural population cells with less than 2000 inhabitant /km2 (FIVIMS).

• Distribution of rural settlement population for 2000 by 5 arc-minute cells derived from the aggregation of 30 arc-second rural population cells with more than 2000 inhabitants /km2 (FIVIMS).

• Distribution of urban population for 2000 by 5 arc-minute cells derived from the aggregation of 30 arc-second urban population cells (FIVIMS).

The population distribution datasets were provided by the Geographic Information Systems Group of SDRN working on the Food Insecurity Vulnerability Mapping System (FIVIMS). These maps are based on Landscan Global Population Database 2002 and adjusted to 2000 UN population data. The urban boundaries, necessary to separate and distribute urban and rural populations, were generated by FAO/SDRN on the basis of Radiance Calibrated Lights of the World, 2000, and UN urban population data for 2000.

Process of estimation

The approach followed for estimating per capita consumption went as follows:

Rural settlements

Designation of woodfuel consumption by rural settlements (in addition to rural and urban) was done for the countries with densely populated rural areas (rural areas with over 2000 persons per km2)6. In these cases woodfuel consumption was assumed to have a consumption pattern somewhere between the urban and average rural levels. In general, rural settlement presented a higher charcoal consumption and lower fuelwood consumption rate relative to average rural conditions. The consumption in the remaining rural areas (with population density below 2000 inh/km2), labelled “rural sparse”, was derived from the remaining “unallocated” consumption and resulted in a higher fuelwood and lower charcoal consumption relative to average rural conditions.

The per capita consumption values, which referred to UN population statistics, were finally adjusted to the actual number of rural and urban population reported in the maps (see details in Annex 3, Summary table). The final per capita consumption values are shown in Table 1.

Table 1
Per capita consumption of wood for energy, in m3 of fuelwood and wood used for charcoal, in all sectors
(household and non-household)

 

Summary values of per capita total wood consumption for energy (hh + ind) adjusted to 5min population map's values
(m3 / person / year)

Country

Rural sparse

Rural settlements

Rural (general)

Urban

Burundi

1.48

1.08

 

0.70

Congo, Democratic Republic

   

1.17

1.97

Egypt

0.35

0.24

 

0.21

Eritrea

0.90

0.74

 

0.59

Kenya

0.78

1.03

 

0.83

Rwanda

0.50

1.00

 

1.86

Somalia

   

0.66

0.53

Sudan

   

1.09

1.09

Tanzania

   

1.33

1.76

Uganda

0.86

1.36

 

1.70

Supply module

The analysis and spatial representation of woodfuel supply sources includes several phases of progressive refinement that may be summarized as follows:

• estimation and distribution of woody biomass stocking of natural formations (forests, other wooded lands) and anthropic landscapes (trees outside forests, forestry and agricultural plantations, farmlands and settlements);

• estimation and distribution of annual sustainable productivity and the share available for energy use; and

• segmentation of wood resource data by legal and physical accessibility classes.

The first phase represented an essential pre-requisite to the subsequent analytical steps on productivity and accessibility and constituted the main focus of the present study’s supply module. The second phase, (estimation of annual productivity) was carried out by applying generic average growth rates due to lack of adequate reference data and to time constraints. The third phase, concerning physical and legal accessibility, requires considerable additional spatial processing work that could not be undertaken. To reduce the impact of the missing accessibility parameters, the analysis of supply/demand balance was constrained within 5 arc-minute cells (approximately 9 x 9 km) and therefore limited to the resources accessible to poor households given assumed gathering capacities.

The definition of the study areas, i.e. selected East and Central African countries, was done by taking into account the specific contribution that recent land cover data available for the 10 countries could make towards the assessment of biomass stocking. The land cover information was based on the Land Cover Classification System (LCCS), which was developed and applied in the framework of Project Africover (Di Gregorio and Jansen, 2000). The new land cover classification encompasses one third of Africa and offers a uniform and coherent support to the estimation/stratification of woody biomass into discreet density classes and subsequently, to the assessment of the state and distribution of woodfuels resources.

Of particular relevance for the present study was the on-going activity, supported by the Italian Istituto Agronomico per l’Oltremare and carried out by Valerio Avitabile, on the estimation and distribution of biomass and carbon stocking using LCCS data. The supply module of the present study benefited from collaboration with IAO in the definition of the methodology and in the collection and review of existing literature references on volumes and biomass stocking. The biomass stocking data used for the supply module are based on the first comprehensive set of volume/biomass reference values collected by ecological zones resulting from the FAO/IAO collaboration. However, since the IAO initiative will continue beyond the completion of the present study, a more advanced biomass and carbon stocking data set will be available at a later stage.

Estimation of woody biomass stocking and distribution

Direct field measurements of woody biomass are extremely rare. Relatively more common are forest inventories although they are usually limited to the “commercial” assortments (higher diameter classes of timber species) of productive forests. Unproductive forests, in terms of timber quality, degraded forest formations, fallows, shrub formations, trees outside forests, farm trees, etc. are systematically excluded from conventional surveys, although they usually represent the main sources of fuelwood and wood for charcoal.

The comparative advantage of LCCS data for estimating biomass stocking rests with the detailed description of the physiognomic characteristics of land units, which are qualified through a system of classifiers that provide a detailed description of tree, shrub and grass layers.

The method for the estimation of biomass density (biomass stocking in tonnes per hectare) was based on the combination of two data sets:

1. Volume and biomass indicators based on field inventory results and other surveys of the main formations and ecological zones, providing minimum, maximum and mean volume and biomass density values in “normal” conditions or referring to specific crown cover densities.7

2. LCCS data providing actual crown cover conditions for the main life forms (trees, woody, shrubs and herbaceous) and for all possible combinations of agricultural and natural formations.

Ecological stratification

Several existing ecological classification systems were considered (ICIV 1980, White, 1983). Given the limited number and uneven spatial distribution of field data on volumes and biomass, preference was given to a relatively simple classification system, with few classes within which an acceptable number of reference values could be found.

The ecological stratification was based on the FRA 2000 ecological zone map (Figure 13), which indicates seven main zones in the ten countries covered by this study:

For practical reasons, the drier zones (steppe, desert and shrub land) were grouped to form a single class and therefore the ecological zones of interest remained five only.

Estimating biomass density of LCCS classes

In total 525 single land cover classes were found in the maps, which gave origin to as many as 2947 individual LCCS codes, including single classes but also numerous class combinations (land units presenting a mixture of two or three single classes). These figures are a good indication of the wide variety of conditions described by LCCS and also of the relative complexity of assigning biomass values to each LCCS class.

In the process of assigning biomass density values, volume and biomass data was used as a reference for the potential stocking (minimum, maximum) in the various ecological zones while LCCS data was used to adjust the biomass stocks according to actual physiognomic conditions of land cover types and their geographic distribution. Annex 3 provides the values assigned to the LCCS crown cover categories of all life forms (trees, woody and shrubs) in each ecological zone and other land cover types.

Depending on the availability of reference data, minimum, maximum and mean values of biomass stocking were identified for all life forms, and ecological zones. In the subsequent phases of analysis, however, the mean values were used as main reference.

Biomass stocking in forest plantations

Estimates of woody biomass stocking and productivity of forestry plantation for the countries of the study are rare, scattered and probably biased since they often refer to successful plantation sites or controlled test areas while excluding poorly stocked ones. The values of Mean Annual Increments (MAI) and rotation periods reported in FRA databases provided an indication of the range of values but realistic average values are difficult to determine because the weight and representation of the existing values are not known.

On the other hand, an overview analysis of fuelwood plantation in developing countries (FRA 2000), the average productivity assumed for Africa was 6 m3/ha/yr. For Ethiopia and Sudan, in which fuelwood plantations represent 88 and 78% of all plantations, respectively, the average productivity assumed was 11 and 5 m3/ha/yr. Moreover, the FRA country report for Ethiopia indicated an average woody biomass stocking for plantations at 40 tons/ha, equal to the average value given for natural forests. These values are lower than those reported by plantation statistics.

Consequently, lacking reliable estimates of actual plantations, the stocking values of closed tree formations for the corresponding ecological zone were used as reference. It was assumed that a plantation of average condition could reach, at end rotation, a biomass density comparable to that of a closed canopy natural forests of the same site. Since the age class of plantations is not reported in LCCS, the stocking was assumed to be mid-term, i.e. the value assumed at end rotation.

Given the limited availability of data on woody biomass of orchards, and other agricultural crops, the estimates for the classes occurring in LCCS were done on inference and more or less educated guesses. It is hoped that in time, these approximate estimates will be replaced by more reliable values.

Estimates for sustainable production of wood for energy

Mean Annual Increment

Estimating woody biomass in the area studied and included in the LCCS legend was a complex task, aggravated by the virtual absence of reliable field data for the study area.

For the scope of the present study a simple approach was adopted, under the assumption that in normal conditions there is a direct positive relation between the stocking and the mean annual increment (MAI) of natural formations (Openshaw, 1982). This assumption, supported by increment data (Micski, 1989, Bowen et al., 1987, FAO 1982), and the fraction applied by Openshaw (2.5 percent) appeared realistic. Therefore, the MAI was estimated as 2.5 percent of biomass stocking for all formations except forest plantations.

As mentioned above, the MAI values for forest plantations reported by the literature were extremely variable (FAO 2001, 2002). However, considering the various references available, a MAI of 5 percent of the stocking at end rotation appeared realistic. Consequently, since the biomass stocking of plantations was considered as of that at end rotation, the MAI applied for forest plantation was estimated as 10 percent of the assumed “mid-rotation” biomass stocking.

Fraction of woody biomass used for energy

In the countries of this region woody biomass is predominantly used for energy. This is clearly shown in Table 2, which reports the ratio between FAOSTAT’s information on woodfuel production and on total roundwood production. On average, the ratio for year 2000 was estimated at 0.94. This factor was systematically applied to the total woody biomass productivity values to quantify the amount of woody biomass available for energy uses after deduction of the amount utilized for other purposes.

Table 2
Fraction of woodfuel production in total roundwood production at year 2000 as reported by FAOSTAT.

Country

Woodfuel / total roundwood

Burundi

0.94

Congo, Dem Republic of

0.95

Egypt

0.98

Eritrea

1.00

Kenya

0.91

Rwanda

0.93

Somalia

0.99

Sudan

0.88

Tanzania, United Rep of

0.90

Uganda

0.91

Average

0.94

Source: FAOSTAT 2005

Integration module and definition of priority areas

The scope of the integration module was to combine, by land units (5 arc minutes cells or sub-national units), the parameters developed in the demand and supply modules to highlight areas of potential deficit or surplus according to estimated consumption levels and sustainable production potentials.

The main indicator so far produced was represented by the balance, within the 5-arcminute cells, between the fraction of the potential sustainable productivity available for energy uses and the total woodfuel consumption. This parameter does not consider the transportation of woodfuels between distant production and consumption sites—an element that would require additional analytical steps.

As is, this parameter provides a useful indication of the ease, or difficulties, that poor rural households face in acquiring their daily subsistence energy.

In order to visualize these results under the administrative angle, the results by 5 arc minute cells were subsequently aggregated at sub-national unit level.

3 Definitions of main terms are reported in Annex 1
4 WISDOM is the fruit of collaboration between FAO’s Wood Energy Programme and the Institute of Ecology of the National University of Mexico. To date, WISDOM was implemented in Mexico (Masera et al.,2005), in Slovenia (Drigo 2004) and Senegal (Drigo, 2004).
5 The sample area is located in Northwest Tanzania, along Lake Victoria (provinces of Kagera and, partly, Mwanza and Shinyanga)
6 Egypt, Eritrea, Kenya, Uganda, Burundi and Rwanda
7 The main references resulting from the bibliographic search and the system of reference values adopted in this study are reported in Annex 3.

Previous PageTop Of PageNext Page