Click here to access the entire suite of spreadsheets in a zipped file.
It is highly recommended that individuals using the database read the paper "Country Specific Global Forest Data Set V.1" by Brent Sohngen and Colleen Tennity." This paper provides important background information on how the data was developed and data limitations.
This forestry data is provided in individual country spreadsheets. The names of the countries with available data and the corresponding spreadsheet name are given in the attached zipped file. In addition, general information on the sources for the original forest inventory data is provided in the table. These sources are provided in more detail on the references page. Each spreadsheet is constructed similarly, although the timber types in each country vary. This section describes the general layout of the spreadsheets so that readers can get a sense for the type of data provided in them.
Each spreadsheet contains 3 worksheets. Individuals must use later versions of Microsoft Excel that support individual worksheets to view the spreadsheets. The three datasets contained within each spreadsheet are:
- "Data_Output": Contains basic economic data on timber types within the country
- "GTAP_foracre": Contains inventory data on the hectares of land in each timber type, age class (where age class information is available), and AEZ
- "GTAP_CO2": Contains information on carbon in each timber type, age class, and AEZ.
The timber types are designated M1 through M14. Only the United States has 14 timber types. All other countries have fewer types. The main reason for this is that substantial information is available for forest economic modeling within the United States, so disaggregated data for this region has been developed more extensively. It is generally not possible to compare timber types across different countries. For example, M1 in the United States is not the same forest type as M1 in Argentina or Canada.
The data for individual countries is allocated into six regions. The countries included in this dataset, as well as the regions to which they are assigned are shown in table A1. Briefly, the regions are:
(2) Central Asia
(3) Southeast Asia
(5) Central and South America
(6) Developed, Large, and Other Countries
Economic data for each timber type is provided in year 2000 US $. The values are obtained from the global timber model developed by Sedjo and Lyon (1990) and Sohngen et al. (1999). Forest inventories for each of the timber types have been further disaggregated into AEZ’s using the methods described above. As noted, it is currently not possible to take the economic data associated with each forest type and present specific estimates of economic parameters for each AEZ.
WORKSHEET "Data Output"
The information contained in the "Data Output" worksheet provides the fundamental economic values associated with forestry activity and carbon sequestration for the particular timber types. The data is provided only for the timber types identified for each country. The values for particular timber types in countries within a particular region are similar because the data have been obtained form a more aggregated global timber model. The parameters contained in this dataset are listed in Table 2. An additional description of each variable is provided below.
The total area in hectares for each forest type is derived from various sources for each country. Total hectares for each country are obtained from FAO (2003).
Land rental values are estimated for each forest type based on the value of the land associated with the forest type. Land rental is given in year 2000 US $ per hectare per year.
With the exception of a few regions, such as the United States, timber production data by timber type is difficult to obtain. As a consequence, this value is estimated for each timber type in each region with the following methods. First, aggregate national timber production for each region is obtained from the United Nations Food and Agricultural Organization FAOSTATS database (FAOSTATS, 2003). This data is distributed to the specific timber types in each region using timber production proportions derived from Sohngen et al. (1999). These proportions are generated for larger aggregated regions than the individual country data described in this study, so the production numbers for specific timber types should only be considered estimates of actual production of those types within a region.
QA Timber Log Price ($ per m3):
Quality adjustment factors for log prices have been developed for each region. The numeraire price is US southern softwood timber. Prices for all other types are adjusted so that they are relative to this type. Quality adjustment factors take into account price differentials (using average price data from FAOSTATS, 2003), and other factors, such as whether the type typically provides raw material mainly for sawtimber or pulpwood markets. These estimates were developed in the mid-1990's based on data from the 1960's – early 1990's.
QA Net Stumpage Price ($ per m3):
Stumpage prices are the quality adjusted timber log price less the costs of accessing the timber (maintaining and building roads), logging, and hauling to mills. The original data for this was derived from Sedjo and Lyon (1990) and adjusted in Sohngen et al. (1999) to reflect changes in general price levels.
Merchantable Yield function parameters:
Merchantable yield functions represent estimates of the merchantable yield for the timber type in m3 per hectare of roundwood. There are three parameters used in the merchantable yield functions. The form of the merchantable yield function is:
m3 per hectare = exp(a - b/(age - c)) if age > c
m3 per hectare = 0 if age < c
Average regeneration costs are given in year 2000 US $ per hectare. These averages represent average intensity of regeneration effort in the timber type for the country in question. The regeneration estimates are obtained from the model described in Sohngen et al. (1999) for larger aggregated regions than the countries contained in this dataset, and thus can be considered only estimates of the regeneration costs for a particular timber type.
Net Present Value:
Average net present value for each timber type is given in year 2000 US $ per hectare. It is estimated as the site expectation value for the timber type using the following formula,
Where PQA is the quality adjusted net stumpage price, yield is the merchantable yield of the timber type at age a, r is the discount rate (5% used here), and C is the regeneration cost.
Annual Forest Area Harvested:
Annual forest areas harvested are estimated for each timber type using information on the optimal rotation of the species from Sohngen et al. (1999), the area of the species, and total timber harvest in the country.
Forest Carbon Stock:
The stock of carbon in million metric tons (1 metric ton = 1 Mg = 1000 kg ) for each forest type is estimated by using information on the area of forests, the age class distribution, the merchantable yield functions, and the carbon conversion factor to convert merchantable forest stock to tons of carbon. Forest carbon stock (FCS) for each timber type is:
where Areaage is the area of land in each age class in the timber type, Yield is the merchantable yield for timber at a specific age, and α is the carbon conversion factor for forest stock described below.
Forest Carbon Sequestration 10 yr:
Forest carbon sequestration in million metric tons (106 Mg) per year for each timber type is estimated from Sohngen et al. (1999). The estimates are first generated for the broad regional types described in that model. Since that model looks forward, estimates are derived for the period 2000 – 2010, as projected by that model. The results for broad regions are disaggregated to specific countries and timber types based on the proportion of forestland areas of each type in each country.
Forest Carbon Sequestration 50 yr:
Forest carbon sequestration in million tons per year for each timber type is estimated from Sohngen et al. (1999). The estimates are first generated for the broad regional types described in that model. Since that model looks forward, estimates are derived for the period 2000 – 2050, as projected by that model. The results for broad regions are disaggregated to specific countries and timber types based on the proportion of forestland areas of each type, in each country.
Carbon associated with forest stock:
This is the tons (Mg) of carbon per m3 of merchantable wood. The units of this measure are Mg/m3. Note that this is a single parameter used to account for density of specific species (typically around 0.5), whole tree factors (typically 1.4 – 1.6), and forest floor carbon. It includes only above-ground storage, and does not include forest soil carbon. These parameters have been calibrated from numerous data sources, and are described in more detail in Sohngen and Sedjo (2000).
Note however, that the merchantable yield function described above must be slightly adjusted to account for carbon in young forests. In particular, the following yield function should be used:
Yield = exp(a – b/age)
Where the parameters a and b are the same as used in the merchantable yield function described above. The parameter c is not used for estimating forest carbon.
Carbon associated with products:
This is the tons (Mg) of carbon per m3 of roundwood, given in Mg/m3. This parameter should be used only for wood removed from the forest and incorporated into wood products. It is an average value for both sawnwood and pulpwood products and thus can be used for both.
Long term storage percent:
This estimates the proportion of wood that enters into long term wood products. For this study, a 30% average has been assumed for the world, based on studies by Winjum et al., 1998.
Net forest area change (FAO data):
This estimates the net forest area change for each timber type in each country in the dataset. The sum across the regions equals the estimated forest area change based on the recent State of the World's Forests 2003 report from FAO (2003). Country level estimates have been disaggregated to specific timber types using proportions projected by the model described in Sohngen et al. (1999).
Net forest area change 10 yr:
An alternative projection of net change in forest area is estimated using the model described in Sohngen et al. (1999) and Sohngen and Mendelsohn (2003). The estimates are first generated for the broad regional types described in that model. Since that model looks forward, estimates are derived for the period 2000 – 2010, as projected by that model. The results for broad regions are disaggregated to specific countries and timber types based on the proportion of forestland areas of each type, in each country.
Net forest area change 50 yr:
An alternative projection of net change in forest area is estimated using the model described in Sohngen et al. (1999) and Sohngen and Mendelsohn (2003). The estimates are first generated for the broad regional types described in that model. Since that model looks forward, estimates are derived for the period 2000 – 2050, as projected by that model. The results for broad regions are disaggregated to specific countries and timber types based on the proportion of forestland areas of each type in each country.
Marginal Access Cost for inaccessible forest types:
Marginal access costs are applied only for certain inaccessible types occurring in temperate and boreal regions. These marginal access costs represent the cost of building additional access roads and infrastructure to access forests in these regions. They represent the marginal value of the stumpage on that site (the value of harvesting current old growth forests plus the net present value of the future forest on the site). Note that the net stumpage value for these sites will thus be $0 per hectare when subtracting marginal access costs from the value of current and future harvests.
The worksheet "GTAP_foracre" contains information on the area of timberland in different age classes for forests in the country. The data is organized by timber type class (M1 – M14), 10-year age class, and agro-ecological zones. The data used to generate the age class information is given in appendix A, Table A1. Table A1 shows broad regions, individual countries included in the dataset, the source of the total timberland area data, and the source of the age class data. Note that for forestry, over 70% of timber production is derived from harvests in the United States, Canada, Europe, and Russia. Thus, for a large part of industrial timber activity, age class information is available. For other regions, however, this data is not available.
For the countries noted in table A1 as having age class information available, the data on age class distributions is derived from the dataset originally developed in Sohngen et al. (1999) and extended to additional regions in Sohngen and Mendelsohn (2003). For most developed countries, information on age classes has been obtained from original inventory reports developed by those countries, or from reports on the inventories from those countries available in the literature. In particular, for most countries in the classifications "developed and large" and "Europe," age class distributions have been developed.
For most developing countries, information on age class distribution is generally unavailable. There is one exception. High value plantation species have emerged in many subtropical countries as an important source of world fiber supply in the last 25 – 30 years. Age class distributions have been developed for these fast growing plantations based on historical establishment rates as well as information on rotation ages to define harvesting dates. This data was originally developed for study by Sohngen et al. (1999) using historical information from the Pandey (1992), Bazett (1993), ABARE-Jaako Poyry (1999) on establishment rates and harvest rates.
In developing countries, plantation species are maintained in timber type M1 and in some regions, also in timber type M2. A general description of timber types in each region is shown in Table A2. In general, the regions have plantations maintained in the following timber types:
South and Central America: Plantations only in M1 and M2
Africa: Plantations only in M1 and M2
Central Asia: Plantations only in M1
SE Asia: Plantations only in M1
For other forest types in developing countries, age classes have been assumed. These are reflected in the worksheet by showing forest types in different age classes, typically through age class 50. The oldest age classes are usually assumed to hold the most forestland area. Age class 50 is used as the maximum age in the regions listed above because forests in these regions typically mature at this age. Local site conditions for or within specific countries could vary substantially. Beyond year 50, one would expect little additional accumulation of forest biomass or carbon, thus users of the data should consider imposing constraints in their models that limit additional growth beyond year 50 for the regions listed above.
There are two issues that individuals should recognize about timber types M3 – M14 in South and Central America, Africa, Central Asia, and SE Asia. For these timber types and regions, age classes have been assumed. Allowing the age class distribution to adjust in modeling could affect estimates of carbon sequestration in modeling efforts. Users should carefully consider this issue before using this age class information directly in modeling, and consult experts from different regions when developing their models. Second, if using this data for modeling purposes, allowing mature forests in types M2 – M14 (i.e. those already in age class 50) to age beyond year 50 in these same regions could lead to large estimates of carbon sequestration. Users should consider setting maximum ages in their models because many of these forests area already mature and sequester very little additional carbon unless they are harvested and returned to younger age classes. Note that users can modify the age class information with more specific country-level information if it is available.
In addition to the age class distributions, the area distribution of forests is defined for specific AEZ's. The distribution of forests into AEZ's has been accomplished by overlaying the distribution of forest types in the model by Sohngen et al. (1999) over the agro-ecological zones provided by Ramankutty and Foley (1999), as noted above. Forest inventories have also been disaggregated to the AEZ's as well. It is important to recognize, however, that age class distributions have been assumed to be the same for the timber types in each AEZ. This is not likely to hold in reality, but we do not have age class information available by agro-ecological zones for this dataset.
This worksheet provides information on carbon located in different timber types and different agro-ecological zones. In the worksheet carbon is calculated for each age class and AEZ using the yield functions and carbon conversion parameters available in the worksheet “Data_output” and the inventories located in the worksheet “GTAP_foracre”. For this section, age classes are denoted “a”, timber types are denoted “t”, and AEZ's are denoted “z”. Forest carbon stock for an age class in a timber type in an AEZ is given as:
where αt is the carbon conversion parameter described above. To calculate carbon for particular agro-ecological zones or timber types, individuals can aggregate across a, t, or z. Note that αt in the data set varies across timber types, but does not vary across AEZ. In reality, this may vary across both AEZ's and age classes, however, that detailed level of information is not available for this data set.