Comparing Forest Area 1990 derived from FRA 1990 (3,442 million ha) with the area for 1990 derived from FRA 2000 (3,963 million ha) there is a difference of 521 million ha. In other words FRA 2000 arrives at a global forest area that is approximately 15 % larger than the 1990 estimate. The 1990 area figures derived from FRA 2000 were for all regions larger than the figures derived from FRA 1990. The main reason for this is the change in FRA definitions of forest from FRA 1990 to FRA 2000, leading to an expansion of the areas defined as forest for FRA 2000. The lowering of the threshold value for crown cover and height criteria from 20% to 10% and 7 to 5 meters respectively in industrialized countries from FRA 1990 to FRA 2000 is the most significant of these changes. This led to increased forest areas being reported for the industrialized countries and had the greatest impact on reported forest area for Australia and Other Former USSR. Due to the substantial difference in the areas reported by FRA 1990 and FRA 2000 area figures from the two assessments should not be used for change assessments.
In spite of FRA 2000 reporting a 15 % larger forest area than FRA 1990, the global net annual forest area change reported by FRA 2000 was -9.4 million ha. This represented an apparent decrease in the global net annual rate of forest area loss by 3.7 million ha from the -13.1 million reported for FRA 1990.
A number of changes in methodology and in the quality of the underlying datamaterial between FRA 1990 and FRA 2000 contributed to the differences in the size of estimated forest area change from the two assessments.
Concerning changes in methodology, the discontinuation of the use of the deforestation model for generating forest area change figures for the developing countries was of particular importance. The use of the model was abandoned for FRA2000 after acknowledging that the correlation between forest area change and demographic parameters used as inputs was weak at national level. For FRA 2000 the estimates for forest area change were generated where possible by timeseries information and elsewhere by expert opinion and secondary source information. The model may for some countries have generated high estimates of net annual forest area loss during the 1980's.
Concerning the quality of the underlying data, new surveys and inventories had become available for a number of countries between FRA 1990 and FRA 2000. Some major reporting units e.g. Other Former USSR did not report on forest area change for FRA 1990 and their contribution to the forest area change figures from FRA 1990 is therefore 0. For FRA 2000 these missing forest change figures had become available. The average reference year of the source data used for compiling the FRA 2000 figures was 1994.
Although the source data provided for FRA 2000 represented an improvement from the source data of FRA 1990, many countries still lack comprehensive and comparable forest inventories. This is particularly the case for many African countries.
The change in FRA definitions of forest did not have a major impact on the rate of forest area change as the change rates were generated by date and definitions from the respective assessments only - i.e. not using data from two points in time based on different definitions.
Compared with the findings of the FRA 2000 RSS, the FRA 2000 country data overestimated the rate of deforestation for Africa, necessitating a positive adjustment of the regional forest area change figure for Africa. At a global level this adjustment was balanced by a downward adjustment of the figure for global plantations establishment assuming a success rate of 70% for plantations.
The main trends at regional level resulting from this adjustment was a + 1,279 million ha net adjustment of the figures for AreaChangeFRA2000 for Africa (resulting in a net difference between the forest area change figures of FRA 1990 and FRA 2000 for Africa of only 34,000 ha).
For Asia there was a net adjustment for plantations survival of 1.050 million ha (resulting in a net difference between the forest area change figures of FRA 1990 and FRA 2000 for Asia of -174,000 ha).
The adjustment for the remaining regions was smaller due to smaller plantation areas.
After adjustment at regional level, Europe incl. Former USSR (difference between FRA 1990 and FRA 2000 net annual forest area change figures of 927,000 ha), North & Central America (difference of 851,000 ha) and in particular South America (difference of 2,321,000 ha) are identified as the main contributors to the observed positive development in the forest area change rate.
As mentioned a number of changes in methodology and source data between the assessments make a direct comparison problematic, even after the above regional adjustments, for the below main reasons:
Methodology change. The methodology for estimating deforestation rate of the developing countries changed between FRA 1990 and FRA 2000. FRA 2000 RSS did not find significant differences between the forest area change rates of FRA 1990 and FRA 2000 for the tropical domain. This makes it likely that the above difference for South America is overestimated and that part of the reason may be the model generated FRA 1990 figures for forest area change during the 1980's.
Availability of improved data. For many countries and most clearly for the reporting units that reported AreaChangeFRA1990 as n.a. (e.g. Other Former USSR) new data had become available between the assessments. This is a source of error of the reported global figure for AreaChangeFRA1990.
The current comparison has been performed on two datasets relating to two different periods in time and that have been generated using different methodologies (for the developing countries) and different definitions of forest (mainly affecting the industrialized countries). The comparison is further complicated by the difference in quality and timeliness of the underlying source documents.
With due respect to these differences it has nonetheless been attempted to make a comparison between the datasets by comparing and making adjustments for the findings of the remote sensing survey and by adjusting for plantation success rate at regional level. The figures indicate a global decrease in the rate of net forest area loss owing mainly to developments in the industrialized countries primarily Europe and North America and in particular to a slowdown in the reported rate of deforestation for South America.
Concluding comments that may be drawn from the study based on the reported net forest area change figures is that there seems to be a decrease in the rate of net forest area loss from the 1980's (FRA 1990) to the 1990's (FRA 2000). Due to the nature of the FRA source data it is not been possible to generate confidence intervals for the estimates. FRA 2000 RSS has however not been able to prove statistically significant differences between the forest cover change rates of the tropical regions for the two decades.
FRA 2000 estimates a net tropical natural forest area loss of 14.2 million ha per year, meaning that almost 1 percent of the tropical forest was lost every year during 1990's. In spite of an apparent improvement from the 1980's to the 1990 owing mainly to a natural expansion of forest in the industrialized countries, natural forests in the tropics are still being lost at an alarmingly high rate. At global level the loss of natural forest has continued at roughly the same high levels over the last 20 years.
Although the source data used for FRA 2000 was more timely and of a better quality for assessing forest area and forest area change than was the case for FRA 1990, there is still room for improvement of the national statistics that form the foundation of FRA. As mentioned less than half the countries relied on time series for estimating forest area change for FRA 2000. Obvious areas for future work at national level are the production of comparable timeseries for producing reliable estimates of forest area change with increased focus on qualitative changes.
