2 Comparison of Forest Area 1990 derived from FRA1990 respectively FRA2000

The primary reason for undertaking the current study of the forest area by the year 1990 derived from FRA 1990 and FRA 2000 is to evaluate the comparability of the area data from the two assessments.

2.1. Starting point for comparing Forest Area 1990

The number of countries and territories encompassed by FRA 2000 totals 213, distributed as follows:

Africa 56 Asia 49 Oceania 20

Europe 40 N + C America 34 S. America 14

In order to compare the 1990 FRA 1990 forest area (Area90_FRA1990) with FRA 2000 figures for 1990 forest area (Area90_FRA2000) as well as the forest area change figures from the two assessments, the country must be present in both reporting periods [FRA 1990 (1980-90) and FRA 2000 (1990-2000)].

Due to a number of changes of national boundaries through the 1990's a number of new countries have emerged between FRA 1990 and FRA 2000. By region there are the following changes in reporting units mainly due to the break up of earlier countries (units highlighted with grey in table 2-01 below have been used for the comparison of Area90_FRA1990 and Area90_FRA2000 as well as for the comparison of forest area change between FRA 1990 and FRA 2000):

Table 2-01. Aggregated units employed in the comparison between FRA 1990 and FRA 2000.

Continent	Countries FRA 1990	Countries FRA 2000
Africa	Ethiopia	Ethiopia, Eritrea
	Morocco	Morocco, Western Sahara
Asia	Indonesia	Indonesia, West Timor
	Israel	Israel, West Bank, Gaza Strip
Europe	Former USSR (excl. Belarus & Ukraine)	Armenia, Azerbaijan, Georgia, Kazakhstan, Kyrgyzstan, Tajikistan, Turkmenistan, Uzbekistan, Estonia, Latvia, Lithuania, Republic of Moldova, Russian Federation
	Former Yugoslavia SFR	Yugoslavia, The FYR of Macedonia, Slovenia, Bosnia & Herzegovina, Croatia
	Former Czechoslovakia	Czech Republic, Slovakia
	Belgium Luxembourg	Belgium & Luxembourg

Furthermore the 15 small countries with limited forest areas (FRA 2000 forest area totalling 129.000 ha) mentioned in table 2-02 were not reported for FRA 1990 and have therefore been excluded from the comparison:

Table 2-02. Countries excluded from comparison between FRA 1990 and FRA 2000.

Africa	Sao Tome & Principe
Asia	Maldives
Oceania	Cook Islands, Marshall Islands, Micronesia, Nauru, Northern Mariana Islands, Palau
Europe	Lichtenstein, Andorra, Malta, San Marino, Greenland
North & Central America	Saint Pierre & Miquelon
South America	Falkland Islands

This reduces the number of comparable units to 176, distributed as follows:

Africa 53 Asia 37 Oceania 14

Europe 27 N + C America 32 S. America 13

According to FRA 2000 area data these 176 countries and territories account for 3,963,300 ha (or 99.997% of the global net forest area according to FRA 2000) by the year 2000.

2.2. Comparison of Area90_FRA1990 and Area90_FRA2000.

The FRA 1990 forest areas (Area90_FRA1990) have been extracted from FAO Forestry Paper 124 table 3. Forest Area 1980 has been calculated as Area90_FRA1990 minus 10 times annual forest area change.

The 1990 forest areas FRA 2000 (Area90_FRA2000) have been extracted from FAO Forestry paper 140 Appendix 3. The FRA 2000 global tables are available at FAO Forestry website http://www.fao.org/forestry/fo/fra/index_tables.jsp

If the datasets of FRA 1990 and FRA 2000 were completely comparable and if the annual forest area change reported for FRA 2000 was constant throughout the decade 1990-2000 it follows that Area90_FRA1990 should equal Area90_FRA2000.

In reality however the two datasets from FRA 1990 and FRA 2000 deviate from this "perfect fit". Table 2-03 below presents the main 4 causes (points 1-4) identified in explaining differences between Area90_FRA1990 and Area90_FRA2000. Point 5 is applicable to countries that have a good correspondence between Area90_FRA1990 and Area90_FRA2000 and where the same set of inventory data had been used for the two assessments.

One factor affecting the 1990 area data of developing countries is the change of methodology between the estimates of Area90_FRA1990 and Area90_FRA2000 following the discontinuation of the use of the forest area adjustment function for estimating forest area at a standard reference year. The model is described further in FAO Forestry paper 112.

Table 2.3. Classification of differences between Area90_FRA1990 and Area90_FRA2000

1. Improved reclassification of National Classes	Where the reclassification of national vegetation classes into FRA terminology has changed between FRA 1990 and FRA 2000 and thereby includes or excludes forest vegetation that were included or excluded in previous assessments. E.g. a substantial part of the woodland in Kenya was classified as other wooded land during FRA 1990 although it is composed of forests with trees having a crown cover falling into the category of forest according to FRA definitions. For FRA 2000 it has been reclassified as forest.
2. Changes in definitions of FRA classes	The definitions of forest employed by FRA 2000 differed from the definitions employed by FRA 1990, thereby causing the Area90FRA2000 to differ from Area90FRA1990. FRA 2000 employed a homogenous set of definitions1 for all countries whereas FRA 1990 used separate sets of definitions to assess the forest resources of developing countries and the forest resources of the Industrialized countries. The changes in definitions between FRA 1990 and FRA 2000 affecting forest areas all result in a higher Area90FRA2000 simply because the areas defined as forest for the FRA 2000 are broader than areas defined as forest for the FRA 1990. For the developed countries, the height criteria for when vegetation is classified as forest changed from 7 to 5 meters from FRA 1990 to FRA 2000 and the crown cover criteria changed from 20% to 10%. This had lead to increased forest areas of certain countries from FRA 1990 to FRA 2000, and has had the greatest impact on Australia and Other Former USSR Rubber plantations are defined as forest in FRA 2000. Their inclusion as forest plantation in FRA 2000 increases the Area90FRA2000 of rubber producing countries. The minimum area requirement for being included in FRA 1990 was 100 hectares while for FRA 2000 it is 0.5 hectares. This will include of areas of woody formations in the Area90FRA2000 that were not included in Area90FRA1990
3. Improved country data	A number of countries produced new national forest inventories between FRA 1990 and FRA 2000. The results of new inventories and (for the developing countries) the abandonment of the model approach for estimating Area90FRA1990 may explain differences in level of Area90FRA1990 and Area90FRA2000
4. Secondary source information	For countries where national information sets for FRA 2000 were not produced from national forest inventories due to the unavailability of regular forest inventory data.
5. Data sets comparable	This designation has been used where country data sets used by FRA 1990 and FRA 2000 are based on the same inventory data and that show small (<10,000 ha) difference between Area90FRA1990 and Area90FRA2000.

To estimate the comparability of Area90_FRA1990 and Area90_FRA2000 for all countries the table in Appendix 1 was compiled. For each country the main cause of difference between Area90_FRA1990 and Area90_FRA2000 was evaluated (1-4 from table 2-03 above) or cases where the data sets were based on the same inventory and showed little difference, number 5 was ticked.

The size of the difference between Area90_FRA1990 and Area90_FRA2000 (in hectares and percent) has also been indicated for each country.

As mentioned (point 2 in table 2-03) the FRA 2000 definition of forest is broader than the FRA 1990 definition. This is especially the case for the industrialized countries reported by UNECE/FAO where the height criteria for vegetation defined as forest has been reduced from 7 to 5 meters and the crown cover criteria has been reduced from 20 to 10% from FRA 1990 to FRA 2000. Unlike FRA 1990, FRA 2000 applies the same set of definition for the forest resources globally.

For all countries it applies that the minimum area defined as forest decreased from 100 ha for FRA 1990 to 0.5 ha for FRA 2000.

Rubberwood plantations were included in the FRA 2000 figures. This is of particular importance to some South East Asian rubber producing countries.

It follows that for the developed countries, Area90_FRA2000 will generally be larger than Area90_FRA1990. For countries reported by UNECE/FAO² it has therefore been assumed that the primary cause for difference is changes in definitions of FRA classes between FRA 1990 and FRA 2000.

6 industrialized countries deviate from this main trend and show an Area90_FRA2000 which is smaller than Area90_FRA1990 (Austria, Norway, Denmark, Cyprus, Israel and Canada). In these cases the reason for difference has been indicated as new inventory information.

Each of the developing countries has been given a classification of the major cause of difference between Area90_FRA1990 and Area90_FRA2000. This grading is based on comparison of the source datasets underlying the two figures.

For the developing countries the comparison is further complicated by the change of methodology between the figures for Area90_FRA1990 and Area90_FRA2000. The countries affected by change of methodology are indicated by comments in appendix I.

In reality any difference between Area90_FRA1990 and Area90_FRA2000 is due to a mixture of causes. However for the sake of simplicity and to avoid providing an impression of false accuracy the indication has been limited to one cause per country. The findings are indicated in table 2-04.

Table 2-04: Summary of estimated causes of differences between Area90_FRA1990 and Area90_FRA2000

Cause	Number of countries	FRA 2000 Total forest area 1990	FRA 1990 Total forest area 1990	Difference 000 ha	Difference %
1	12	117,570	104,320	13,250	2.54
2	34	1,585,705	1,284,286	301,419	57.86
3	61	1,822,858	1,699,969	122,889	23.59
4	64	422,810	339,327	83,483	16.03
5	5	14,357	14,473	-116	-0.02
Total	176	3,963,300	3,442,375	520,925	100

According to this classification, the main cause of difference between Area90_FRA2000 and Area90_FRA1990 is change of forest definitions between FRA 1990 and FRA 2000 accounting for about 58% of the net difference in forest area between the 2 assessments. Improved data becoming available between FRA 1990 and FRA 2000 in the form of regular inventory data accounts for 24% and secondary source information accounts for about 16% of the net difference each. Improved reclassification of national vegetation classes into FRA classes accounts for 2.5%. Differences between Area90_FRA2000 and Area90_FRA1990 attributed to category 5 are marginal (which also follows naturally from the definition of category 5).

The table in appendix 1 provides the difference between Area90_FRA2000 and Area90_FRA1990 for each country in hectares and as a percentage.

For categories 1-4, countries where the difference between Area90_FRA2000 and Area90_FRA1990 exceed 10 million ha are commented in the following and listed in Table 2-05. This threshold value has been chosen arbitrarily to allow highlighting the countries showing the greatest differences between the area data of the assessments.

Category 1 - Improved reclassification. Countries accounting for main differences.

As improved reclassification can lead to a country showing an Area90_FRA2000 that is greater or smaller than Area90_FRA1990, this category consists both of positive and negative differences. Positive differences indicate Area90_FRA2000 being greater than Area90_FRA1990. This applies for 6 countries totalling 23,789,000 ha difference. Negative differences indicate the opposite (i.e. Area90_FRA1990 being greater than Area90_FRA2000) and applies for 6 countries totalling 10,539,000 ha difference.

With a difference between the assessments of 16,722,000 ha, Kenya is the only country in this category where the difference between Area90_FRA2000 and Area90_FRA1990 exceeds 10 million ha. This is because large areas that were classified as other wooded land during FRA 1990 were reclassified as forest for FRA 2000 as the crown cover exceeds 10%.

Category 2 - Changed FRA definitions. Countries accounting for main differences.

The changes in FRA forest definitions cause a broader area to be defined as forest. Countries falling in this category will therefore always have an Area90_FRA2000 which is bigger than Area90_FRA1990, and hence show a positive difference.

250 million ha (or approximately 84.8%) of the difference within category 2 is composed of two countries [Australia and ex. USSR (excluding Belarus and Ukraine) accounting for 117,522,000 ha and 138,157,000 ha respectively]. This is due to the presence of large areas of open / low height woody vegetation in these countries, that have been included as forest in FRA 2000 with the redefinition of the height criteria from 7 to 5 meters and the crown density criteria from 20 to 10%. USA accounts for a difference of 12,540,000 ha (or 4.2 %) of the difference due to change of FRA definitions. Likewise this is assumed to be due to redefining of some sparsely stocked /low height vegetation types from non-forest in FRA 1990 into forest in FRA 2000. For many industrialized countries there is only a minor difference between Area90_FRA2000 and Area90_FRA1990 as indicated in the table in appendix 1³.

Inclusion of rubber plantations as forest plantations and therefore in the net forest area for FRA 2000 has only had a minor effect on the global figures as the size of the global area of rubber plantations is about 10 million ha. It is however the main explanation to the differences between Area90_FRA2000 and Area90_FRA1990 for certain major rubber producing countries like Thailand and Indonesia.

The effect of lowering the area threshold from 100 ha for FRA 1990 to 0.5 ha for FRA2000 is not possible to evaluate with the current data.

Category 3 - Improved Inventory Data. Countries accounting for main differences.

New inventory data becoming available can result in Area90_FRA2000 exceeding Area90_FRA1990 (this applies for 32 countries providing a total positive difference of 147,374,000 ha). New data can also have the opposite effect, i.e. Area90_FRA2000 being estimated as a smaller figure than Area90_FRA1990, this applies for 29 countries providing a total negative difference of 24,485,000 ha. The resulting net difference within category 3 is therefore 122,889,000 ha, as indicated in table 2.4. Expressed in another way improved inventory data made available during the 1990's is held as the main reason for an additional 123 million ha of forest being "registered" between the two assessments. It is noteworthy that the three main contributors are African countries with a high proportion of open /fragmented forest where the forest edge towards drier woodlands can be difficult to determine. This increase in estimated forest area may indicate, at least for the main contributors below, a tendency of having provided conservative estimates of the forest area for FRA 1990 due to incomplete information.

The following 7 countries are the main contributors in this category (i.e. countries where Area90_FRA2000 exceeds Area90_FRA1990 by more than 10 million ha) and together account for a net difference between Area90_FRA2000 and Area90_FRA1990 of approximately 100 million ha:

Sudan (28,037,000 ha), Mozambique (13,881,000 ha), Zimbabwe (13,258,000 ha), China (11,618,000 ha), Mexico (12,816,000 ha), Paraguay (11,734,000 ha) and Myanmar (10,497,000 ha).

Category 4 - Data based on Secondary sources. Countries accounting for main differences.

For Democratic Republic of the Congo Area90_FRA2000 exceeds Area90_FRA1990 by 27,214,000 ha and for Angola the difference is 47,804,000 ha. In both cases the figures for both assessments are based on secondary references from the 1980's.

The differences between Area90_FRA1990 and Area90_FRA2000 may be attributed to differences in the annual change rates employed to generate the figures for FRA 1990 and FRA 2000 - most likely an effect of the discontinued use of the forest area adjustment function between the assessments.

Table 2-05 Regional / subregional comparison between Area90_FRA1990 and Area90_FRA2000:

Subregion/Region	FRA 2000 Total forest area 1990	FRA 1990 Total forest area 1990	Difference
	000 ha	000 ha
			000 ha	%
North Africa	5,930	6,906	-976	-14.1
West Africa	98,586	96,847	1,739	1.8
East Africa	148,995	99,693	49,302	49.5
Central Africa	236,532	204,815	31,717	15.5
Southern Africa	212,260	136,684	75,576	55.3
Insular Africa	172	144	28	19.4
Total Africa	702,475	545,089	157,386	28.9
West Asia *	22,531	12,767	9,764	76.5
Central Asia **	11,245	9,406	1,839	19.6
South Asia	77,643	77,763	-120	-0.2
East Asia	183,973	170,418	13,555	8.0
South East Asia	235,203	219,318	15,885	7.2
Total Asia	530,595	489,672	40,923	8.4
Total Australia & New Zealand	164,915	47,309	117,606	248.6
Total Other Oceania	36,261	40,945	-4,684	-11.4
Total Oceania	201,176	88,254	112,922	128.0
Northern Europe ***	57,566	53,246	4,320	8.1
Central Europe	50,358	47,553	2,805	5.9
Southern Europe	49,397	39,398	9,999	25.4
Belarus, Moldova, Other Former USSR and Ukraine ****	894,000	754,958	139,042	18.4
Total Europe	1,051,321	895,155	156,166	17.4
North America excl. Mexico	466,684	456,737	9,947	2.2
Central America & Mexico	82,738	70,287	12,451	17.7
Caribbean	5,580	4,252	1,328	31.2
Total North and Central America	555,002	531,276	23,726	4.5
Tropical South America	868,702	849,647	19,055	2.2
Non-Tropical South America	54,029	43,282	10,747	24.8
Total South America	922,731	892,929	29,802	3.3
TOTAL WORLD (excluding countries mentioned in Table 2.2)	3,963,300	3,442,375	520,925	15.1

* Area90_FRA2000 for West Asia is excluding forest area for Armenia, Azerbaijan and Georgia. These did not exist as separate countries for FRA 1990 and for reasons of comparison they have been included under the Other Former USSR.

** Area90_FRA2000 for Central Asia is only for Mongolia. The remaining 5 countries of the subregion (Kazakhstan, Usbekhistan, Turkmenistan, Tadjikistan and Kyrgistan) are included under the Other Former USSR for the same reasons as mentioned above.

*** Area90_FRA2000 for Northern Europe is excluding forest area for Estonia, Latvia and Lithuania. These are included under the Russian federation for reasons of comparability between the assessments.

**** Area90_FRA2000 for Other Former USSR includes the area figures for Armenia, Azerbaijan and Georgia Kazakhstan, Usbekhistan, Turkmenistan, Tadjikistan, Kyrgistan, Estonia, Latvia and Lithuania to make it comparable with Area90_FRA1990

For all regions Area90_FRA2000 is higher than Area90_FRA1990 (most pronounced for Africa, Oceania & Europe).

For Oceania & Europe the larger forest area reported for FRA 2000 is largely due to changed definitions for the industrialized countries reported by UNECE/FAO (all except 4 European countries have been classified in category 2) and Australia dominates the figures for Oceania.

The only subregions that show higher figures for Area90_FRA1990 than for Area90_FRA2000 are Other Oceania, North Africa and South Asia:

Other Oceania: The reduction is primarily due to new forest inventory data becoming available for FRA 2000 for Papua New Guinea providing smaller figures for the total national forest resource for FRA 2000 than the ones used for FRA 1990.

North Africa: The reduction is mainly due to new forest inventory data becoming available for FRA 2000 for Morocco, providing smaller figures for the total national forest resource for FRA 2000 than the ones used for FRA 1990).

For South Asia the difference is due to new inventory data becoming available between FRA 1990 and FRA 2000 for all countries of the subregion. This leads to a Area90_FRA2000 which is lower than the Area90_FRA1990 for India and Nepal while the opposite trend applies for the remaining four countries.

2.3 Summing up the 1990 area comparison.

For the 176 countries compared, Area90_FRA2000 is reported as 3,963 million hectares or about 521 million hectares (15.1%) larger than Area90_FRA1990.

Area90_FRA2000 is larger than Area90_FRA1990 for all regions (most pronounced for Africa, Oceania & Europe).

The difference between Area90_FRA2000 and Area90_FRA1990 is primarily due to changes in FRA definitions of forests between the two assessments causing a broader area to be defined as forest for FRA 2000 than for FRA 1990. The increased Area90_FRA2000 compared with Area90_FRA1990 is primarily due to the inclusion of forest in the industrialized countries within the crown density interval 10%-20% and within the height interval 5-7 meters as a consequence of applying a homogenous set of definitions for all countries for FRA 2000. The change in forest definitions had the greatest impact on the forest area of the Former USSR and Australia. For many other industrialized countries there was a good consistency between the Forest Areas of the two assessments.

FRA 2000 estimates the global area of rubber plantations at about 10 million ha. The inclusion of rubber plantations in FRA 2000 therefore only accounts for a minor part of the difference between Area90_FRA2000 and Area90_FRA1990 although for certain rubber producing countries it is the main explanatory factor e.g. Thailand and Indonesia.

Improved forest inventory information becoming available is evaluated to be the second most important explanation for the difference between Area90_FRA2000 and Area90_FRA1990, followed by differences due to Area90_FRA2000 being evaluated using secondary source information and finally differences owing to improved reclassification into FRA classes of national vegetation classes.

A cause of difference for developing reporting units is the change of methodology between FRA 1990 and FRA 2000 for generating forest area for standard reference year. For FRA 1990 the forest areas of developing countries were generated using the forest area adjustment function which was partly driven by population density and growth. With the acknowledgement of low correlation between demographic parameters and forest area change this practice was discontinued for FRA 2000. FRA 2000 generated figures for the standard reference years 1990 and 2000 by extrapolation using (where available) data sets from two or more points in time.

The overall effect of the changed definitions, changed methodologies and new data becoming available is that the FRA 2000 reports a 15.1 % greater area for the year 1990 as forests compared with the figures from FRA 1990. A general conclusion is therefore that area data from the two assessments should not be used for direct calculation of forest area changes.

When compared with the findings of the FRA 2000 RSS with the data from the country reports there is a statistically significant difference between the two reported areas for all tropical regions. However this was a consistent discrepancy as the figures of the remote sensing survey were lower than the country data for all 3 tropical regions.

The area figures from FRA 1990 and FRA 2000 for change assessments are net forest areas i.e. composed of figures for natural forest and plantations. The figures do not reflect qualitative changes (e.g. reforestation of plantations, regeneration of natural forests, forest degradation and forest improvement). For many countries the national statistics underlying the FRA 1990 and FRA 2000 data are not comprehensive enough to make definitive estimates regarding qualitative changes.

¹ For Terms and definitions employed by FRA 2000 see FRA working paper 1 http://www.fao.org/forestry/fo/fra/index.jsp

² Countries reported by UN-ECE/FAO are the countries of Europe, CIS, North America, Australia, Japan and New Zealand.

³ It is noteworthy that the figures for Canada do not show a similar increase in the 1990 area estimates from FRA 1990 to FRA 2000. Being a large boreal heavily forested country one would expect the change of definitions to cause an increase in the reported forest area. Instead Canada shows a smaller figure for Area90_FRA2000 than for Area90_FRA1990. According to UNECE/FAO 2000 the reason for this is that the figure reported for FRA1990 (from Canadas Forest Inventory 1986) included an estimate of 55 million ha of forest lands in uninventoried portions of Quebec, Ontario and Manitoba. This estimate was replaced in the 1991 national forest inventory by low intensity samples of the previously uninventoried area. The historical estimates were considerably higher than the new estimates hence an apparent reduction in forest area from FRA1990 to FRA2000. Canada has therefore been placed in category 3 where improved inventory data accounts for the main differences.