7.1.1 The distribution of monitoring sites in the Terrestrial Ecosystems Monitoring Sites (TEMS) database was evaluated with respect to region, biome, land use, etc.. The analysis seemed to reveal less about the gaps in coverage, than about the shortcomings of the database itself. These shortcomings are probably a consequence of the way the database was assembled. Questionnaires were distributed from the GTOS Switzerland office with an approximately 30% return on the initial mailing. When data from the questionnaires and those from additional networks were entered in the TEMS database, little or no additional support was available to improve, check and/or complete the necessary site data. The TEMS database therefore retains many inconsistencies, errors and missing values, including several sites which are listed without their specific location (Table 1). The inconsistency of the database may be its largest drawback.
7.2.1 The shortcomings are obvious when one analyses the distribution of the individual sites.. Europe is well represented, while substantially fewer sites (Table 2) are located on other continents. Latin America, Africa and the Indonesian Archipelago have the worst coverage with only 90, 82 and 12 sites, respectively. The few sites on these latter continents also are relatively small. The monitoring sites only cover less then 0.005% the total land area, although site size does not need to be a limitation. The sites available for analysis can also generate significant biases. Less than 40% (Table 1) of the sites had size data associated with them. The sites analysed all belong to GTOS site tiers 2-4. Large-scale experiments and remotely sensed data are absent from the current version of TEMS.
7.2.2 The continental distribution is skewed which also has consequences for assessing site representation of the different biomes. The observed biome distribution created by Olson et al. (1993) and the simulated biome distribution by Prentice et al. (1992; Table 3) have been used. The distribution of Olson et al. (1983) has the advantage that it characterises not only natural vegetation types but also some human-dominated land-cover types. The biome distribution of Prentice has the advantage of predicting the kind of ecosystems which should appear as remnants in areas otherwise too intensely used to be so classified by Olson, et al (1983). However, both databases have a relatively coarse resolution (0,5 x 0.5o longitude and latitude) and were developed for different purposes than monitoring land-cover change As a result of these limitations, our current analysis defers examination of gaps in coverage by site numbers, instead illustrating some of the improvements of TEMS which needs to precede such analysis.
7.3.1 Both databases show that temperate deciduous and mixed forests are the best represented biomes in the TEMS database (Table 3). These land cover types coincide with the major European land cover types and are a consequence of the global distribution of sites in the most heavily developed countries. All other land cover types are less well represented. Both the databases (simulated by Biome and observed by Olson et al.) agree in this respect. The European domination of site numbers also results in the majority of human-dominated land cover types delimited by the Olson database being those found in Europe. The data clearly reveal the skewed distribution of the biome numbers in the database.
7.4.1 This examination can also be readily extended to other overlays. Tables 4 and 5 show an example of the TEMS database with the wetland and related land-cover and the agricultural classes, respectively, of the Olson et al.(1983) database. While monitoring sites in individual agricultural land cover classes appear reasonably distributed, almost all wetland classes clearly are inadequately represented. Similar evaluations can be performed with other data sets. Groombridge (1992), for example, presents a map with biodiversity hotspots. The basic TEMS map can be used to evaluate whether the TEMS sites adequately cover these hotspots and therefore, are capable of assessing changes in regions important to biodiversity. The linkage between such capability and the TEMS database must be developed if GTOS and GCOS are to provide answers to the questions asked and issues raised by the Subsidiary Body on Scientific and Technical Advice (SBSTA) and the international conventions.
Table 1: Overall availability of major characteristics of the TEMS database with 1398 sites.
Availability of data by |
Number of sites |
% |
Location |
1289 |
92 |
Site level |
837 |
60 |
Altitude |
548 |
39 |
Extent |
532 |
38 |
Precipitation |
218 |
16 |
Temperature |
194 |
14 |
Site History |
118 |
8 |
Table 2: Distribution of the sites in the TEMS database by continent.
Class name |
Number of sites in class |
% |
Number of sites used |
Extent total class in Mha |
Area covered by sites in Mha |
% |
Tier undefined |
Tier 1 |
Tier 2 |
Tier 3 |
Tier 4 |
Tier 5 |
Undefined |
119 |
8.9 |
59 |
136122 |
0.097 |
0 |
52 |
0 |
5 |
19 |
43 |
0 |
North America and Canada |
137 |
10.3 |
83 |
19614.3 |
0.173 |
0.001 |
64 |
0 |
17 |
47 |
9 |
0 |
Mexico, Central America |
23 |
1.7 |
17 |
2817.6 |
0.081 |
0.003 |
21 |
0 |
1 |
1 |
0 |
0 |
South America |
57 |
4.3 |
28 |
17941.1 |
0.485 |
0.003 |
44 |
0 |
2 |
11 |
0 |
0 |
Europe (including western U.S.S.R) |
719 |
54.0 |
175 |
10376.7 |
0.098 |
0.001 |
180 |
0 |
4 |
107 |
428 |
0 |
Africa |
82 |
6.2 |
47 |
30028.8 |
0.34 |
0.001 |
59 |
0 |
0 |
16 |
7 |
0 |
Asia |
129 |
9.7 |
76 |
41847.5 |
1.267 |
0.003 |
57 |
0 |
3 |
61 |
8 |
0 |
Indonesian and other Pacific I |
12 |
0.9 |
2 |
3100.3 |
0.01 |
0 |
8 |
0 |
0 |
3 |
1 |
0 |
Australia and New Zealand |
49 |
3.7 |
16 |
8101.4 |
0.047 |
0.001 |
12 |
0 |
1 |
35 |
1 |
0 |
Antarctica |
1 |
0.1 |
0 |
0 |
0 |
n.a. |
0 |
0 |
1 |
0 |
0 |
0 |
Greenland |
4 |
0.3 |
2 |
2279.6 |
0.97 |
0.043 |
2 |
0 |
0 |
0 |
2 |
0 |
Total: |
1332 |
100 |
505 |
136122 |
3.569 |
0.003 |
0 |
0 |
34 |
300 |
499 |
0 |
Table 3: Distribution of the sites in the TEMS database by Biome (Prentice et al., 1992).
Class name |
Number of sites in class |
% |
Number of sites used |
Extent total class in Mha |
Area covered by sites in Mha |
% |
Tier undefined |
Tier 1 |
Tier 2 |
Tier 3 |
Tier 4 |
Tier 5 |
Undefined |
120 |
9 |
59 |
136197.4 |
0.097 |
0 |
52 |
0 |
6 |
19 |
43 |
0 |
Ice/Polar Desert |
6 |
0.5 |
2 |
1839.1 |
0.004 |
0 |
5 |
0 |
0 |
1 |
0 |
0 |
Semidesert |
33 |
2.5 |
27 |
5735.2 |
0.099 |
0.002 |
20 |
0 |
2 |
11 |
0 |
0 |
Tundra |
34 |
2.6 |
11 |
8449 |
0.973 |
0.012 |
26 |
0 |
1 |
4 |
3 |
0 |
Taiga 97 |
97 |
7.3 |
19 |
13619.3 |
0.011 |
0 |
45 |
0 |
0 |
23 |
29 |
0 |
Cold Deciduous Forest |
9 |
0.7 |
1 |
4199.2 |
0.003 |
0 |
4 |
0 |
0 |
2 |
3 |
0 |
Cool Grass/Shrub 26 |
26 |
2 |
13 |
4861.9 |
0.025 |
0.001 |
11 |
0 |
3 |
9 |
3 |
0 |
Cool Conifer Forest 43 |
43 |
3.2 |
10 |
3340 |
0.056 |
0.002 |
10 |
0 |
0 |
12 |
21 |
0 |
Cold Mixed Forest |
9 |
0.7 |
5 |
666.1 |
0.004 |
0.001 |
3 |
0 |
1 |
4 |
1 |
0 |
Cool Mixed Forest |
179 |
13.4 |
64 |
5910.6 |
0.113 |
0.002 |
58 |
0 |
6 |
41 |
74 |
0 |
Temperate Deciduous Forest |
402 |
30.2 |
118 |
5983.9 |
1.085 |
0.018 |
74 |
0 |
8 |
62 |
258 |
0 |
Evergreen/Warm mixed Forest |
74 |
5.6 |
29 |
5783.6 |
0.152 |
0.003 |
20 |
0 |
1 |
27 |
26 |
0 |
Warm Grass/Shrub |
50 |
3.8 |
31 |
10807 |
0.095 |
0.001 |
30 |
0 |
0 |
10 |
10 |
0 |
Hot Desert |
25 |
1.9 |
18 |
20893.6 |
0.218 |
0.001 |
18 |
0 |
2 |
4 |
1 |
0 |
Xerophytic Woods/Shrub |
94 |
7.1 |
30 |
10770 |
0.035 |
0 |
35 |
0 |
2 |
35 |
22 |
0 |
Tropical Rain Forest |
40 |
3 |
21 |
8133 |
0.124 |
0.002 |
27 |
0 |
2 |
10 |
1 |
0 |
Tropical Seasonal Forest |
33 |
2.5 |
21 |
8272.9 |
0.099 |
0.001 |
28 |
0 |
0 |
5 |
0 |
0 |
Tropical Dry Forest/Savannah |
58 |
4.4 |
26 |
16917.9 |
0.375 |
0.002 |
33 |
0 |
0 |
21 |
4 |
0 |
Total |
1332 |
100 |
505 |
136197.4 |
3.569 |
0.003 |
0 |
0 |
34 |
300 |
499 |
0 |
Table 4: Distribution of the sites in the TEMS database by the Olson et al. (1983) wetland land-cover classes.
Class name |
Number of sites in class |
% |
Number of sites used |
Extent total class in Mha |
Area covered by sites in Mha |
% |
Tier undefined |
Tier 1 |
Tier 2 |
Tier 3 |
Tier 4 |
Tier 5 |
Undefined |
755 |
56.7 |
324 |
30624.4 |
2.464 |
0.008 |
361 |
0 |
19 |
150 |
225 |
0 |
Cool Crops |
62 |
4.7 |
22 |
0 |
0.007 |
n.a. |
18 |
0 |
0 |
14 |
30 |
0 |
Warm Farms |
259 |
19.4 |
77 |
0 |
0.415 |
n.a. |
51 |
0 |
9 |
59 |
140 |
0 |
Paddyland |
19 |
1.4 |
12 |
0 |
0.618 |
n.a. |
6 |
0 |
1 |
11 |
1 |
0 |
Warm Irrigated |
13 |
1 |
4 |
0 |
0.009 |
n.a. |
2 |
0 |
0 |
8 |
3 |
0 |
Cold Irrigated Drylands |
3 |
0.2 |
2 |
0 |
0 |
n.a. |
2 |
0 |
1 |
0 |
0 |
0 |
Cool Irrigated Drylands |
0 |
0 |
0 |
0 |
0 |
n.a. |
0 |
0 |
0 |
0 |
0 |
0 |
Med. Grazing |
8 |
0.6 |
5 |
0 |
0.015 |
n.a. |
6 |
0 |
0 |
2 |
0 |
0 |
Cool Field/Woods |
12 |
0.9 |
3 |
0 |
0.003 |
n.a. |
4 |
0 |
1 |
5 |
2 |
0 |
Warm Woods/Fields |
63 |
4.7 |
17 |
0 |
0.029 |
n.a. |
16 |
0 |
1 |
17 |
29 |
0 |
Cool Woods/Fields |
78 |
5.9 |
28 |
0 |
0.007 |
n.a. |
25 |
0 |
2 |
21 |
30 |
0 |
Warm Field/Woods |
60 |
4.5 |
11 |
0 |
0.002 |
n.a. |
8 |
0 |
0 |
13 |
39 |
0 |
Total |
1332 |
100 |
505 |
30624.4 |
3.569 |
0.012 |
0 |
0 |
34 |
300 |
499 |
0 |
Table 5: Distribution of the sites in the TEMS database by the Olson et al. (1983) agricultural land-cover classes
Class name |
Number of sites in class |
% |
Number of sites used |
Extent total class in Mha |
Area covered by sites in Mha |
% |
Tier undefined |
Tier 1 |
Tier 2 |
Tier 3 |
Tier 4 |
Tier 5 |
Undefined |
1291 |
96.9 |
490 |
6006.9 |
3.545 |
0.059 |
481 |
0 |
33 |
294 |
483 |
0 |
Antarctica |
1 |
0.1 |
0 |
0 |
0 |
n.a. |
0 |
0 |
1 |
0 |
0 |
0 |
Bogs, Bog Woods |
2 |
0.2 |
0 |
0 |
0 |
n.a. |
2 |
0 |
0 |
0 |
0 |
0 |
Mangroves |
10 |
0.8 |
6 |
0 |
0.006 |
n.a. |
5 |
0 |
0 |
3 |
2 |
0 |
Heaths, Moors |
7 |
0.5 |
1 |
0 |
0.002 |
n.a. |
2 |
0 |
0 |
1 |
4 |
0 |
Marsh, Swamp |
4 |
0.3 |
4 |
0 |
0.003 |
n.a. |
3 |
0 |
0 |
0 |
1 |
0 |
Coastal Edges |
13 |
1 |
2 |
0 |
0.003 |
n.a. |
2 |
0 |
0 |
2 |
9 |
0 |
Ice |
3 |
0.2 |
1 |
0 |
0.004 |
n.a. |
3 |
0 |
0 |
0 |
0 |
0 |
Water |
1 |
0.1 |
1 |
0 |
0.005 |
n.a. |
1 |
0 |
0 |
0 |
0 |
0 |
Total |
1332 |
100 |
505 |
6006.9 |
3.569 |
0.059 |
0 |
0 |
34 |
300 |
499 |
0 |
7.5.1 The meeting participants discussed the analysis to date and the next steps. It was agreed that such an analysis is essential to: a) identify the greatest weaknesses of the present surface observation networks, and b) guide priority efforts for future improvements. It was agreed that future gap analyses should be undertaken with respect to only the networks/sites that have agreed to be part of the global terrestrial observing networks and that capabilities of the TEMS 2 should be used for this purpose. Once the required data and analysis capabilities are available, the priorities for gap analysis should be based on the importance and urgency of specific climate change terrestrial issues at the time.
Recommendation 15: Future gap analyses should be undertaken with respect to only the networks/sites that have agreed to be part of the global terrestrial observing networks.