Acknowledgement ................................................. I
Abstract ...................................................... 111
Zusammenfassung ................................................. V
Table of Contents .............................................. XV
Figures ....................................................... XIX
Tables ...................................................... XXIII
Abbreviations ................................................. XXV
1. Introduction ................................................. 1
2. Scientific background ........................................ 5
2.1 Greenhouse gases and global warming ...................... 5
2.2. The carbon cycle ........................................ 6
2.2.1. Carbon Dioxide ................................... 9
2.2.2. Methane ......................................... 12
2.3. Parameters and processes affecting carbon fluxes
and pools with emphasis on boreal ecosystems ........... 14
2.3.1. Terrestrial vegetation .......................... 14
2.3.1.1. Parameters affecting terrestrial
primary production ..................... 14
2.3.1.2. Vegetation communities and plant
species ................................ 16
2.3.1.3. Developmental stage and activity of
plants ................................. 18
2.3.2. Soil ............................................ 19
2.3.2.1. Climate ................................ 20
2.3.2.2. Overlying vegetation ................... 21
2.3.2.3. Topography ............................. 22
2.3.2.4. Soil characteristics ................... 23
2.3.3. Disturbances .................................... 24
2.3.4. Impact of future climate change on the role
of boreal forests for the global carbon
balance ......................................... 27
2.4. Derivation of carbon cycle components .................. 29
2.4.1. Direct methods .................................. 29
2.4.2. Remote sensing .................................. 31
2.4.3. Spatial modelling with Geographic
Information Systems ............................. 36
2.4.3.1. Predictive vegetation mapping .......... 36
2.4.3.2. Soil landscape modelling ............... 38
2.4.3.3. Regionalisation of landscape
elements and scaling problem ........... 41
2.4.4. Biosphere models for carbon accounting in
boreal ecosystems ............................... 43
2.4.4.1. Process-based models ................... 43
2.4.4.2. Empirical models ....................... 45
2.4.5. Uncertainties in carbon accounting .............. 47
3. Objective and methodology ................................... 49
3.1. Scientific Need ........................................ 49
3.2. Objectives ............................................. 50
3.3. Methodology ............................................ 51
4. Description of study region ................................. 53
4.1. Geographic location .................................... 53
4.2. Climate ................................................ 54
4.3. Topography ............................................. 55
4.4. Soils, permafrost and wetlands ......................... 56
4.5. Vegetation patterns .................................... 58
4.6. Disturbances ........................................... 64
5. Data Basis .................................................. 67
5.1. Climate data ........................................... 67
5.2. Soil database .......................................... 67
5.3. IIASA's vegetation map and forest inventory data ....... 68
5.4. Land cover ............................................. 71
5.5. MODIS and AVHRR Vegetation Continuous Field (VCF)
products ............................................... 73
5.6. Digital Elevation Models ............................... 74
5.7. Water bodies and wetland types ......................... 75
5.8. Disturbances ........................................... 78
6. IIASA's landscape ecosystem model ........................... 79
6.1. Model description ...................................... 79
6.2. Model uncertainties .................................... 83
6.3. Sensitivity Analysis ................................... 84
6.3.1. Approach ........................................ 84
6.3.2. Results of sensitivity analysis ................. 85
6.3.3. Conclusions ..................................... 88
7. Spatial Modelling of Greenhouse Gas Compartments ............ 89
7.1. Predictive vegetation mapping for derivation of
enhanced land cover map ................................ 89
7.1.1. Decision rules .................................. 90
7.1.1.1. Application of ecoregions and
plausibility rules ..................... 91
7.1.1.2. Fuzzy land cover classification
using Vegetation Continuous Field
products ............................... 92
7.1.1.3. Digital elevation models and
topographic rules ...................... 94
7.1.1.4. Classification of wetland types and
tree species in non-mountainous
regions using soil characteristics ..... 95
7.1.1.5. Classification of tundra wetlands
and riparian vegetation using water
bodies map ............................. 97
7.1.1.6. Disturbances ........................... 97
7.1.2. Propagation of uncertainties and
inconsistencies ................................. 98
7.1.3. Sensitivity of post-classification process
and class requirements of initial land
cover map ....................................... 99
7.1.4. Result of rule-based post-classification ....... 102
7.2. Refinement of soil database and estimation of soil
carbon density ........................................ 111
7.2.1. Refinement of soil database using terrain
parameters ..................................... 111
7.2.1.1. Digital terrain analysis .............. 1ll
7.2.1.2. Derivation of combined soil index ..... 116
7.2.1.3. Application of soil index for
prediction of soil properties ......... 118
7.2.2. Estimation of soil organic carbon (SOC) ........ 119
7.2.2.1. Estimation of SOC from original
soil database ......................... 120
7.2.2.2. Estimation of SOC with
consideration of surface area ......... 121
7.2.2.3. Estimation of SOC with
consideration of topographic
effects ............................... 124
7.3. Regionalisation and derivation of Greenhouse Gas
Compartments .......................................... 126
8. Estimation of carbon fluxes and pools based on GGCs ........ 127
8.1. Implementation of Greenhouse Gas Compartments
in IIASA model ........................................ 127
8.1.1. Compilation of input database .................. 127
8.1.2. Modifications of HAS A model ................... 129
8.2. Results and discussion ................................ 130
9. Conclusions and Recommendations ............................ 137
References .................................................... 141
Appendix ...................................................... 165
Appendix A: Used data sets .................................... 165
Appendix B: Attributes of soil database ....................... 166
Appendix C: Attributes and parameters of IIASA's
vegetation database and model ..................... 168
Appendix E: Requirements for input files of land
cover post-classification process ................. 173
Appendix F: Decision rules for land cover post-
classification process ............................ 175
Appendix G: Error propagation and compensation in post-
classification process ............................ 189
Appendix H: Produced rule-based land cover map ................ 191
List of publications .......................................... 201
Figures ....................................................... XIX
Figures
Figure 1: Pools and fluxes of the carbon cycle .................. 7
Figure 2: The range of time scales of major processes within
the global carbon cycle ............................... 8
Figure 3: Terrestrial carbon uptake and storage ................ 11
Figure 4: Location of study region and administrative
regions .............................................. 54
Figure 5: Topography of study region ........................... 55
Figure 6: Distribution of permafrost types and wetlands
in the study region .................................. 57
Figure 7: Ecological zones of the study region ................. 58
Figure 8: Typical tundra with open water bodies ................ 59
Figure 9: Dark coniferous forest on western site of Yenisey
River near Bor (61.5° N, 90° E) ...................... 61
Figure 10:Pine {Pinus silvestris) stand with lichens on
the eastern side of the Yenisey (light coniferous
forest) in the middle taiga near Zotino (61° N,
89° E) ............................................... 61
Figure 11:Closed cedar (Pinus sibiricd) forest in southern
mountain taiga of Eastern Sayan Mountains near
Ermakovskaya (52.5°N, 92.5°E) (August 2004) .......... 62
Figure 12:Typical landscape with sparse forest, alpine
meadow and bare rocks in top aititudinal belts
in Eastern Sayan Mountains (around 52.5°N, 92.5°E) ... 63
Figure 13:Steppe vegetation in Khakasia (54.5 N, 91 ° E) ....... 63
Figure 14:Typical post-fire succession with birch as
dominant species in the middle taiga (61° N,90°E) .... 64
Figure 15:Parent material of soil database at scale 1:1 Mio
for Krasnoyarsk Kray and Khakass Republic ............ 68
Figure 16:Forest inventory test sites in the study region ...... 70
Figure 17:Regional land cover product from MODIS data
for 2003 ............................................. 73
Figure 18:MODIS VCF for 2001, a) tree cover, b) herbaceous
vegetation cover, c) barren ground cover
(in percent) ......................................... 74
Figure 19:Digital Elevation Model (SRTM and GTOPO'30),
500 m resolution ..................................... 75
Figure 20:Permanently open water bodies in the western
Taimyr tundra based on summer acquisitions in 2003 ... 76
Figure 21:Tundra wetiand zone derived from ASAR WS open water
bodies map and Russian base data by density
analysis of lakes ‹ 30 ha ............................ 76
Figure 22:Non-forested peadands in middle and southern taiga
as identified from ASAR WS, location of inventory
site Nizhne Yeniseysk and more detailed looks in
two test sites with open bogs from inventory in
comparison with peadands as derived from ASAR WS ..... 77
Figure 23:Disturbance map for the period 1992 to 2003 .......... 78
Figure 24:Ecoregions used in IIASA model ....................... 80
Figure 25:Uncertainty in NEP across the study region ........... 83
Figure 26:Flowchart of data fusion and application of
decision rules with two examples ..................... 91
Figure 27:Aggregated ecoregions for development and
application of different classification rules ........ 92
Figure 28:Rules for forest type classification using
the AVHRR VCF for pixels, which were not
classified as forest in the original land
cover map, but with at least 5% tree cover
in the MODIS VCF .................................... 93
Figure 29:Rules for classification of barren ground ............ 94
Figure 30:Types of soil parent material, which were used
for differentiation between evergreen needleleaf
species .............................................. 96
Figure 31:Decision tree for differentiation of evergreen
needleleaf trees in non-mountainous regions
using information about soil parent material ......... 96
Figure 32:Differences between input and output land cover
maps ................................................ 100
Figure 33:Differences in tree species distribution between
the four output land cover maps ..................... 101
Figure 34:Result of rule-based post-classification: 500 m
land cover map consisting of three layers for
dominant tree species, non-forest classes and
unvegetated areas ................................... 103
Figure 35:Comparison of the rule-based land cover map with
two forest inventory test sites ..................... 104
Figure 36:Overall pixel-based agreement between aggre-gated
classes of rule-based land cover map and GLC2000 .... 105
Figure 37:Percent of all pixels per ecoregion of rule-based
land cover map that have the same classification
in the IIASA map (user's accuracy) .................. 105
Figure 38:Example for better georeferencing of the
rule-based land cover map compared to the
IIASA map ........................................... 106
Figure 39:Generalised polygons in IIASA map compared to
single or groups of pixels as islands in a
dominating matrix in the rule-based land
cover map ........................................... 107
Figure 40:Examples for different classifications of the
IIASA map and the rule-based land cover map in
comparison with Landsat 7 ETM imagery ............... 108
Figure 41:Frequency distribution of aggregated land cover
classes in percent of total area .................... 109
Figure 42:Frequency distribution of tree species in number
of pixels ........................................... 109
Figure 43:Differentiation of sun and shadow slopes
according to Bodenkundliche Kartieranleitung ........ 112
Figure 44:Slope in % derived from SRTM DEM with 100 m and
500 m resolution. Due to interpolation coarser
resolution leads to a flattening of the relief
and hence to lower slopes ........................... 113
Figure 45:Curvature derived from SRTM DEM with 100 m and
500 m resolution .................................... 114
Figure 46:Classified topographic wetness index derived
from SRTM DEM with 100 m and 500 m resolution ....... 115
Figure 47:Combined soil index derived from addition of
slope, curvature and aspect indices ................. 118
Figure 48:Comparison of original soil polygons with soil
regions derived from digital terrain analysis ....... 119
Figure 49:Mean soil carbon density for the complete soil
profile derived from soil database at scale
1:1 Mio ............................................. 121
Figure 50:Relationship between surface area and pixel area .... 122
Figure 51:Different scales and location of topographically
different test sites for the calculation of the
surface area ........................................ 122
Figure 52:Soil carbon density in the study region derived
from the refined soil database ...................... 125
Figure 53:Comparison between carbon density map with
consideration of topography and map from original
soil database ....................................... 126
Figure 54:NPP density for the year 2003 calculated with
IIASA model using GGC-database with 500 m spatial
resolution and IIASA vegetation database at
a scale of 1:1 Mio, and with LPJ with 0.5° spatial
resolution .......................................... 134
Figure 55:Phytomass С density for the year 2003 calculated
with IIASA model using GGC-database with 500 m
spatial resolution and IIASA vegetation database
at a scale of 1:1 Mio, and with LPJ with 0.5°
spatial resolution. 134
Tables
Table 1: Residence time, global warming potentials for 20,
100 and 500 years, atmospheric concentrations and
annual increase rates of major greenhouse gases ........ 6
Table 2: Sinks and sources of atmospheric CH4 .................. 13
Table 3: Average maximum values for net photosynthesis
under natural atmospheric CO, concentration and
optimal site conditions ............................... 17
Table 4: Maintenance respiration of leaves and needles
in summer at 20°C at the beginning of the night.
Respiratory intensity declines in the course of
the night ............................................. 18
Table 5: Comparison of global emissions from biomass burning
with emissions from all sources (including biomass
burning) .............................................. 25
Table 6: Global estimates of annual amounts of biomass
burning and the resulting release of carbon into
the atmosphere ........................................ 26
Table 7: Selected investigations on soil landscape modelling,
including the modelled soil parameters and the used
topographic attributes ................................ 40
Table 8: Results of selected investigations on the
estimation of the carbon sink by increment of
woody biomass in Russian forests ...................... 47
TABLE 9: Key climatic data for Siberia ......................... 55
Table 10:Land cover classes of SIBERIA-II land cover map
and training data ..................................... 72
Table 11:Ecoregions of IIASA's vegetation database,
selected for sensitivity analysis of IIASA model ...... 84
Table 12:Attributes of IIASA vegetation database, which
do not affect the output parameters of the IIASA
model (directly) ...................................... 85
Table 13:Influence of input parameters on output parameters
of IIASA model ........................................ 86
Table 14:Differences of output parameters produced by
different tree species ................................ 87
Table 15:Deviations of outputs produced from average values
for each tree species and ecoregion to original
outputs in percent .................................... 87
Table 16:Minimum requirements for class definitions of
initial land cover map with current data situation
and with the availability of a new MODIS VCF with
differentiation of leaf type and longevity ........... 102
Table 17:User's and producer's accuracy of some land cover
classes of the rule based land cover product
compared to the forest inventory test sites .......... 104
Table 18:Percentage of overall pixel-based agreement
between rule-based (rb) land cover map, forest
inventory test sites, 1IASA vegetation map and
GLC2000 .............................................. 110
Table 19:Soil index derived from terrain attribute aspect
with thresholds by Bodenkundliche Karueranleitung
(AG Boden 1994) ...................................... 112
Table 20:Soil index derived from terrain attribute slope
with thresholds from Bodenkundliche
Karueranleitung (AG BODHN 1994) ...................... 113
Table 21:Soil index derived from terrain attribute slope ...... 114
Table 22:Soil index derived from inverse topographic
wetness index ........................................ 116
Table 23:Correlation between terrain attributes ............... 116
Table 24:Combination of slope and curvature index ............. 117
Table 25:Combination of slope, curvature and aspect index
to derive combined soil index ........................ 117
Table 26:Impact of surface area (SA) on calculation of soil
carbon content ....................................... 123
Table 27:Results of carbon accounting with and without
consideration of topographic effects ................. 124
Table 28:Application of MODIS VCF tree cover classes for
assignment of statistical growing stock values
to pixels of specific tree species ................... 128
Table 29:Input and output parameters of original I1ASA
model, which have been deleted from C++ source
code for simplification and adaptation of IIASA
model to new input database .......................... 130
Table 30:Results of carbon accounting using the IIASA model
with the produced GGC-database, the IIASA
vegetation database and from independent
calculations with the DGVM LPJ ....................... 131
Table 31:NPP densities (gC/m2-a) in the study region from
various references and measurement methods ........... 133
|
