Impact of Terrestrial Ecosystems of

Russia on the Global Carbon Cycle for

2003-2008: An Attempt of Synthesis

A. Shvidenko, D. Schepaschenko, S. Maksyutov,IIASA (Laxenburg, Austria), Institute of Forest SB RAS (Krasnoyarsk,

Russia), Moscow State University of Forest (Russia), NIES (Tsukuba, Japan)

ENVIROMIS-2010, 5-11 July 2010, Tomsk,

Russia

Prerequisites

• Post Kyoto developments versus

Terrestrial Ecosystems Full Carbon

Account (FCA)

• High variability of reported results of

estimation

• High and mostly unknown uncertainty

• Could the uncertainty of the FCA be made

acceptable for policy makers?

Need of Terrestrial Biota Verified Full

Greenhouse Gas Account

• Key words: Full, Verified, Uncertainty

• Full: ALL ecosystems, ALL land classes and ALL processes – spatially explicit and continuously in time

• Verified: (1) reliable and comprehensive assessment of uncertainties; (2) possibility to manage uncertainties up to an acceptable level

• Uncertainty is an aggregation of insufficiencies of outputs of the accounting system, regardless of whether those insufficiencies result from a lack of knowledge, intricacy of the system, or other causes

• Need of synthesis: what is current state of knowledge of terrestrial ecosystems carbon accounting?

Major principles of the FCA: Integration,

harmonization and multiple constraints

Landscape-ecosystem approach

Process-based models

Flux measurements

Multi-sensor remote sensing concept

Inverse modelling

Terrestrial Biota Full Carbon Account

is a dynamic very complicated

open stochastic fuzzy system (... full

complexity problem)

The direct verification of results of

FCA is not possible

Structural uncertainty cannot be

reliably recognized within any

individually used method

IIASA landscape-ecosystem approach: a semi-empirical background of FCA

• As comprehensive as possible following the requirements of the applied systems analysis

• Relevant combination of flux- and pool-based approaches• Strict mono-semantic definitions and proper classification

schemes; harmonization of these with other approaches• Explicit intra- and intersystem structuring: optimization of

input data; explicit algorithmic form of accounting schemes, models and assumptions

• Spatially and temporally explicit distribution of pools and fluxes

• Correction of many year average estimates for environmental and climatic indicators of individual years

• Assessment of uncertainties at all stages and for all modules of the account – intra-approach uncertainty

• Comparative analysis with independent sources, harmonizing and multiple constraints of the intermediate and final results

Structure of the Integrated Land Information System

Multi-sensor remote sensing concept

• NOAA AVHRR

• MODIS

• GLC-2000

• MODIS-VCF

• LANDSAT TM

• ENVISAT MERIS

• ENVISAT ASAR

• JERS

• ERS-1 and ERS-2

• ALOS PALSAR

Biomass by radars

Last results (Santoro et al. 2010)

report possibility for assessing the

growing stock up to 300-350 m3 with

uncertainty of 10-15%

Courtesy by C.Schmullius

Hybrid land cover – a background of the

Integrated Land Information System(1 km resolution)

Method: Schepaschenko et al. 2010

Results: carbon pools of terrestrial ecosystem

(an example for 2005)

Area, mln ha

Forests 794.7

Open woodland 82.6

Agricultural land 218.6

incl arable land 109.2

Wetland 146.9

Burnt area 27.5

G & Sh 300.8

Productive land 1571.4

Carbon stock, Pg C

Soil 324.0

Incl surface organic

layer 14.2

Live biomass 42.1

Incl forest LB 34.5

Dead wood in forest 8.6

Soil / Biomass C

in forest 3.5 : 1

Reanalysis: Net primary production, Tg C yr-1

by vegetation classes and vegetation zone

Land class Polar TundraSparse

taiga

Middle

taiga

Southern

taiga

Temperate

forestSteppe Desert Total

Forest 0.0 48.4 337.2 1,363.5 636.1 133.4 66.4 9.8 2,594.7

Arable 0.0 0.0 0.0 2.0 44.5 70.6 294.0 1.8 412.8

Hayfield 0.0 0.0 0.3 11.5 25.1 9.4 33.5 15.0 94.8

Pasture 0.0 0.2 0.6 20.1 29.7 22.7 128.2 86.1 287.6

Fallow 0.0 0.0 0.1 4.3 7.1 4.2 5.5 0.1 21.2

Abandoned

arable0.0 0.1 0.5 11.0 59.1 24.1 51.4 5.3 151.6

Wetland 0.0 53.4 76.7 113.4 63.1 7.6 68.2 12.6 395.0

Open

woodland0.0 15.2 34.9 44.0 26.9 4.8 2.7 0.5 129.1

Burnt area 0.0 2.7 4.4 40.0 3.8 0.4 0.8 0.1 52.2

Grass &

shrubland0.3 181.4 42.9 590.9 48.5 42.8 77.0 15.6 999.3

Total 0.3 301.4 497.6 2,200.7 944.0 320.0 727.7 146.7 5,138.3

Net primary production, g C m-2 yr-1

by vegetation classes and vegetation zone

Land class Polar TundraSparse

taiga

Middle

taiga

Southern

taiga

Temperate

forestSteppe Desert Total

Forest - 231 241 291 431 508 445 442 318

Arable - 250 269 377 452 591 533 534 530

Hayfield 98 - 381 366 409 414 363 473 395

Pasture - 313 304 316 382 374 383 605 422

Fallow - - 403 362 491 465 383 307 424

Abandoned

arable- 344 421 520 516 542 485 459 507

Wetland 0 121 213 260 403 652 2031 1380 273

Open

woodland- 246 240 334 486 457 470 619 314

Burnt area 69 139 126 113 448 511 519 517 151

Grass &

shrubland0 126 141 228 571 428 507 322 316

Total 60 126 214 322 444 537 521 572 323

An example of reanalysis: NPP of Russian forests

(2009) based on a new empirical method

Components

14.7%

5.5%

27.9%

29.0%

6.3%

16.6%

Stem Branches Foliage Roots Understory GFF

Age groups

10.6%

30.4%

12.4%

26.8%

19.8%

Young Middleaged Immature Mature Overmature

Dominant species

14.3%

12.1%

2.0%

32.1%

6.9%3.6%

17.9%

3.7%7.4%

Pine Spruce Fir Larch Cedar HWD Birch Aspen Ohters

NPP 2.59 Pg C yr-1, or

318 g C ha-1 yr-1

Uncertainty 7% (CI 0.9)

Difference with a

previous inventory ~1/3

Method:

Shvidenko et al.,

Ecol. Model. 2007

Heterotrophic respiration, Tg C yr-1

by vegetation classes and vegetation zoneLand class Polar Tundra

Sparse

taiga

Middle

taiga

Southern

taiga

Temperate

forestSteppe Desert Total

Forest - 24.9 185.9 870.2 404.3 95.0 49.7 6.9 1,637.0

Arable - 0.0 0.0 1.7 34.4 34.2 210.1 0.8 281.2

Hayfield 0.0 - 0.1 9.8 23.5 8.3 30.7 7.1 79.5

Pasture - 0.1 0.6 19.3 28.1 21.8 110.5 31.6 212.0

Fallow - - 0.1 3.5 5.5 3.1 4.5 0.1 16.7

Abandoned

arable- 0.0 0.3 8.0 39.3 16.5 37.6 2.8 104.5

Bare fellow - - 0.0 0.3 4.2 6.2 37.8 0.6 49.2

Wetland 0.0 44.5 67.4 112.6 62.3 5.7 21.5 3.5 317.5

Open

woodland- 10.9 31.8 48.5 19.0 3.2 2.3 0.4 116.0

Burnt area - 2.0 3.3 30.0 2.6 0.3 0.6 0.1 38.9

Grass &

shrubland0.2 175.5 40.0 272.1 33.0 23.3 58.1 9.1 611.4

Total 0.2 258.0 329.5 1,376.1 656.0 217.7 563.4 63.0 3,463.8

Heterotrophic respiration, g C m-2

by vegetation classes and vegetation zoneLand class Polar Tundra

Sparse

taiga

Middle

taiga

Southern

taiga

Temperate

forestSteppe Desert Total

Forest - 121 132 185 274 359 333 318 199

Arable - 128 191 322 349 286 380 242 361

Hayfield 42 - 190 311 381 366 333 224 331

Pasture - 186 277 304 361 359 330 222 311

Fallow - - 276 300 378 352 311 236 334

Abandoned

arable- 112 236 376 343 370 355 243 349

Bare fellow - - 157 328 358 338 358 220 352

Wetland 0 100 187 259 397 493 640 389 219

Open

woodland- 114 146 232 310 369 357 387 193

Burnt area 32 99 115 124 316 336 428 382 129

Grass &

shrubland0 95 106 171 390 296 379 240 147

Total 35 99 137 188 303 340 364 240 215

Disturbances

Several facts

the total area of wild vegetation fires in Russia in 2003

enveloped 23 million ha including 17 million ha of forests

(4.4 times all Austrian forests);

▲these fires produced direct carbon emissions at ~270

million ton of carbon– more than overall target of the Kyoto

Protocol; the average flux for 2003-2008 is 160 mln t

▲an outbreak of Siberian moth in Russia in 2001 covered

~10 million ha

▲during the recent years insects damaged Canadian

forests at the area of above 20 million ha

Way to estimate uncertainty

• Assessment of precision

• Standard sensitivity analysis (Monte Carlo,

error propagation)

• Transformation precision into uncertainty

• Harmonizing and multiple constraints of

results obtained by independent

methodologies

Fire 2009

Emissions, g C per m2

< 10

11 -

25

26 -

50

51 -

100

101

- 250

251

- 500

501

- 1 0

00

1 00

1 - 1

 500

Source: Global Fire Database GFED3, Giglio et al. 2010, van der Werf et al. 2010

Fire 1997-2009: Average annual area 8.8

mln ha, carbon emissions ~130 Tg C yr-1

Emissions, g C per m2 and year

< 10

11 -

25

26 -

50

51 -

75

76 -

100

101

- 200

201

- 300

> 300

Source: Global Fire Database GFED3, van der Werf et al. 2010

Net Ecosystem Carbon Balance for Russia (average fluxes for 2003-2008, Tg C yr-1, sign “-“ means

sink)

Land classes and components Flux, Tg C yr-1

Forest -563 250

Open woodland -28 21

Shrubs -22 12

Natural grassland -58 26

Agriculture land -32 28

Wetland (undisturbed) -47 26

Disturbed wetland +36 20

Wood products +48 20

Food products (import-export) +18 16

Flux to hydro- and lithosphere +81 36

NECB (NBP) -567 259

NPP: comparison of independent

estimates

0

50

100

150

200

250

0 50 100 150 200 250

Phytomass by [Shvidenko et al., 2002, 2007]

Ph

yto

ma

ss

by

[U

so

lts

ev

, 1

99

8, 2

00

7]

1

2

0

5

10

15

20

25

30

35

40

45

50

2 4 6 8 10 12 14

NPP by [Shvidenko et al., 2004]

Me

tho

ds

1, 2

[U

so

lts

ev

, 2

00

7]

1

2

3

NPP calculated for Russia by 17 DGVM (Gusti 2009) gave the

result +11 % to this study; variability of results by different models

22%

Integration: consistency of information, check of

temporal trends, model-data fusion and model-

data synthesis, etc.Empirical NPP vs. MODIS NPP

MODIS NPP = -105.5381+2.6561*x-0.0048*x^2+2.9488E-6*x^3 (R2 = 0.46)

0 100 200 300 400 500 600 700 800 900 1000 1100

Empirical NPP

0

100

200

300

400

500

600

700

800

900

1000

MO

DIS

NP

P

NPPE

NPPM

Inverse modeling

• Inverse modeling – Results for Eurasia, Pg C year-1

Fan et al.,1999, Science +0.1 0.7

Bousquet et al., 1999, JGR -1.8 1.0

Rodenback et al., 2003, AChPh +0.2 0.3

Gurney et al., 2004, GChB -0.7 1.0

• Inverse modeling – Estimates for boreal Asia, Pg C year-1

Maksyutov et al., 2003 (1992-1996) -0.63 0.36

Gurney et al.,2003 (1992-1996) -0.58 0.56

Baker et al. (1988-2003) -0.37 0.24

Patra et al., 2006 (1999-2001) -0.33 0.78

• Inverse modeling – Results for Russia, Pg C year-1

Ciais et al (2010),4 inversions for 2000-2004 -0.65 0.12

This study (2003-2008), LEA -0.57 0.26

Correction of many year empirical averages for actual climate of

individual seasons: Temperature impact on forest NPP

Examination of different regression models

ΔNPP = F(ΔDD>5oC, ΔP>5oC, Δ[CO2])

ΔHR = Φ(N>0oC, P>0oC, ΔT>0oC, W)

ΔHR = φ (11 seasonal climatic indicators)

Inter-seasonal variability of NPP and HR can

reach 15-20%

Diverse published results

• NPP (Pg C yr-1, for all ecosystems): 2.75 (Filipchuk,

Moiseev 2003), 4.35 (Nilsson et al. 2003), 4.41 (Voronin

et al. 2005), 4.73 (Zavarzin 2007), 5.14 (this study)

• NPP for forest ecosystems (g C m-2 yr-1): 204 (Filipchuk,

Moiseev 2003), 275 (Zamolodchikov, Utkin 2000), 318

(this study), 614 (Gower et al. 2001)

• HSR (Pg C yr-1, for all ecosystems): from 2.78

(Kurganova 2002) to 3.46 (this study)

• Disturbances (forest ecosystems, Tg C yr-1): from about

50 (“managed forests”, Zavarzin 2007) to 200-400

(different studies, including this one)

Reasons for diversity

• Incompleteness of the account

• Oversimplification of accounting schemes

and models used

• Different system boundaries

• Obsolete and uncertain information

• Lack of system design of the account

During 2003-2008

In 2003-2008 land of Russia provided

the carbon sink between 0.6-0.7 Pg C

per year

Carbon balance of selected NH regions from

compiled land-based C accounting data

-750

-250

250

750

1250

1750

Canada USA Mexico EU-25 Russia China NH

Carb

on flu

x into

land e

cosyste

ms (

Tg C

yr-

1) Food products trade

Wood products (incl.Trade)

Peatlands degraded +peat use

Rivers to ocean andlakes

Peatlands & wetlandsundisturbed

Grassland & steppe

Cropland

Shrubland & desert

Forest

Source: Ciais et al. (2010, in press)

Global forest carbon fluxes (Pg C yr-1) of 1990s

and changes (%) over 2000s (vs. 1990s)

Source: Birdsey et al. (2010) in preparation

Warm but not very optimistic future:

the world should be ready to increasing the

global temperature by 4oC

Global average surface temperature scenarios for peak emissions at three

different dates with 3%-per-year reductions in greenhouse gas emissions.

Source: Parry et al. Nature 458, 30 April 2009

There are many unresolved questions and uncertainties

• How are terrestrial ecosystems functioning under dynamic conditions of multiple limitations for life resources?

• How much stable is the direct stimulation of photosynthesis and NPP by the environmental change?

• To what extent do the limitations bound CO2 fertilization effect and how long?

• How much nitrogen deposition is able to eliminate lack of available nitrogen in high latitudes?

• How do all these changes interact with the hydrological cycle, particularly with water stress?

• How will destruction of permafrost impact forest ecosystems of high latitudes?