Module 2.7 Estimation of uncertainties REDD+ training materials by GOFC-GOLD, Wageningen University, World Bank FCPF 1 Module 2.7 Estimation of uncertainties

Module 2.7 Estimation of uncertaintiesREDD+ training materials by GOFC-GOLD, Wageningen University, World Bank FCPF

1

Module 2.7 Estimation of uncertainties

Module developers:Giacomo Grassi , Joint Research Centre

Suvi Monni, Benviroc

Frédéric Achard, Joint Research Centre

Andreas Langner, Joint Research Centre

Martin Herold, Wageningen University

Country examples:

1. Biomass burning

2. LULUCF in Finland

3. Appling the conservativeness approach to the DRC example (matrix approach); this example also relates to Module 3.3

V1, May 2015

Source: IPCC GPG LULUCF

Creative Commons License


2

Example 1: Biomass burning (1/2)

This country example shows a combination of uncertainties for non-CO2 emissions from biomass burning for an Annex I party.

Note that no uncertainty is assumed for GWP values.

The table below shows the data used in the calculations.

Value Uncertainty GWP value

Area burned 1.16 kha ±10%

CH4 EF 43 Mg CH4/kha ±70% 21

N2O EF 0.3 Mg N2O/kha ±70% 310


3

Example 1: Biomass burning (2/2)


4

Example 2: LULUCF in Finland (1/3)*

For its GHG inventory, . T.

* Synthesis from the National Inventory Report of Finland (Statistics Finland

2013).

Table: Inventory uncertainties


5

Example 2: LULUCF in Finland (2/3)

For forest land remaining forest land, uncertainty analysis is done by pools: living biomass, mineral, and organic soils. Both sampling uncertainty of national forest inventory (NFI) and model uncertainty (used for reporting) are considered.

gain-loss method is applied, which requires data regarding increment and losses:

• Uncertainty (U%) includes sampling of volume increment (4-13%), BCEF (0.5-2.5%),


6

Example 2: LULUCF in Finland (3/3)

Conversion to/from forest land and related KP activities: estimation of C stock change in all pools is done by: AD x EF.

Uncertainty of AD due to sampling was estimated from NFI:

●Because of small land areas involved, a high sampling error is reported: e.g., U% for deforestation is 30%

U% in the increment of living biomass and in the mineral and organic soil emission factors is based on expert judgement.

For emissions from soils under conversions of forest land to cropland and grassland, preliminary estimates are 60–150%.


7

Example 3: Appling the conservativeness approach to the Democratic Republic of Congo (DRC) example (matrix approach) (1/14)

(This example also relates to exercise 4 and Module 3.3)

IPCC basics to estimate forest C stock changes Emissions = activity data (AD) x emission factor (EF)

Six land uses: forest land, cropland, grassland, wetlands, settlements, other lands

Methods to estimate C stock changes:Gain-loss: growth minus harvest minus other losses (all tiers)

Stock change: difference of C stock over time (only Tiers 2–3)

IPCC would require Tier 2/3 methods for EF in "Key Categories" (likely including deforestation and degradation in most cases), but most developing countries are not ready yet for Tier 2/3.


8

Example 3: REDD+ matrix (2/14)

ToFrom

Forest land Other land

Forest land Forest degradation Forest conservation

Sustainable management of forestsEnhancement of carbon stocks

Deforestation

Other land Enhancement of carbon stocks(Afforestation/ Reforestation)

How would REDD+ activities fit into IPCC land uses?

Stock change method: C before minus C after

Gain-loss: growth minus harvest minus other losses

IPCC (very uncertain) FAOSTAT: very difficult to get the right data!

Difficult to get data

Overall, unlikely to estimate C stock changes from degradation with tier 1


9


Don’t forget degradation!

Estimates of carbon emissions from degradation(expressed as an additional percentage to the emissions from deforestation)

Study areaAdditional emissions

due to forest degradation

Reference

Humid tropics +6% Achard et al. 2004

Brazilian Amazon, Peruvian region

+25-47% Asner et al. 2005

Tropical regions +29% Houghton 2003

South East Asia +25-42% Houghton and Hackler

1999

Tropical Africa +132% Gaston et al. 1998


10


To

From

Forest land

Other landIntact (natural) forest

Non–intactforest

Forest land

Intact (natural)

forestForest conservation

Forest degradation

Deforestation

Non–intact forest

Enhancement ofC stocks

(forest restoration)

Sustainable management of

forestsDeforestation

Other land -Enhancement of

C stocks(A/R)

Modified IPCC land transition matrix (REDD+ matrix)

Stock change method: C before – C after

Gain-loss: growth – harvest – other losses


11


How to identify non–intact forests?

Among different possible approaches, forest edges may be used as a simple and pragmatic proxy to identify non–intact areas (boundary forests), or at least may be a first step to be complemented by other more accurate approaches (i.e., high-resolution remote sensing).

The underlying assumption is that forests that are sufficiently remote from nonforested areas (i.e., at a certain distance from roads, navigable waters, crops, grasslands, mines, etc.) are protected against significant anthropogenic degradation.


12


Input: binary forest maps using the methodology of FAO Remote Sensing Survey

Intensified sampling 60x60 m²

Treatment: morphological spatial pattern analysis (MSPA)

Biome specific: rainforest in Congo Basin (Edge size=500m)

Could as well be called exposed, potentially degraded, managed, or simply other forests.

Example of identification of boundary forests


13


Area transition matrices for a biome: Congo rainforests (ha 000s)

a. 2000–2005 b. 2005–2010

NFL 2005

BFL 2005 OL 2005 Total

2000 NFL 2010

BFL 2010 OL 2010 total

2005

NFL 2000 78,424 828 26 79,278 NFL

2005 76,950 1407 66 78,424

BFL 2000 - 24,747 316 25,063 BFL

2005 - 24,976 599 25,575

OL 2000 0 - 123,839 123,839 OL 2005 0 - 124,182 124,181

Total2005 78,424 25,575 124,181 228,180 Total

2010 76,950 26,383 124,847 228,180

Case study in DRC

NFL = natural forest land; BFL = boundary forest; OL = other land.


14


Deforestation (in 5 yrs)

Degraded (in 5 yrs)

Sust. mgd.

forests

Conser-vation

Total

NFL to OL

EFL to OL

NFL to EFL

EFL to EFL

NFL to NFL

Area

(103 ha)

Historical 2000-2005 26 316 828 24,747 78,424 228,180

Ref. level 2005-2010 +100%

=52

+100%=

632

+100% =1,656

24943 76,716 228,180

Actual 2005-2010 66 599 1,407 24,976 76,950 228,180

Difference actual - RL

27 125 -249 -125 221 0

Area-based hypothetical reference level

NFL = natural forest land; BFL = boundary forest; OL = other land


15


Estimating C stock changes for REDD activities

Once the transition matrix for AD is done, each AD will need to be multiplied by the relevant EF to get C stock change for each REDD+ activity:

• For natural forest, Tier 1 EF are available from IPCC

• For boundary forests, data may be taken from the literature (or a crude assumption of half of C stock of NFL may be considered)

Uncertainties values need to be associated with each EF.

The proposed approach requires that the same Tier 1 EF (stratified by forest and climate type) be used in both reference level (RL) and in the accounting period. This means that the errors of EF in the RL and accounting period are fully correlated.


16


Deforestation (in 5 yrs)

Degraded (in 5 yrs)

Sust. mgd.

forestsConservation

Total

NFL to OL

EFL to OL

NFL to EFL

EFL to EFL NFL to NFL

Area (103 ha)

Difference actual - RL 15 -33 -249 -125 221 0

C losses (-), tC/ha (a) -150 -73 -78

C increment (+), tC/ha/yr (...) (…)

Cumulated credits(+) or debits (-) in 2010, MtC (b) -2,3 2,4 19,3 (…) (…) 19.4

NFL = natural/intact forest land; BFL = boundary forest; OL = other land.

(a) Assuming these values of biomass C stocks: NFL, 155 tC/ha (IPCC 2006); EFL, NFL/2 (or 50% degradation on average in exposed forests); OL, 5 tC/ha.

(b) Calculated as the difference in area (actual minus RL) x the C stock change.


17


NFL to OL BFL to OL NFL to BFLTier-1 C stock change (tC/ha) 150 73 78

Uncertainty % (95%CI) 52 80 125

NFL BFL OLTier-1 C stocks (tC/ha) 155 78 5Uncertainty % (95%CI) 50 75 50

Assume that estimates for (accounting period minus RL) obtained with adequate methods for AD but not for EF (Tier 1)

When the uncertainties above are combined, total uncertainty of the emission reduction (19,4 Mt C) becomes >100% (95%CI)

Taking uncertainties into account

How to deal with the fact that this country used Tier 1 (highly uncertain) EF for a key category? see next slides


18


As part of the Kyoto Protocol review process, UNFCCC has approved conservativeness factors linked to specific uncertainty ranges. Essentially, these factors use the 50% confidence interval.


19

0

10

20

30

40

50

Red

uced

em

issi

ons

(M tC

)


95% confidence interval

50% confidence interval

In this example, by discounting the emissions reduction by about 30% (following the approach of KP review), the risk of overestimating the reduction of emissions is significantly reduced.

Lower bound of 50% CI (≈14MtC)


20


In conclusion, the REDD+ matrix may allow one to estimate C stock change from deforestation/degradation based on IPCC Tier 1.

The application of a conservative discount to address the high uncertainty of Tier 1–based estimates increases the credibility of any possible claim of result-based payment.

The simplicity and cost-effectiveness of this approach may allow:

• Broadening the participation to REDD+, allowing those countries with limited forest monitoring capacity to join

• Increasing the credibility of emission reductions estimated with Tier 1, while maintaining strong incentives for further increasing the accuracy of the estimates, i.e., to move to higher tiers


21

Recommended modules as follow-up

Module 2.8 to learn more about the role of evolving technologies for monitoring of forest area changes and changes in forest carbon stocks

Modules 3.1 to 3.3 to proceed with REDD+ assessment and reporting


22

References

Achard, F., Eva, H. D., Mayaux, P., Stibig, H.-J. and Belward, A., 2004. Improved estimates

of net carbon emissions from land cover change in the tropics for the 1990s. Glob.

Biogeochem. Cycles 18, GB2008.

Asner, G. P., Knapp, D. E., Broadbent, E. N., Oliveira, P. J. C., Keller, M. and Silva, J. N.,

2005. Selective logging in the Brazilian Amazon. Science 310, 480–2.

Bucki, M., D. Cuypers, P. Mayaux, F. Achard, C. Estreguil, and G. Grassi. 2012. “Assessing

REDD+ Performance of Countries with Low Monitoring Capacities: The Matrix Approach.”

Environmental Research Letters 7 (1) 014031.

Gaston, G., Brown, S., Lorenzini, M. and Singh, K. D., 1998. State and change in carbon

pools in the forests of tropical Africa. Glob. Change Biol. 4, 97–114.

Houghton, R. A., 2003. Revised estimates of the annual net flux of carbon to the

atmosphere from changes in land use and land management 1850–2000. Tellus, B, 55,

378–90.


23

Houghton, R. A. and Hackler, J. L., 1999. Emissions of carbon from forestry and land-use

change in tropical Asia. Glob. Change Biol. 5, 481–92.

IPCC (Intergovernmental Panel on Climate Change). 2000. Good Practice Guidance and

Uncertainty Management in National Greenhouse Gas Inventories. (Often IPCC GPG.)

Geneva, Switzerland: IPCC. http://www.ipcc-nggip.iges.or.jp/public/gp/english/.

IPCC, 2003. 2003 Good Practice Guidance for Land Use, Land-Use Change and Forestry,

Prepared by the National Greenhouse Gas Inventories Programme, Penman, J., Gytarsky,

M., Hiraishi, T., Krug, T., Kruger, D., Pipatti, R., Buendia, L., Miwa, K., Ngara, T., Tanabe, K.,

Wagner, F. (eds.). Published: IGES, Japan.

http://www.ipcc-nggip.iges.or.jp/public/gpglulucf/gpglulucf.html (Often referred to as IPCC

GPG)

Statistics Finland. 2013. Greenhouse Gas Emissions in Finland, 1990–2011: National

Inventory Report under the UNFCCC and the Kyoto Protocol. Helsinki: Statistics Finland.

http://unfccc.int/national_reports/annex_i_ghg_inventories/national_inventories_submissio

ns/items/7383.php.

Documents

Module 2.7 Estimation of uncertainties REDD+ training materials by GOFC-GOLD, Wageningen University, World Bank FCPF 1 Module 2.7 Estimation of uncertainties