Developing FRELs/FRLs and Potential Implications of Economic Incentives

Developing FRELs/FRLs and Potential Implications of Economic Incentives

Series of modules for GHG National Inventory from LULUCF with REDD+ Focus

MINAM MRV REDD+ Workshop

Lima, Peru, 28-30 April 2014

Contributors: Arief Wijaya, Lou Verchot, Daniel Murdiyarso, Martin Herold,

Arild Angelsen, Erika Romijn, Sabastine Ugbaje and Kalkidan Mulatu

Outline

Forest reference emission level/reference level definition

FRELs/FRLs estimation approaches

Stepwise approach proposal

Case study in setting up REL in four tropical countries

Impacts to REDD+ financial incentives

CIFOR Global Comparative Study on REDD+

GCS Module 3: REDD+ MRV and Carbon Emissions measurement

– Assessment of major deforestation drivers

– Setting national reference emission levels (RELs)

– Monitoring, reporting, verification (MRV) for REDD+

– Six case study countries: Brazil, Peru, Indonesia, Vietnam, Tanzania and Cameroon

Setting a forest reference emissions levels/forest reference levels (FRELs/FRLs)

Forest reference levels (FRLs) and forest reference emission levels (FRELs) are a business as usual (BAU) baseline of carbon emission projection to assess a country’s performance in implementing REDD+ (UNFCCC, 2011)

The uses of FRLs:– to establish a reference point or benchmark against which actual

emissions (and removals) are compared

– to serve a benchmark for payments in a results-based REDD+ mechanism

FRELs/FRLs development is data-driven approach (need exhaustive and reliable activity data and emission factors)

Proposals for estimating FRELs/FRLs

REL using historical emission (Santilli, et al, 2005 and Mollicone, et al, 2007)

REL using adjusted historical emission. This approach assumed that future deforestation is estimated from historical data with some adjustments (e.g. population density, agriculture land demand, GDP, etc) (Amano, et al, 2008)

REL with forward looking (may or may not consider historical emission) (e.g. Strassburg, et al, 2009)

Capacity gaps of 99 non-annexes I countries

Consideration of factors for capacity assessment:

1.Requirements for monitoring forest carbon on national level (IPCC GPG)

2.Existing national capacities for national forest monitoring

3.Progress in national GHG inventory and engagement in REDD

4.REDD particular characteristics: importance of forest fires, soil carbon, deforestation rate

5.Specific technical challenges (remote sensing)

Proposing stepwise approach for setting up FRELs/FRLs

Guidance suggests to use historical data; adjusted for national circumstances

Data driven approach: the less data a country has the more it should rely on the data – need to manage the uncertainties

Why a stepwise approach?• Data availability and quality varies• IPCC GPG LULUCF (2003) use Tiers as a mechanism to deal with

uncertain & incomplete data for estimation on national level• Match data availability and uncertainty and allow for broad country

participation• Motivation to reduce uncertainties over time

Proposing a Tier-ed approach – Tier 1

Simple extrapolation using historical forest area estimates – assumes no change in trend

May use IPCC approach 1 data (FAO FRA)

No consideration of drivers information

Importance of consistency and transparency

Uncertainty: ± 75% of prediction and accuracy is based on available trend data (bias?) – corridor approach

Suitable for exploration and international comparison

Can be applied to most of developing countries

Simple REL for 4 countries using FAO FRA data

Higher tier approaches for REL development

Retain predictive power of historical trend data and move to more driver-based assessment and predictions

Include data-driver reasoning for deviations from historical trend (i.e. national circumstances)

Higher tiers use national data:– Deforestation and emissions and understanding of historical processes

using data on drivers and activities causing forest carbon stocks change

– Establish relationship with underlying causes (proxies)

– Justification why and how deforestation varies from historical trend on the level of drivers and activities

Example of spatial analysis of national deforestation drivers

Stepwise approach for FRELs/FRLs

(Herold, et.al, 2011)

Modeling deforestation drivers

Variables M1 M2 M3 M4

Historical annual deforestation (2000-2005) ✓ ✓   ✓

Other land use/socio-economic factors (2005)   ✓ ✓  

Forest cover (share of land area) (2005)   ✓ ✓ ✓

Square of forest cover (share of land area) (2005)       ✓

Identification of deforestation driversCountries DriversVietnam 

- Drivers vary from region to region- Conversion to agriculture & shrimp Farms- unsustainable extraction of woods/land clear- Land conversion to infrastructure- Unsustainable logging (legal, illegal)- Forest fires

Indonesia 

- Oil palm- Mining- Other: Roads, Settlement, aqua culture dev’t- Forest fires- Degradation and logging (Sumatra, Kalimantan)- Swidden (slash-and-burn) Agriculture

Brazil    

- Agriculture and ranching, pasture lands- Human settlement- Logging- Soy bean/sugarcane plantation- Infrastructure

Peru 

- Mining sector- Fast economic growth- Expansion of Agriculture- Logging for domestic uses and trade- Unplanned human settlement and infrastructure

Deforestation drivers modeling

Available Variables(Mean Values)

CountriesVietnam Indonesia Brazil Peru

Annual Deforestation (2005-2010) (ha) 518,192 769,162 1,275,148 163,335Annual Deforestation (2000-2005) (ha) 420,712 527,024 3,266,150 91,096

Forest cover (2005) (ha) 10,257,429 96,475,685 316,285,800 63,045,790Planted forest (2005) (ha) 26133.25 Agriculture and similar land uses (2005) 180771.7 141179.6 Bare land (2005) (ha) 119386 Residence (2005) (ha) 10448.22 Population Density 2005 (km2) 404 20.46 20.53 GDP (current prices) 228406.9 Mining (2005) (ha) 1520.80 Palm oil (2005) (ha) 33445.16 Commercial plantation (ha) 37812.31 Timber Concession (ha) 131610.2 Soy bean (ha) 16427.64 GDP cattle ranching (2000) 14294.23 Concession area (ha) 151130Protected Area (ha) 390798.7 Number of Observation 64 226 719 63

Step 2: Brazil

Predict deforestation rates for legal Amazon2005- 2009

Category Regression coefficientHistorical deforestation (2000-2005)

0.628***(0.02)

Forest stock (share of land) 0.009***(0.00)

0.026***(0.00)

Soybean plantation 0.026(0.02)

0.089***(0.03)

Rubber plantation -0.043***(0.01)

0.044*(0.02)

Population density 0.036(0.03)

0.104**(0.05)

Agriculture GDP (%) 0.161***(0.03)

0.480***(0.05)

Per Capita GDP -0.313***(0.07)

0.076(0.11)

GDP based on current price (2005)

-0.112***(0.02)

-0.405***(0.04)

R2 0.914 0.767N 719 719

***p<0.01, **P<0.05, *p < 0.1

Step 2: Vietnam

Predict deforestation rates 2005- 2010

Category Regression coefficient

Historical deforestation (2000 - 2005)

0.520***(0.12)

Forest stock (share of land)

0.029(0.02)

0.711***(0.02)

Planted forest area -0.102(0.12)

-0.002(0.13)

Agricultural land 0.284***(0.07)

0.320***(0.08)

Settlement -0.087(0.08)

-0.056(0.09)

Total bare land 0.210**(0.12)

0.401***(0.12)

R2 0.961 0.947N 64 64

***p<0.01, **P<0.05, *p < 0.1

Step 2: Peru

Predict deforestation rates 2005- 2010

***p<0.01, **P<0.05, *p < 0.1

Category Regression coefficientHistorical deforestation (2000-2005)

.955***(0.03)

Forest stock (share of land)

.001(0.00)

0.069***(0.01)

Forest concession -0.017(0.02)

0.033(0.08)

Protected area 0.054***(0.02)

0.274***(0.06)

Population density 1.378***(0.44)

8.789***(1.48)

R2 0.994 0.907N 63 63

Projection of deforestation 2010 - 2015

Countries Deforestation Forecast 2010-2015 (thousand ha)M1 M2 M3 M4 M5 (BAU1) M6 (BAU 2)

Vietnam 547 (2.1)

654(0.35)

661(1.86)

514(0.54)

518 469

Indonesia 924(2.19)

888(1.67)

605(1.98)

755(0.49)

769 623

Brazil 285 (0.74)

339 (0.66)

410 (0.93)

1,318 (0.29)

939

2,006

Peru 347(1.11)

295(0.99)

279(6.91)

302(0.29)

163 123

Forest carbon emissions equation

Carbon biomass estimation approach

Field data (e.g. National forest inventory)

Direct remote sensing measurement

– Empirical models where RS data is calibrate to field estimates (Baccini et al. 2004, 2008, Saatchi et al. 2007, Blackard et al. 2008)

Stratify and Multiply (SM) method

– Assign an average biomass value to land cover/vegetation type map (Asner et al. 2010)

Combine and Assign approach

– Extension of SM, GIS and multi-layers information (Gibbs et al. 2007)

Ecological Models approach

– RS data to parameterize the biomass model (Hurtt et al. 2004)

Available carbon stocks maps and its uncertainty

Source: Avitabile, et.al. University of Wageningen (in preparation)

Implications to national carbon emissions estimate (Indonesia)

SourceCumulative Emission from

LUCF 2000 -2009(in Gt CO2e)*

Methods Remarks

FAOStat 3.140 FRA country report(EF = 138 ton C/ha) Net forest conversion

MoE - Second National Communication to UNFCCC 7.443 IPCC Guidelines 2006 Net forest conversion

Winrock International (Harris, 2012) 3.468 Carbon Bookkeping model

(RS + Field) Gross deforestation

MOF (official) 1.760 Approach 1 + NFI(Tier 1 or 2)

Net forest conversion (peat?) - carbon emissions potential

MOF + Saatchi (CIFOR) 1.811 Approach 1 + Global EF(Tier 1 or 2) Net forest conversion

* does not include peat emissions and peat fire

Lesson learnt from CIFOR Global Comparative Study on REDD+ on setting up FRELs/FRLs

Countries should start as soon as possible for estimating their national REL/RL based on the best available data (tier 1 estimate)– If we have less data the more we depend on those data

Countries should invest for collecting national level data (i.e. time series forest cover change and local emissions factor data)

Comparison of global forest cover loss and national prediction (Brazil)

Lesson learnt from CIFOR Global Comparative Study on REDD+ on setting up FRELs/FRLs

Countries should start as soon as possible for estimating their national REL/RL based on the best available data (tier 1 estimate)– If we have less data the more we depend on those data

Countries should invest for collecting national datasets (i.e. time series forest cover change and local emissions factor data)

Estimate on carbon stocks change at national/sub-national/project scale should comply with the convention (i.e. IPCC guidelines)

Indicative national REL estimate to 2020 (Indonesia)

Source: Indonesian REDD+ Agency, MRV working group

Lesson learnt from CIFOR Global Comparative Study on REDD+ on setting up FRELs/FRLs

Participation of local people is important to significantly reduce risks of REDD+ and to improve credibility of national climate change mitigation programs

Source: Indonesian REDD+ Agency

Impacts to REDD+ incentives

Emissions reductions that can claim for REDD+ financial incentives should be additional of those resulted from national climate change mitigation program (i.e. based on self financed/national efforts)– Differentiate between REL as business as usual (BAU) and crediting baseline

(or financial incentive benchmark)

Source: redd-net.org

Impacts to REDD+ incentives

Emissions reduction that can claim REDD+ incentives should be additional of those of national climate change mitigation program (based on self financed/national efforts)

Carbon emissions reduction should be estimated following Tier 2 approach

Trade off between climate change mitigation and national economic development sectors is the main challenge to engage in phase 3 of REDD+ implementation (i.e. results-based payment at national scale)– Phase 1 and 2 funding of REDD+ project established at

subnational/project levels can be financed through non-market mechanism (e.g. international donors, project based, etc)

– Phase 3 REDD+ project should be ideally financed through market based mechanism