Methodology

Carbon Stock Estimation and Economic Valuation Framework

Authors:

1. Fairuz Akmal Pradana (Methodology)

2. Aulia Nur Fajriyah (Visualization)

1. Forest Carbon Stock Estimation

1.1 IPCC Framework

This research adopts the four carbon pools approach according to the 2019 Refinement to the 2006 IPCC Guidelines for National Greenhouse Gas Inventories (IPCC, 2019). Total forest carbon stock is calculated using the formula:

C_total = C_above + C_below + C_soil + C_dead

• C_above = above-ground biomass carbon (Mg C/ha)

• C_below = below-ground biomass/root carbon (Mg C/ha)

• C_soil = soil organic carbon 0-30 cm (Mg C/ha)

• C_dead = dead organic matter/litter carbon (Mg C/ha)

1.2 Remote Sensing and Deep Learning Integration

A. Land Cover Detection with U-Net AI Model

Land cover classification is performed using Deep Learning with U-Net architecture (Ronneberger et al., 2015). U-Net was chosen for its excellence in semantic segmentation, capable of maintaining spatial location information while capturing feature context through contracting (encoder) and expanding (decoder) pathways.

The model is trained using Sentinel-2 multispectral imagery to separate land cover classes (Forest, Non-Forest, Shrubland, etc.). U-Net segmentation results are used as the spatial basis for determining soil carbon (C_soil) and dead organic matter (C_dead) parameters according to land cover type.

B. Biomass Estimation from NDVI (Sentinel-2)

Normalized Difference Vegetation Index (NDVI) is calculated from Sentinel-2 Surface Reflectance using the standard formula (Tucker, 1979):

NDVI = (ρ_NIR - ρ_Red) / (ρ_NIR + ρ_Red)

Above-ground biomass (AGB) is estimated using an exponential model calibrated for Sumatran Dipterocarp forests (Basuki et al., 2009):

AGB_Sentinel = exp(2.13 + 2.55 × NDVI)
R² = 0.72, RMSE = 45.3 Mg/ha

C. Integration with WHRC Reference Data

To reduce uncertainty, Sentinel-2 estimates are combined with the Woods Hole Research Center (WHRC) pantropical biomass dataset at 500m resolution (Baccini et al., 2012) using weighted averaging:

AGB_combined = 0.5 × AGB_Sentinel + 0.5 × AGB_WHRC

D. Biomass to Carbon Conversion and Other Pool Estimation

Biomass is converted to carbon using a carbon fraction of 0.47 (IPCC, 2019):

C_above = AGB_combined × 0.47

Below-ground biomass carbon is estimated using a root-to-shoot ratio of 0.25 for tropical forests (IPCC, 2019, Table 4.4):

C_below = C_above × 0.25

Soil carbon (C_soil) and dead organic matter (C_dead) are obtained from IPCC 2019 Refinement carbon density tables for Tropical Asia region (IPCC, 2019, Tables 4.7, 4.9, 4.12) mapped based on U-Net land cover classification.

Table 1. Carbon Density per Land Cover Class (IPCC, 2019)

Land Cover Class	C_above	C_below	C_soil	C_dead	Total
Forest (Tree Cover)	180	45	90	15	330
Shrubland	40	10	75	5	130
Grassland	15	7.5	80	2.5	105
Agricultural Land	5	2.5	65	0	72.5
Mangrove	150	37.5	100	12.5	300
Wetland	30	15	100	10	155

Note: All values in Mg C/ha (megagram carbon per hectare)

1.3 Data and Analysis Period

Temporal analysis is conducted for the period 2020-2025 using:

• Land Cover: Generated from U-Net model using training data referenced from ESA WorldCover v100/v200 (Zanaga et al., 2021) and Google Dynamic World (Brown et al., 2022)
• Biomass: Sentinel-2 SR Harmonized (10m resolution, cloud cover <20%) and WHRC Pantropical Biomass (500m resolution) (Baccini et al., 2012)

Area of Interest (AOI) covers Aceh, North Sumatra, and West Sumatra provinces with a total area of ~173,000 km². Analysis is performed using Google Earth Engine platform for cloud-based computation (Gorelick et al., 2017).

2. Carbon Economic Valuation

2.1 Carbon to CO₂ Equivalent Conversion

Carbon stock is converted to CO₂ equivalent using the standard IPCC stoichiometric conversion factor (IPCC, 2006, Volume 1, Chapter 3):

CO₂e = C_stock × 3.67

The factor 3.67 is derived from the molecular mass ratio of CO₂ (44 g/mol) to atomic mass of C (12 g/mol). This conversion is necessary because carbon trading mechanisms and UNFCCC reporting use CO₂ equivalent units, not elemental carbon (UNFCCC, 2015).

2.2 Economic Value Calculation

Economic value of carbon per hectare is calculated using the formula:

USD_ha = CO₂e × P_carbon

Where P_carbon is the carbon price in USD per tonne CO₂e. This research uses four price scenarios based on global carbon market conditions (World Bank, 2023):

Scenario	Price (USD/tonne CO₂e)	Basis
Conservative	6.34	Average REDD+ projects Asia-Pacific 2020-2022
Moderate	15.00	Global voluntary carbon market median 2023
High	27.00	Projects with high co-benefits
Compliance	50.00	Average compliance market (EU ETS)

The conservative scenario (USD 6.34) is chosen as baseline as it is representative for Indonesian REDD+ projects and consistent with historical prices of the Indonesia-Norway REDD+ program (Government of Indonesia, 2020; Ecosystem Marketplace, 2023).

2.3 Temporal Change and Regional Total Analysis

Carbon emission/sequestration change (2020-2025):

ΔCO₂e = CO₂e₂₀₂₅ - CO₂e₂₀₂₀

Negative values indicate emissions (deforestation/degradation), positive values indicate sequestration (reforestation/forest growth).

Economic value change:

ΔUSD = ΔCO₂e × P_carbon

Total regional economic value:

USD_total = Σ(CO₂e_i × A_i) × P_carbon

Where A_i is the area of pixel i in hectares (0.09 ha for 30m resolution).

2.4 Computational Implementation

The valuation algorithm is implemented in Python with GeoJSON input data extracted from Google Earth Engine. For each feature/pixel:

1. Conversion: CO₂e = C_stock × 3.67 for each year (2020-2025)
2. Valuation: USD_ha = CO₂e × P_carbon according to scenario
3. Temporal change: ΔCO₂e and ΔUSD for period 2020-2025
4. Aggregation: Total regional CO₂e and USD, average per hectare

Output includes additional properties: {year}_CO2e, {year}_USD_per_ha, CO2e_change_2020_2025, USD_change_2020_2025, and metadata (carbon price, conversion factor, calculation date, price scenario).

2.5 Uncertainty and Applications

Sources of Uncertainty

• Carbon stock estimation: ±25-35% (U-Net classification error, NDVI model, spatial variability) (Mitchard et al., 2013)
• Carbon price: ±50-100% (market volatility, project type differences) (Ecosystem Marketplace, 2023)
• Total uncertainty: ±60-110%

Valuation Applications

1. Baseline for REDD+ projects according to UNFCCC framework (UNFCCC, 2015)
2. Contribution to Indonesia's Nationally Determined Contributions (NDC) targeting 29-41% emission reduction by 2030 (Republic of Indonesia, 2016)
3. Identification of priority conservation areas based on high carbon value
4. Cost-benefit analysis of conservation vs. land conversion
5. Communication of ecosystem service economic value to stakeholders

Note: This valuation is suitable for initial screening and strategic planning, but requires field verification for actual carbon credit transactions according to Verified Carbon Standard (Verra, 2023).

References

Baccini, A., et al. (2012). Estimated carbon dioxide emissions from tropical deforestation improved by carbon-density maps. Nature Climate Change, 2(3), 182-185.

Basuki, T. M., et al. (2009). Allometric equations for estimating the above-ground biomass in tropical lowland Dipterocarp forests. Forest Ecology and Management, 257(8), 1684-1694.

Brown, C. F., et al. (2022). Dynamic World, near real-time global 10 m land use land cover mapping. Scientific Data, 9(1), 251.

Ecosystem Marketplace. (2023). State of the voluntary carbon markets 2023. Forest Trends.

Gorelick, N., et al. (2017). Google Earth Engine: Planetary-scale geospatial analysis for everyone. Remote Sensing of Environment, 202, 18-27.

Government of Indonesia. (2020). Indonesia's REDD+ results and the Norway partnership. Ministry of Environment and Forestry.

IPCC. (2006). 2006 IPCC Guidelines for National Greenhouse Gas Inventories, Volume 4: Agriculture, Forestry and Other Land Use.

IPCC. (2019). 2019 Refinement to the 2006 IPCC Guidelines for National Greenhouse Gas Inventories, Volume 4: Agriculture, Forestry and Other Land Use.

Mitchard, E. T., et al. (2013). Uncertainty in the spatial distribution of tropical forest biomass: A comparison of pan-tropical maps. Carbon Balance and Management, 8(1), 10.

Republik Indonesia. (2016). First nationally determined contribution Republic of Indonesia. Submitted to UNFCCC.

Ronneberger, O., Fischer, P., & Brox, T. (2015). U-Net: Convolutional Networks for Biomedical Image Segmentation. Medical Image Computing and Computer-Assisted Intervention (MICCAI), 9351, 234–241.

Tucker, C. J. (1979). Red and photographic infrared linear combinations for monitoring vegetation. Remote Sensing of Environment, 8(2), 127-150.

UNFCCC. (2015). Adoption of the Paris Agreement. UNFCCC Decision 1/CP.21.

Verra. (2023). VCS Standard, Version 4.5. Verified Carbon Standard.

World Bank. (2023). State and trends of carbon pricing 2023. World Bank Group.

Zanaga, D., et al. (2021). ESA WorldCover 10 m 2020 v100. European Space Agency.