WildFire Emissions

Author

Published

March 30, 2026

Preface

This resource provides comprehensive methodological guidance for deriving IPCC Tier 1 Forest Fire Activity Data, with particular emphasis on burned area estimation following FAOSTAT and IPCC 2019 Refinement protocols. The workflow integrates satellite-based fire detection, land cover classification, fuel consumption modeling, and emission factor application to produce country-level greenhouse gas emission estimates from biomass burning. The eBook uses an example AOI derived from the Republic of Ireland national border.

The methodology presented here addresses key challenges in fire emissions accounting: inter-annual variability in natural disturbances, the managed land proxy approach for anthropogenic attribution, uncertainty quantification across multiple data sources, and the balance between immediate emissions and subsequent carbon removals. Each chapter builds systematically from data acquisition through quality control, providing transparent, reproducible workflows suitable for national greenhouse gas inventory reporting and validation.

Chapter 1: Burn Area

Extract MODIS MCD64A1 data, apply quality filters, aggregate temporally and spatially, handle missing data, and validate against alternative fire products.

Chapter 2: Fuel Stratification

Classify burned pixels into Forest (Humid Tropical, Other), Savanna, and Organic Soil categories using MODIS land cover and auxiliary datasets.

Chapter 3: Fuel Consumption

Assign IPCC default fuel consumption values based on vegetation type and climate zone stratification, calculate total burned biomass.

Chapter 4: Emission Factors

Apply IPCC emission factors, compute CH₄, N₂O, and CO₂ emissions, handle CO₂ reporting exclusions for non-stand-replacing fires.

Chapter 5: QA/QC & Audit-Ready SOPs (Pending)

Cross-validation with UNFCCC data, uncertainty propagation, sensitivity analysis, and documentation for NGHGI reporting.

IPCC Tier 1 Fire Emissions

The IPCC Tier 1 approach for estimating greenhouse gas emissions from fires provides a standardized, globally applicable framework based on readily available data and default emission factors. This methodology applies Equation 2.27 from the 2019 IPCC Refinements to Guidelines for National Greenhouse Gas Inventories (IPCC, 2019):

\[ L_{fire} = A \times M_B \times C_f \times G_{ef} \times 10^{-3} \]

Where:

L_fire = greenhouse gas emissions from fire (tonnes of CH₄, N₂O, or CO₂)
A = area burnt (ha)
M_B = mass of fuel available for combustion (tonnes ha⁻¹)
C_f = combustion factor (dimensionless, proportion of fuel consumed)
G_ef = emission factor (g kg⁻¹ dry matter burnt)

Vegetation-Specific Applications

Forest Fires:

Only non-CO₂ emissions (CH₄, N₂O) are reported
CO₂ emissions excluded as they are captured in carbon stock change calculations
Applies to fires in Evergreen, Deciduous, and Mixed Forest land cover types

Savanna Fires:

Non-CO₂ emissions only (CH₄, N₂O)
Assumes CO₂ balance through annual vegetation regrowth
Covers Grassland, Savanna, Woody Savanna, and Shrubland types

Organic Soil Fires:

Reports both CO₂ and CH₄ emissions
Uses IPCC Wetlands Supplement parameters (IPCC, 2014)
Currently validated for Southeast Asia only

Managed Land Proxy Approach

The methodology adopts the Managed Land Proxy (MLP) as the primary framework for attributing anthropogenic emissions. Under MLP:

All emissions and removals on managed lands approximate anthropogenic effects
Natural disturbances (e.g., wildfires) are included in total managed land fluxes
Optional disaggregation separates anthropogenic activities from natural disturbance impacts
Expectation that emissions from natural disturbances balance with subsequent removals over time

The 2019 IPCC Refinement provides enhanced guidance for addressing inter-annual variability (IAV) caused by:

Natural disturbances (wildfires, insects, windthrow, ice storms)
Climate variability affecting photosynthesis and respiration
Variation in human activities (harvesting, land-use change)

Countries experiencing high IAV from natural disturbances may optionally disaggregate these fluxes to better identify trends in anthropogenic emissions and evaluate mitigation actions.

FAOSTAT Fires Domain

Database Overview

The FAOSTAT Emissions from Fires domain provides spatially- and temporally stratified estimates of CH₄, N₂O, and CO₂ emissions from biomass burning across all countries and territories. The database:

Covers the complete time series 1990–2024
Updates annually with new satellite observations
Disseminates activity data and emission estimates (kt CH₄, N₂O, CO₂)
Provides regional aggregations for Annex I / non-Annex I country groups
Integrates country-reported data from UNFCCC submissions for validation

Methodological Foundation

FAOSTAT implements IPCC Tier 1 methodology using:

Pixel-level processing in Google Earth Engine (GEE)
Country-level aggregation via Global Administrative Unit Layers (EUROSTAT, n.d.)
Climate-dependent fuel consumption values stratified by FAO-FRA Ecological Zones
Quality-filtered satellite observations (uncertainty < 20%)

The database serves as an official IPCC-recommended tool for National Greenhouse Gas Inventory (NGHGI) QA/QC processes and validation of activity data and emission estimates (IPCC 2019).

Three Fire Categories

1. Forest Fires
Emissions from fires in Forest Land, disaggregated into:

Humid Tropical Forest: Forest pixels overlaid with FAO-FRA tropical zones
Other Forest: All remaining forest types (boreal, temperate, extra-tropical)

Land cover identification aggregates MODIS IGBP classes:

Evergreen Needleleaf Forest
Evergreen Broadleaf Forest
Deciduous Needleleaf Forest
Deciduous Broadleaf Forest
Mixed Forests

2. Savanna Fires
Emissions from Grassland and woody vegetation:

Grassland
Savanna
Woody Savanna
Open Shrubland
Closed Shrubland

3. Fires in Organic Soils
Emissions from peat fires identified by overlaying burned area with Histosols class from the Harmonized World Soil Database. Due to high uncertainty and limited validation outside Southeast Asia, emissions are set to zero for all countries except those in Southeast Asia (Indonesia, Malaysia, Brunei Darussalam). Complete global estimates are provided in supplementary files for reference and research purposes.

Temporal Gap-Filling

Primary Satellite Era (2001–2024): Direct observations from MODIS MCD64A1 Collection 6.1 provide monthly burned area at 500m resolution.

Historical Period (1990–2000): Values backfilled using two approaches:

1996–2000: Earlier FAOSTAT estimates (Rossi et al., 2016)
1990–1995: Averaged values from reconstructed time series (1996–2002 mean)

This approach ensures complete temporal coverage required for UNFCCC reporting while acknowledging higher uncertainty in pre-MODIS years.

QA/QC and Validation

National GHG Inventories

Primary Use: The IPCC explicitly recommends FAOSTAT as a tool for NGHGI QA/QC processes (IPCC 2019). Countries can:

Cross-validate activity data: Compare satellite-derived burned areas against national fire databases
Evaluate emission factors: Test sensitivity to country-specific vs. default IPCC parameters
Identify data gaps: Detect missing or inconsistent reporting in national systems
Assess trends: Evaluate whether reported trends align with satellite observations

Comparison Frameworks

Top-Down vs. Bottom-Up Reconciliation:

FAOSTAT (satellite-based, top-down) <> NGHGI (ground-based, bottom-up)

Systematic differences may indicate:

Detection limitations in satellite products (e.g., small fires, smoldering)
Incomplete ground reporting
Different definitions of “managed land”
Biomass consumption parameter uncertainties

Multi-Source Fire Validation:

Fire Detection: MODIS active fire products (MCD14ML, MYD14) vs. burned area (MCD64A1)
Land Cover: MODIS MCD12Q1 vs. national forest inventories
Emission Factors: IPCC defaults ~~vs. regional measurements from literature~~¹

Tier Advancement Pathways

Countries may use FAOSTAT as a stepping stone to higher-tier methods:

Tier 1 => Tier 2:

Replace default emission factors with country-specific values
Refine fuel consumption parameters using field measurements
Improve combustion factor estimates via burn severity mapping

Tier 2 => Tier 3:

Integrate process-based models (e.g., FOFEM, CONSUME)
Develop national fire monitoring systems
Implement measurement-based carbon stock change assessments

Mitigation Action Tracking

FAOSTAT time series enable evaluation of:

Prescribed burning programs
Fire suppression effectiveness
Land management policy impacts
Climate change effects on fire regimes

Countries can quantify whether trends in anthropogenic fire emissions diverge from natural disturbance variability, supporting Nationally Determined Contribution (NDC) reporting and verification.

Uncertainty Considerations

Sources of Uncertainty

1. Activity Data Uncertainty

Satellite detection limitations:

Small fires (< 500m): May be missed by MODIS 500m pixels
Canopy-obscured fires: Difficult to detect in dense forests
Smoldering fires: Peat fires may not produce strong thermal signatures
Cloud contamination: Persistent cloud cover reduces observation frequency
Rapid vegetation regrowth: Short time lag between fire and regrowth obscures burn scars

MCD64A1 applies quality filters:

Only pixels with uncertainty < 20% are included
Commission/omission errors vary by biome
Global burned area may be underestimated by 20-30% (Giglio et al., 2010)

2. Land Classification Uncertainty

MODIS MCD12Q1 accuracy:

Overall accuracy: ~75% globally
Confusion between woody savannas and forests
Temporal lag in land-use change detection
Mixed pixels in heterogeneous landscapes

Impact on emissions: Misclassification affects fuel consumption values (Table 2.4 ranges: 2.6 to 163.6 t ha⁻¹).

3. Fuel Consumption Uncertainty

IPCC default values (Table 2.4) reflect:

High natural variability: Standard errors 30-50% of mean
Limited data for some vegetation types
Climate dependency: Values vary by FAO-FRA Ecological Zones, but within-zone heterogeneity remains

Example uncertainties (Mean ± SE):

Primary Tropical Forest: 119.6 ± (range: 83.9 to 163.6) t ha⁻¹
Savanna Woodland: 2.6 ± 0.1 (early dry season) to 4.6 ± 1.5 (late dry season) t ha⁻¹

4. Combustion Factor Uncertainty

Proportion of fuel consumed (Table 2.6):

Influenced by fire intensity, fuel moisture, weather conditions
Variability within vegetation types: SD often 0.1-0.2 (for factors ranging 0.2-0.9)
Incomplete combustion of coarse woody debris vs. near-complete combustion of grasses

5. Emission Factor Uncertainty

IPCC defaults (Table 2.5) based on (Andreae & Merlet, 2001):

Values represent global averages across fire types
Standard deviations: 20-60% of mean values
Influenced by combustion efficiency (flaming vs. smoldering)

Example (Savanna fires):

CO₂: 1613 ± 95 g kg⁻¹
CH₄: 2.3 ± 0.9 g kg⁻¹
N₂O: 0.21 ± 0.10 g kg⁻¹

Uncertainty Aggregation

For IPCC Tier 1, cumulative uncertainty typically:

Activity data: ±30% for burned area
Emission estimates: ±50% for CH₄ and N₂O, ±30-40% for CO₂

Organic soil fires have particularly high uncertainty:

Remote sensing detection challenges
Limited field validation outside Southeast Asia
Difficulty distinguishing surface fires from subsurface peat combustion
Unknown proportion of fires on drained vs. undrained peatlands

Uncertainty Reduction

Country-Specific Improvements:

Integrate national fire databases with satellite observations
Conduct field campaigns to measure fuel loads and combustion factors
Develop regional emission factors from smoke sampling
Implement Tier 2/3 methods in high-emission areas

Methodological Enhancements:

Multi-sensor fusion (MODIS + VIIRS + Sentinel)
Sub-pixel burned area algorithms
Burn severity mapping for combustion factors
Seasonal stratification of fuel moisture

Managing Inter-Annual Variability

High IAV from natural disturbances complicates trend analysis:

Wildfires can cause 2-3 orders of magnitude annual variation
Obscures anthropogenic emission trends
Challenges mitigation policy evaluation

Optional Disaggregation Approach (IPCC, 2019)²:

Countries may separate:

Natural disturbance component: Emissions beyond normal variability or 95% CI
Anthropogenic component: Remaining managed land fluxes

This disaggregation:

Improves clarity of human activity impacts
Maintains MLP total for official reporting
Requires documentation of methods and thresholds
Assumes eventual balance between disturbance emissions and regrowth

Data Sources

MCD64A1.061: Burn Areas

Product Overview: The MCD64A1 Version 6.1 Burned Area product provides monthly, global gridded burned area at 500m spatial resolution, derived from combined Terra and Aqua MODIS observations.

Key Features:

Temporal Resolution: Monthly composites, 2001–present
Spatial Resolution: 500m nominal pixel size
Algorithm: Decision tree classifier using surface reflectance time series
Output: Burn date encoded as ordinal day of year, with quality indicators

Data Layers:

Burn Date (day of year when burn detected)
Burn Data Uncertainty (percentage uncertainty, 0-100%)
Quality Assurance flags

FAOSTAT Implementation:

Quality filter applied: Only pixels with uncertainty < 20% are retained
Aggregation: Pixel-level burn dates accumulated to annual country totals
GEE Asset: MODIS/061/MCD64A1

Access:

NASA LP DAAC: https://lpdaac.usgs.gov/products/mcd64a1v061/
Google Earth Engine: ee.ImageCollection("MODIS/061/MCD64A1")
User Guide: (Giglio et al., 2024)

Limitations:

May miss small fires (< 500m²)
Reduced accuracy in densely forested areas
Cloud obscuration limits temporal coverage in some regions
Detection lag in slow-spreading fires

MCD12Q1.061: IGBP Land Cover

Product Overview: MCD12Q1 provides annual global land cover classification at 500m resolution using supervised classification of MODIS reflectance data.

Classification Scheme: FAOSTAT uses Land Cover Type 1 (IGBP 17-class legend):

Forest Types (used for Forest Fire category):

Class 1: Evergreen Needleleaf Forests
Class 2: Evergreen Broadleaf Forests
Class 3: Deciduous Needleleaf Forests
Class 4: Deciduous Broadleaf Forests
Class 5: Mixed Forests

Grassland/Savanna Types (used for Savanna Fire category):

Class 6: Closed Shrublands
Class 7: Open Shrublands - Class 8: Woody Savannas
Class 9: Savannas
Class 10: Grasslands
Other Classes: Croplands, Urban, Wetlands, etc. (Classes 11-17)

FAOSTAT Implementation:

Annual land cover maps matched to fire detection year
Overlay burned pixels with contemporaneous land cover
Exclude built-up, barren, and water bodies
GEE Asset: MODIS/061/MCD12Q1

Access:

NASA LP DAAC: https://lpdaac.usgs.gov/products/mcd12q1v061/
Google Earth Engine: ee.ImageCollection("MODIS/061/MCD12Q1")
User Guide: (Sulla-Menashe et al., 2018)

Accuracy:

Overall accuracy: ~75% globally
Higher accuracy in homogeneous landscapes
sConfusion common between woody savannas and open forests

FAO Global Ecological Zones

Dataset Overview:
The Global Ecological Zones (GEZ) classification provides a climate-vegetation framework for stratifying fuel consumption parameters.

Classification System Combines:

Climate domains (Polar, Boreal, Temperate, Subtropical, Tropical)
Ecological zones (e.g., Tropical moist forest, Tropical dry forest, Subtropical steppe)

FAOSTAT Implementation:

Overlay burned forest pixels with GEZ to distinguish:
Humid Tropical Forest: Tropical moist ecological zones
Other Forest: All remaining forest types
Assign climate-dependent fuel consumption values from IPCC Table 2.4

Access:

FAO Forestry: http://www.fao.org/forest-resources-assessment/
Document: FAO Global Ecological Zones for FAO forest reporting 2010 Update (FAO, 2012)

Application:

Determines fuel biomass consumption parameters
Stratifies emission factors where climate-sensitive
Links to FAO Forest Resources Assessment (FRA) reporting categories

IPCC Soil Zones Layer

Dataset Overview:
The Harmonized World Soil Database (HWSD v1.2) provides global soil type classifications at 30 arc-second (~1 km) resolution.

Histosols Class:
Organic soils characterized by:

High organic matter content (> 20-30% by weight)
Formed in waterlogged conditions
Includes peatlands, bogs, fens

FAOSTAT Implementation:

Burned area pixels overlaid with Histosols spatial layer
Fires in organic soils identified independently of land cover (except excluding built-up, barren, water)
Different fuel consumption and emission factors applied (IPCC Wetlands Supplement 2014)

Access:

FAO Soils Portal: http://www.fao.org/soils-portal/
IIASA: https://iiasa.ac.at/models-and-data/hwsd

Critical Limitation:

FAOSTAT organic soil fire emissions set to zero outside Southeast Asia due to:
High uncertainty in peat fire detection
Limited validation data - Difficulty distinguishing drained vs. undrained peatlands
Uncertainty in human influence on fire occurrence

Supplementary file AllFires_OrganicSoils.csv contains global estimates for research purposes.

IPCC Climate Zones Layer

Dataset Overview:
IPCC Climate Zones classify global climate into categories used for stratifying biomass growth rates, decomposition rates, and fire parameters.

Classification:

Polar
Boreal (Cold)
Temperate (Warm temperate, Cool temperate)
Tropical (Tropical wet, Tropical moist, Tropical dry, Tropical montane)

Source:

Joint Research Centre (JRC) Climate Zones layer
Based on IPCC climatic zone definitions (IPCC, 2006)

FAOSTAT Implementation:

Savanna fire fuel consumption values vary by climate zone (Table 2.4)
Organic soil emission factors stratified by climate (IPCC, 2013)
Links IPCC methodology to geographic reality

Access:

JRC European Soil Data Centre (ESDAC): https://esdac.jrc.ec.europa.eu/
Reference: JRC (2010) Climatic Zone layer

Google Earth Engine Setup

Platform Overview:
Google Earth Engine is a cloud-based geospatial analysis platform providing:

Petabyte-scale satellite imagery archives
Planetary-scale computational infrastructure
JavaScript and Python APIs
Web-based Code Editor

FAOSTAT Workflow Architecture:

Data Ingestion: All datasets (MCD64A1, MCD12Q1, GAUL, etc.) available as pre-processed GEE assets
Pixel-Level Processing: Fire algorithm applied to all global pixels
Zonal Statistics: Country-level aggregation via GAUL boundaries
Export: Results exported as CSV tables for FAOSTAT database

Getting Started:

Account Registration:
- Sign up at https://earthengine.google.com/signup/
- Academic/NGO accounts typically approved within 1-2 days
Code Editor Access:
- Navigate to https://code.earthengine.google.com/
- JavaScript environment for interactive analysis

Python API Setup (optional):

pip install earthengine-api
earthengine authenticate

Key GEE Concepts:
- ImageCollection: Time series of raster images
- Image: Single raster dataset
- FeatureCollection: Vector dataset (e.g., country boundaries)
- Reducer: Aggregation functions (sum, mean, etc.)

FAOSTAT-Relevant GEE Assets:³

// Burned Area
var burnedArea = ee.ImageCollection("MODIS/061/MCD64A1");

// Land Cover
var landCover = ee.ImageCollection("MODIS/061/MCD12Q1");

// Country Boundaries
var countries = ee.FeatureCollection("FAO/GAUL/2015/level0");

Andreae, M. O., & Merlet, P. (2001). Emission of trace gases and aerosols from biomass burning. Global Biogeochemical Cycles, 15(4), 955–966.

EUROSTAT. (n.d.). Geographic information system of the european commission.

FAO, G. (2012). Global ecological zones for FAO forest reporting: 2010 update. FAO: Rome, Italy.

Giglio, L., Hall, J. V., Humber, M., Argueta, F., Boschetti, L., & Roy, D. (2024). Collection 2 VIIRS burned area product user’s guide version 1.1. NASA EOSDIS Land Processes Distributed Active Archive Center. DOI: Https://Doi. Org/10.5067/VIIRS/VNP64A1, 2.

Giglio, L., Randerson, J., Van der Werf, G., Kasibhatla, P., Collatz, G., Morton, D., & DeFries, R. (2010). Assessing variability and long-term trends in burned area by merging multiple satellite fire products. Biogeosciences (Online), 7(3), 1171–1186.

IPCC. (2006). Chapter 3: Consistent representation of lands. In 2006 IPCC guidelines for national greenhouse gas inventories 3.1 (Agriculture, Forestry and Other Land Use, Vol. 4). Intergovernmental Panel on Climate Change. https://www.ipcc-nggip.iges.or.jp/public/2006gl/pdf/4_Volume4/V4_03_Ch3_Representation.pdf

IPCC. (2013). Revised supplementary methods and good practice guidance arising from the kyoto protocol.

IPCC. (2014). 2013 supplement to the 2006 IPCC guidelines for national greenhouse gas inventories: Wetlands iii. Intergovernmental Panel on Climate Change. https://www.ipcc-nggip.iges.or.jp/public/wetlands/pdf/Wetlands_Supplement_Entire_Report.pdf

IPCC. (2019). Chapter 2: Generic methodologies applicable to multiple land-use categories. In 2019 refinement to the 2006 IPCC guidelines for national greenhouse gas inventories (Agriculture, Forestry and Other Land Use, Vol. 4). Intergovernmental Panel on Climate Change. https://www.ipcc-nggip.iges.or.jp/public/2019rf/pdf/4_Volume4/19R_V4_Ch02_Generic%20Methods.pdf

Rossi, S., Tubiello, F. N., Prosperi, P., Salvatore, M., Jacobs, H., Biancalani, R., House, J. I., & Boschetti, L. (2016). FAOSTAT estimates of greenhouse gas emissions from biomass and peat fires. Climatic Change, 135(3), 699–711.

Sulla-Menashe, D., Friedl, M. A., et al. (2018). User guide to collection 6 MODIS land cover (MCD12Q1 and MCD12C1) product. Usgs: Reston, Va, Usa, 1, 18.

Note to self: 2025 wishlist↩︎
Direct Page Link: https://www.ipcc-nggip.iges.or.jp/public/2019rf/pdf/4_Volume4/19R_V4_Ch02_Generic%20Methods.pdf#page=71 ↩︎
Code provided Python and R, with some cells in Java for use natively in Earth Engine Codespace↩︎

# Preface {.unnumbered} ------------------------------------------------------------------------ ```{r} #| warning: false #| message: false #| echo: false #| comment: NA easypackages::packages( "bslib", "cols4all", "covr", "cowplot", "dendextend", "digest","DiagrammeR","dtwclust", "downlit", "e1071", "exactextractr","elevatr", "FNN", "future", "forestdata", "gdalcubes", "gdalUtilities", "geojsonsf", "geos", "ggplot2", "ggstats", "ggspatial", "ggmap", "ggplotify", "ggpubr", "ggrepel", "giscoR", "hdf5r", "httr", "httr2", "htmltools", "jsonlite", "kohonen", "leaflet.providers", "leafem", "libgeos","luz","lwgeom", "leaflet", "leafgl", "mapedit", "mapview", "maptiles", "methods", "mgcv", "ncdf4", "nnet", "openxlsx", "parallel", "plotly", "randomForest", "rasterVis", "raster", "Rcpp", "RcppArmadillo", "RcppCensSpatial","rayshader", "RcppEigen", "RcppParallel", "RColorBrewer", "reactable", "rgl", "rsconnect","RStoolbox", "rts", "s2", "sf", "scales", "sits","spdep", "stars", "stringr","supercells", "terra", "testthat", "tidyverse", "tidyterra","tools", "tmap", "tmaptools", "terrainr", "xgboost", prompt = F) #mapviewOptions(fgb = FALSE) sf::sf_use_s2(use_s2 = FALSE) ``` ```{r} #| warning: false #| message: false #| echo: false #| comment: NA options(repos = c(CRAN = "https://cloud.r-project.org"), htmltools.dir.version = FALSE, htmltools.preserve.raw = FALSE ) knitr::opts_chunk$set( echo = TRUE, message = FALSE, warning = FALSE, error = FALSE, comment = NA, tidy.opts = list(width.cutoff = 60) ) ``` ```{css, echo=FALSE, class.source = 'foldable'} div.column { display: inline-block; vertical-align: top; width: 50%; } #TOC::before { content: ""; display: block; height:200px; width: 200px; background-image: url('https://raw.githubusercontent.com/seamusrobertmurphy/IPCC-wildfire-emissions/refs/heads/main/01-burn-area/AOI/aoi_burn_aa.png'); background-size: contain; background-position: 50% 50%; padding-top: 80px !important; background-repeat: no-repeat; } ``` This resource provides comprehensive methodological guidance for deriving IPCC Tier 1 Forest Fire Activity Data, with particular emphasis on burned area estimation following FAOSTAT and IPCC 2019 Refinement protocols. The workflow integrates satellite-based fire detection, land cover classification, fuel consumption modeling, and emission factor application to produce country-level greenhouse gas emission estimates from biomass burning. The eBook uses an example AOI derived from the Republic of Ireland national border. The methodology presented here addresses key challenges in fire emissions accounting: inter-annual variability in natural disturbances, the managed land proxy approach for anthropogenic attribution, uncertainty quantification across multiple data sources, and the balance between immediate emissions and subsequent carbon removals. Each chapter builds systematically from data acquisition through quality control, providing transparent, reproducible workflows suitable for national greenhouse gas inventory reporting and validation. #### Chapter 1: Burn Area {.unnumbered} Extract MODIS MCD64A1 data, apply quality filters, aggregate temporally and spatially, handle missing data, and validate against alternative fire products. #### Chapter 2: Fuel Stratification {.unnumbered} Classify burned pixels into Forest (Humid Tropical, Other), Savanna, and Organic Soil categories using MODIS land cover and auxiliary datasets. #### Chapter 3: Fuel Consumption {.unnumbered} Assign IPCC default fuel consumption values based on vegetation type and climate zone stratification, calculate total burned biomass. #### Chapter 4: Emission Factors {.unnumbered} Apply IPCC emission factors, compute CH₄, N₂O, and CO₂ emissions, handle CO₂ reporting exclusions for non-stand-replacing fires. #### Chapter 5: QA/QC & Audit-Ready SOPs (Pending) {.unnumbered} Cross-validation with UNFCCC data, uncertainty propagation, sensitivity analysis, and documentation for NGHGI reporting. ------------------------------------------------------------------------ ## IPCC Tier 1 Fire Emissions The IPCC Tier 1 approach for estimating greenhouse gas emissions from fires provides a standardized, globally applicable framework based on readily available data and default emission factors. This methodology applies Equation 2.27 from the 2019 IPCC Refinements to Guidelines for National Greenhouse Gas Inventories [@ipccChapter2Generic2019]: $$ L_{fire} = A \times M_B \times C_f \times G_{ef} \times 10^{-3} $$ Where: - L~fire~ = greenhouse gas emissions from fire (tonnes of CH₄, N₂O, or CO₂) - A = area burnt (ha) - M~B~ = mass of fuel available for combustion (tonnes ha⁻¹) - C~f~ = combustion factor (dimensionless, proportion of fuel consumed) - G~ef~ = emission factor (g kg⁻¹ dry matter burnt) #### Vegetation-Specific Applications Forest Fires: - Only non-CO₂ emissions (CH₄, N₂O) are reported - CO₂ emissions excluded as they are captured in carbon stock change calculations - Applies to fires in Evergreen, Deciduous, and Mixed Forest land cover types Savanna Fires: - Non-CO₂ emissions only (CH₄, N₂O) - Assumes CO₂ balance through annual vegetation regrowth - Covers Grassland, Savanna, Woody Savanna, and Shrubland types Organic Soil Fires: - Reports both CO₂ and CH₄ emissions - Uses IPCC Wetlands Supplement parameters [@ipcc2013Supplement20062014] - Currently validated for Southeast Asia only #### Managed Land Proxy Approach The methodology adopts the Managed Land Proxy (MLP) as the primary framework for attributing anthropogenic emissions. Under MLP: - All emissions and removals on managed lands approximate anthropogenic effects - Natural disturbances (e.g., wildfires) are included in total managed land fluxes - Optional disaggregation separates anthropogenic activities from natural disturbance impacts - Expectation that emissions from natural disturbances balance with subsequent removals over time The 2019 IPCC Refinement provides enhanced guidance for addressing inter-annual variability (IAV) caused by: 1. Natural disturbances (wildfires, insects, windthrow, ice storms) 2. Climate variability affecting photosynthesis and respiration 3. Variation in human activities (harvesting, land-use change) Countries experiencing high IAV from natural disturbances may optionally disaggregate these fluxes to better identify trends in anthropogenic emissions and evaluate mitigation actions. ------------------------------------------------------------------------ ## FAOSTAT Fires Domain #### Database Overview The FAOSTAT Emissions from Fires domain provides spatially- and temporally stratified estimates of CH₄, N₂O, and CO₂ emissions from biomass burning across all countries and territories. The database: - Covers the complete time series 1990–2024 - Updates annually with new satellite observations - Disseminates activity data and emission estimates (kt CH₄, N₂O, CO₂) - Provides regional aggregations for Annex I / non-Annex I country groups - Integrates country-reported data from UNFCCC submissions for validation #### Methodological Foundation FAOSTAT implements IPCC Tier 1 methodology using: - Pixel-level processing in Google Earth Engine (GEE) - Country-level aggregation via Global Administrative Unit Layers [@eurostatGeographicInformationSystem2025] - Climate-dependent fuel consumption values stratified by FAO-FRA Ecological Zones - Quality-filtered satellite observations (uncertainty \< 20%) The database serves as an official IPCC-recommended tool for National Greenhouse Gas Inventory (NGHGI) QA/QC processes and validation of activity data and emission estimates (IPCC 2019). #### Three Fire Categories 1\. Forest Fires\ Emissions from fires in Forest Land, disaggregated into: - Humid Tropical Forest: Forest pixels overlaid with FAO-FRA tropical zones - Other Forest: All remaining forest types (boreal, temperate, extra-tropical) Land cover identification aggregates MODIS IGBP classes: - Evergreen Needleleaf Forest - Evergreen Broadleaf Forest - Deciduous Needleleaf Forest - Deciduous Broadleaf Forest - Mixed Forests 2\. Savanna Fires\ Emissions from Grassland and woody vegetation: - Grassland - Savanna - Woody Savanna - Open Shrubland - Closed Shrubland 3\. Fires in Organic Soils\ Emissions from peat fires identified by overlaying burned area with Histosols class from the Harmonized World Soil Database. Due to high uncertainty and limited validation outside Southeast Asia, emissions are set to zero for all countries except those in Southeast Asia (Indonesia, Malaysia, Brunei Darussalam). Complete global estimates are provided in supplementary files for reference and research purposes. #### Temporal Gap-Filling Primary Satellite Era (2001–2024): Direct observations from MODIS MCD64A1 Collection 6.1 provide monthly burned area at 500m resolution. Historical Period (1990–2000): Values backfilled using two approaches: - 1996–2000: Earlier FAOSTAT estimates [@rossi2016faostat] - 1990–1995: Averaged values from reconstructed time series (1996–2002 mean) This approach ensures complete temporal coverage required for UNFCCC reporting while acknowledging higher uncertainty in pre-MODIS years. ## QA/QC and Validation #### National GHG Inventories Primary Use: The IPCC explicitly recommends FAOSTAT as a tool for NGHGI QA/QC processes (IPCC 2019). Countries can: 1. Cross-validate activity data: Compare satellite-derived burned areas against national fire databases 2. Evaluate emission factors: Test sensitivity to country-specific vs. default IPCC parameters\ 3. Identify data gaps: Detect missing or inconsistent reporting in national systems 4. Assess trends: Evaluate whether reported trends align with satellite observations #### Comparison Frameworks Top-Down vs. Bottom-Up Reconciliation: FAOSTAT (satellite-based, top-down) \<\> NGHGI (ground-based, bottom-up) Systematic differences may indicate: - Detection limitations in satellite products (e.g., small fires, smoldering) - Incomplete ground reporting - Different definitions of "managed land" - Biomass consumption parameter uncertainties Multi-Source Fire Validation: - Fire Detection: MODIS active fire products (MCD14ML, MYD14) vs. burned area (MCD64A1) - Land Cover: MODIS MCD12Q1 vs. national forest inventories - Emission Factors: IPCC defaults ~~vs. regional measurements from literature~~[^index-1] #### Tier Advancement Pathways Countries may use FAOSTAT as a stepping stone to higher-tier methods: Tier 1 =\> Tier 2: - Replace default emission factors with country-specific values - Refine fuel consumption parameters using field measurements - Improve combustion factor estimates via burn severity mapping Tier 2 =\> Tier 3: - Integrate process-based models (e.g., FOFEM, CONSUME) - Develop national fire monitoring systems - Implement measurement-based carbon stock change assessments #### Mitigation Action Tracking FAOSTAT time series enable evaluation of: - Prescribed burning programs - Fire suppression effectiveness - Land management policy impacts - Climate change effects on fire regimes Countries can quantify whether trends in anthropogenic fire emissions diverge from natural disturbance variability, supporting Nationally Determined Contribution (NDC) reporting and verification. ------------------------------------------------------------------------ ## Uncertainty Considerations #### Sources of Uncertainty ### 1. Activity Data Uncertainty Satellite detection limitations: - Small fires (\< 500m): May be missed by MODIS 500m pixels - Canopy-obscured fires: Difficult to detect in dense forests - Smoldering fires: Peat fires may not produce strong thermal signatures - Cloud contamination: Persistent cloud cover reduces observation frequency - Rapid vegetation regrowth: Short time lag between fire and regrowth obscures burn scars MCD64A1 applies quality filters: - Only pixels with uncertainty \< 20% are included - Commission/omission errors vary by biome - Global burned area may be underestimated by 20-30% [@giglio2010assessing] ### 2. Land Classification Uncertainty MODIS MCD12Q1 accuracy: - Overall accuracy: \~75% globally - Confusion between woody savannas and forests - Temporal lag in land-use change detection - Mixed pixels in heterogeneous landscapes Impact on emissions: Misclassification affects fuel consumption values (Table 2.4 ranges: 2.6 to 163.6 t ha⁻¹). ### 3. Fuel Consumption Uncertainty IPCC default values (Table 2.4) reflect: - High natural variability: Standard errors 30-50% of mean - Limited data for some vegetation types - Climate dependency: Values vary by FAO-FRA Ecological Zones, but within-zone heterogeneity remains Example uncertainties (Mean ± SE): - Primary Tropical Forest: 119.6 ± (range: 83.9 to 163.6) t ha⁻¹ - Savanna Woodland: 2.6 ± 0.1 (early dry season) to 4.6 ± 1.5 (late dry season) t ha⁻¹ ### 4. Combustion Factor Uncertainty Proportion of fuel consumed (Table 2.6): - Influenced by fire intensity, fuel moisture, weather conditions - Variability within vegetation types: SD often 0.1-0.2 (for factors ranging 0.2-0.9) - Incomplete combustion of coarse woody debris vs. near-complete combustion of grasses ### 5. Emission Factor Uncertainty IPCC defaults (Table 2.5) based on [@andreae2001emission]: - Values represent global averages across fire types - Standard deviations: 20-60% of mean values - Influenced by combustion efficiency (flaming vs. smoldering) Example (Savanna fires): - CO₂: 1613 ± 95 g kg⁻¹ - CH₄: 2.3 ± 0.9 g kg⁻¹ - N₂O: 0.21 ± 0.10 g kg⁻¹ #### Uncertainty Aggregation For IPCC Tier 1, cumulative uncertainty typically: - Activity data: ±30% for burned area - Emission estimates: ±50% for CH₄ and N₂O, ±30-40% for CO₂ Organic soil fires have particularly high uncertainty: - Remote sensing detection challenges - Limited field validation outside Southeast Asia - Difficulty distinguishing surface fires from subsurface peat combustion - Unknown proportion of fires on drained vs. undrained peatlands #### Uncertainty Reduction Country-Specific Improvements: 1. Integrate national fire databases with satellite observations 2. Conduct field campaigns to measure fuel loads and combustion factors 3. Develop regional emission factors from smoke sampling 4. Implement Tier 2/3 methods in high-emission areas Methodological Enhancements: 1. Multi-sensor fusion (MODIS + VIIRS + Sentinel) 2. Sub-pixel burned area algorithms 3. Burn severity mapping for combustion factors 4. Seasonal stratification of fuel moisture #### Managing Inter-Annual Variability High IAV from natural disturbances complicates trend analysis: - Wildfires can cause 2-3 orders of magnitude annual variation - Obscures anthropogenic emission trends - Challenges mitigation policy evaluation Optional Disaggregation Approach [@ipccChapter2Generic2019][^index-2]: Countries may separate: - Natural disturbance component: Emissions beyond normal variability or 95% CI - Anthropogenic component: Remaining managed land fluxes This disaggregation: - Improves clarity of human activity impacts - Maintains MLP total for official reporting - Requires documentation of methods and thresholds - Assumes eventual balance between disturbance emissions and regrowth ------------------------------------------------------------------------ ## Data Sources #### MCD64A1.061: Burn Areas Product Overview: The MCD64A1 Version 6.1 Burned Area product provides monthly, global gridded burned area at 500m spatial resolution, derived from combined Terra and Aqua MODIS observations. Key Features: - Temporal Resolution: Monthly composites, 2001–present - Spatial Resolution: 500m nominal pixel size - Algorithm: Decision tree classifier using surface reflectance time series - Output: Burn date encoded as ordinal day of year, with quality indicators Data Layers: - Burn Date (day of year when burn detected) - Burn Data Uncertainty (percentage uncertainty, 0-100%) - Quality Assurance flags FAOSTAT Implementation: - Quality filter applied: Only pixels with uncertainty \< 20% are retained - Aggregation: Pixel-level burn dates accumulated to annual country totals - GEE Asset: `MODIS/061/MCD64A1` Access: - NASA LP DAAC: https://lpdaac.usgs.gov/products/mcd64a1v061/ - Google Earth Engine: `ee.ImageCollection("MODIS/061/MCD64A1")` - User Guide: [@giglio2024collection] Limitations: - May miss small fires (\< 500m²) - Reduced accuracy in densely forested areas - Cloud obscuration limits temporal coverage in some regions - Detection lag in slow-spreading fires ------------------------------------------------------------------------ #### MCD12Q1.061: IGBP Land Cover Product Overview: MCD12Q1 provides annual global land cover classification at 500m resolution using supervised classification of MODIS reflectance data. Classification Scheme: FAOSTAT uses Land Cover Type 1 (IGBP 17-class legend): Forest Types (used for Forest Fire category): - Class 1: Evergreen Needleleaf Forests - Class 2: Evergreen Broadleaf Forests - Class 3: Deciduous Needleleaf Forests - Class 4: Deciduous Broadleaf Forests - Class 5: Mixed Forests Grassland/Savanna Types (used for Savanna Fire category): - Class 6: Closed Shrublands - Class 7: Open Shrublands - Class 8: Woody Savannas - Class 9: Savannas - Class 10: Grasslands - Other Classes: Croplands, Urban, Wetlands, etc. (Classes 11-17) FAOSTAT Implementation: - Annual land cover maps matched to fire detection year - Overlay burned pixels with contemporaneous land cover - Exclude built-up, barren, and water bodies - GEE Asset: `MODIS/061/MCD12Q1` Access: - NASA LP DAAC: https://lpdaac.usgs.gov/products/mcd12q1v061/ - Google Earth Engine: `ee.ImageCollection("MODIS/061/MCD12Q1")` - User Guide: [@sulla2018user] Accuracy: - Overall accuracy: \~75% globally - Higher accuracy in homogeneous landscapes - sConfusion common between woody savannas and open forests ------------------------------------------------------------------------ #### FAO Global Ecological Zones Dataset Overview:\ The Global Ecological Zones (GEZ) classification provides a climate-vegetation framework for stratifying fuel consumption parameters. Classification System Combines: - Climate domains (Polar, Boreal, Temperate, Subtropical, Tropical) - Ecological zones (e.g., Tropical moist forest, Tropical dry forest, Subtropical steppe) FAOSTAT Implementation: - Overlay burned forest pixels with GEZ to distinguish: - Humid Tropical Forest: Tropical moist ecological zones - Other Forest: All remaining forest types - Assign climate-dependent fuel consumption values from IPCC Table 2.4 Access: - FAO Forestry: http://www.fao.org/forest-resources-assessment/ - Document: FAO Global Ecological Zones for FAO forest reporting 2010 Update [@fao2012global] Application: - Determines fuel biomass consumption parameters - Stratifies emission factors where climate-sensitive - Links to FAO Forest Resources Assessment (FRA) reporting categories ------------------------------------------------------------------------ #### IPCC Soil Zones Layer Dataset Overview:\ The Harmonized World Soil Database (HWSD v1.2) provides global soil type classifications at 30 arc-second (\~1 km) resolution. Histosols Class:\ Organic soils characterized by: - High organic matter content (\> 20-30% by weight) - Formed in waterlogged conditions - Includes peatlands, bogs, fens FAOSTAT Implementation: - Burned area pixels overlaid with Histosols spatial layer - Fires in organic soils identified independently of land cover (except excluding built-up, barren, water) - Different fuel consumption and emission factors applied (IPCC Wetlands Supplement 2014) Access: - FAO Soils Portal: http://www.fao.org/soils-portal/ - IIASA: https://iiasa.ac.at/models-and-data/hwsd Critical Limitation: - FAOSTAT organic soil fire emissions set to zero outside Southeast Asia due to: - High uncertainty in peat fire detection - Limited validation data - Difficulty distinguishing drained vs. undrained peatlands - Uncertainty in human influence on fire occurrence Supplementary file `AllFires_OrganicSoils.csv` contains global estimates for research purposes. ------------------------------------------------------------------------ #### IPCC Climate Zones Layer Dataset Overview:\ IPCC Climate Zones classify global climate into categories used for stratifying biomass growth rates, decomposition rates, and fire parameters. Classification: - Polar - Boreal (Cold) - Temperate (Warm temperate, Cool temperate) - Tropical (Tropical wet, Tropical moist, Tropical dry, Tropical montane) Source: - Joint Research Centre (JRC) Climate Zones layer - Based on IPCC climatic zone definitions [@ipccChapter3Consistent2006] FAOSTAT Implementation: - Savanna fire fuel consumption values vary by climate zone (Table 2.4) - Organic soil emission factors stratified by climate [@ipcc2013RevisedSupplementary2014] - Links IPCC methodology to geographic reality Access: - JRC European Soil Data Centre (ESDAC): https://esdac.jrc.ec.europa.eu/ - Reference: JRC (2010) Climatic Zone layer ------------------------------------------------------------------------ #### Google Earth Engine Setup Platform Overview:\ Google Earth Engine is a cloud-based geospatial analysis platform providing: - Petabyte-scale satellite imagery archives - Planetary-scale computational infrastructure - JavaScript and Python APIs - Web-based Code Editor FAOSTAT Workflow Architecture: 1. Data Ingestion: All datasets (MCD64A1, MCD12Q1, GAUL, etc.) available as pre-processed GEE assets 2. Pixel-Level Processing: Fire algorithm applied to all global pixels 3. Zonal Statistics: Country-level aggregation via GAUL boundaries 4. Export: Results exported as CSV tables for FAOSTAT database Getting Started: 1. Account Registration: - Sign up at https://earthengine.google.com/signup/ - Academic/NGO accounts typically approved within 1-2 days 2. Code Editor Access: - Navigate to https://code.earthengine.google.com/ - JavaScript environment for interactive analysis 3. Python API Setup (optional): ``` python pip install earthengine-api earthengine authenticate ``` 4. Key GEE Concepts: - `ImageCollection`: Time series of raster images - `Image`: Single raster dataset - `FeatureCollection`: Vector dataset (e.g., country boundaries) - `Reducer`: Aggregation functions (sum, mean, etc.) FAOSTAT-Relevant GEE Assets:[^index-3] ``` javascript // Burned Area var burnedArea = ee.ImageCollection("MODIS/061/MCD64A1"); // Land Cover var landCover = ee.ImageCollection("MODIS/061/MCD12Q1"); // Country Boundaries var countries = ee.FeatureCollection("FAO/GAUL/2015/level0"); ``` ------------------------------------------------------------------------ [^index-1]: Note to self: 2025 wishlist [^index-2]: Direct Page Link: <https://www.ipcc-nggip.iges.or.jp/public/2019rf/pdf/4_Volume4/19R_V4_Ch02_Generic%20Methods.pdf#page=71> [^index-3]: Code provided Python and R, with some cells in Java for use natively in Earth Engine Codespace