3 Fuel Consumption

Overview

This chapter describes the estimation of fuel mass available for combustion (M_B) and combustion factors (C_f) following IPCC Tier 1 default values. These parameters, combined with burned area and vegetation classifications from Chapters 1-2, determine the total biomass burned in fire events.

Environment Setup

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", "knitr",
  "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", "reticulate",
  "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)

IPCC Fuel Biomass

3.0.1 Component in Equation 2.27

Fuel biomass consumption represents two components of the IPCC fire emissions equation:

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

Where: - M_B = Mass of fuel available for combustion (tonnes dry matter ha⁻¹) - C_f = Combustion factor (dimensionless, 0-1) - M_B × C_f = Actual biomass consumed per unit area (tonnes DM ha⁻¹)

Tier 1 Approach: Use IPCC default values stratified by:

Vegetation type (Forest, Savanna, Organic Soil)
Climate zone (Tropical, Temperate, Boreal)
Forest type (Humid Tropical vs. Other)
Season (Early vs. Late dry season for savannas)

3.0.2 IPCC Default Values - Quick Reference

Forest Fires (IPCC 2019 Table 2.4):

Forest Type	M_B (t DM ha⁻¹)	C_f	M_B × C_f
Humid Tropical Forest	119.6 ± 52.4	0.36 ± 0.09	43.1
Other Forest (Boreal)	41.0 ± 25.3	0.45 ± 0.15	18.5
Other Forest (Temperate)	50.4 ± 30.8	0.47 ± 0.14	23.7
Other Forest (Tropical)	69.4 ± 44.3	0.47 ± 0.14	32.6

Savanna Fires (IPCC 2019 Table 2.4):

Vegetation Type	Season	M_B (t DM ha⁻¹)	C_f	M_B × C_f
Savanna Woodland	Early	9.0 ± 7.5	0.29 ± 0.09	2.6
Savanna Woodland	Late	9.0 ± 7.5	0.51 ± 0.15	4.6
Savanna Grassland	Early	6.8 ± 6.0	0.31 ± 0.09	2.1
Savanna Grassland	Late	6.8 ± 6.0	0.83 ± 0.15	5.6
Other Grassland	-	13.4 ± 8.8	0.52 ± 0.15	7.0

Organic Soil Fires (IPCC 2014 Wetlands Supplement Table 2.6):

Soil Type	M_B (t DM ha⁻¹)	C_f	M_B × C_f
Tropical Peatland	41.2 ± 1.4	1.0	41.2

Note: Organic soil combustion factor = 1.0 assumes complete consumption of affected layer.

3.1 Fuel Available for Combustion (M_B)

3.1.1 Forest Fuel Loads

3.1.1.1 Components of Forest Fuel

Forest fuel includes multiple strata burned during fire:

Surface Fuels:

Litter (freshly fallen leaves, needles)
Duff (decomposed organic layer)
Herbaceous vegetation
Woody debris (< 7.5 cm diameter)

Ground Fuels:

Organic soil horizons (O and A)
Roots (fine and coarse)
Stumps and buried wood

Crown Fuels (stand-replacing fires only):

Foliage
Fine branches
Bark
Lichen and epiphytes

Tier 1 Simplification: M_B represents average fuel consumed across fire severity gradient, from surface fires (low M_B) to crown fires (high M_B).

3.1.1.2 Climate-Dependent Forest Fuel Loads

Boreal Forests (41.0 t DM ha⁻¹):

Slow decomposition
thick organic layers
Coniferous litter high in lignin
Permafrost limits root depth
Fire-adapted species (e.g., Pinus sylvestris)

Temperate Forests (50.4 t DM ha⁻¹):

Moderate decomposition rates
Mixed deciduous/coniferous litter
Seasonal accumulation patterns
Variable fire regimes

Tropical Forests (69.4 t DM ha⁻¹):

Rapid decomposition → less litter, but high biomass density
Fire-sensitive under natural conditions
Droughts enable fire spread

Humid Tropical Forests (119.6 t DM ha⁻¹):

Highest fuel loads due to:
Dense canopy biomass
Multi-story structure
Epiphytes, vines, and bamboo
Deep root systems
Primarily deforestation fires (slash-and-burn)
Stand-replacing events

3.1.2 Savanna Fuel Loads

3.1.2.1 Savanna Fuel Characteristics

Dominant fuel type: Herbaceous grass (70-90% of fuel consumed)

Key determinants of fuel load:

Annual precipitation: Grass productivity increases with rainfall
Season: Fuel cures (dries) during dry season
Fire frequency: Frequent fires -> lower woody component
Grazing pressure: Reduces available fuel

Spatial heterogeneity: Fuel patchy due to:

Termite mounds (fuel-free zones)
Tree shade (reduced grass growth)
Grazing concentration around water
Previous fire scars

3.1.2.2 Climate-Dependent Savanna Fuel Loads

Tropical Savannas (500-1500 mm rainfall):

High grass productivity
Distinct wet/dry seasons
Fuel curing rapid
M_B: 6.8-9.0 t DM ha⁻¹

Temperate Grasslands (250-750 mm rainfall):

Lower productivity than tropical
Variable seasonality
Mixed grass/forb composition
M_B: 13.4 t DM ha⁻¹ (includes crop residues)

Woody Savannas vs. Open Savannas:

Woody component adds coarse fuels, but canopy reduces grass fuel
Net effect: Similar M_B to open savanna
Combustion factor varies more than fuel load

3.1.2.3 Seasonal Dynamics

Early Dry Season:

Fuel partially cured (50-70% dry matter)
Lower combustion efficiency
Lower C_f: 0.29-0.31
Total consumption: 2.1-2.6 t DM ha⁻¹

Late Dry Season:

Fuel fully cured (>90% dry matter)
Complete combustion of grass layer
Higher C_f: 0.51-0.83
Total consumption: 4.6-5.6 t DM ha⁻¹

3.1.3 Organic Soil Fuel Loads

3.1.3.1 Peat Fuel Characteristics

Definition: Organic soils (Histosols) with >30 cm organic layer

Fuel properties:

High carbon content: 50-60% C by dry weight
Variable bulk density: 0.05-0.20 g cm⁻³
High moisture retention: 200-1000% (dry weight basis)
Smoldering combustion: Can burn for months

Fire behavior:

Surface fires: Consume litter and upper peat (10-30 cm)
Deep burns: Consume entire peat layer to mineral soil (>100 cm)
Drainage amplifies risk: Drained peat extremely flammable

3.1.3.2 Fuel Load Calculation

IPCC Wetlands Supplement Approach:

\[ M_B = D_{burn} \times BD \times 10 \]

Where:

$D_{burn}$ = Depth of burn (cm)
$BD$ = Bulk density of peat (g cm⁻³)
10 = Conversion factor (cm = m, g cm⁻³ = t ha⁻¹)

IPCC Default (Southeast Asia):

Depth of burn: 41 cm (range: 18-65 cm)
Bulk density: 0.10 g cm⁻³ (range: 0.05-0.15)
M_B: 41.2 t DM ha⁻¹

High uncertainty: Coefficient of variation ~50%

3.2 Combustion Factor (C_f)

3.2.1 Definition and Interpretation

Combustion Factor (C_f) = Proportion of available fuel actually consumed by fire

\[ C_f = \frac{\text{Mass of fuel consumed}}{\text{Mass of fuel available}} \quad (0 \leq C_f \leq 1) \]

Determinants:

Fuel moisture: Higher moisture = lower C_f
Fire intensity: Hotter fires → higher C_f
Fire type: Surface vs. crown fire
Fuel particle size: Fine fuels (< 0.6 cm) burn more completely
Weather conditions: Wind, humidity, temperature

3.2.2 Forest Combustion Factors

Low C_f for Humid Tropical Forests (0.36):

High ambient moisture
Rapid fuel accumulation = green fuels mixed with dead
Incomplete combustion of large woody debris
Char and partially burned material remain

Moderate C_f for Other Forests (0.45-0.47):

Drier climates = lower fuel moisture
More complete consumption of surface fuels
Variability based on fire severity:
Surface fires: C_f ~ 0.20-0.40
Crown fires: C_f ~ 0.60-0.80
Default represents average across severities

3.2.3 Savanna Combustion Factors

Early Dry Season (0.29-0.31):

Fuel partially cured
High moisture content in live grass (30-50%)
Incomplete combustion - Patchy fire spread
Much grass left standing

Late Dry Season (0.51-0.83):

Fuel fully cured (<10% moisture)
Complete consumption of grass layer
Only woody stems and char remain
Higher C_f for grasslands vs. woody savannas:
Grassland: C_f = 0.83 (almost complete)
Woody Savanna: C_f = 0.51 (trees create shade/barriers)

Physical explanation: \[ C_f \propto \frac{1}{\text{Fuel Moisture}} \times \text{Fire Intensity} \]

Late season = Lower moisture × Higher intensity = Higher C_f

3.2.4 Organic Soil Combustion Factors

Complete Consumption (C_f = 1.0):

Once peat ignites, burns to completion within affected depth
Smoldering combustion eliminates unburned fuel
Char production minimal
Caveat: C_f = 1.0 applies to depth of burn, not total peat column
If burn depth = 41 cm, that 41 cm is fully consumed
Deeper peat layers remain intact (if fire extinguished)

3.3 Applying Fuel Consumption Parameters

3.3.1 Workflow: Vegetation Type to Fuel Values

library(tidyverse)

# IPCC 2019 Table 2.4 - Default fuel consumption values
fuel_consumption_table <- tribble(
  ~fire_category, ~vegetation_type, ~climate_zone, ~season, ~MB_tDM_ha, ~Cf, ~MB_Cf,

  # Forest Fires
  "Forest", "Humid_Tropical_Forest", "Tropical_Moist", NA, 119.6, 0.36, 43.1,
  "Forest", "Other_Forest", "Boreal", NA, 41.0, 0.45, 18.5,
  "Forest", "Other_Forest", "Temperate", NA, 50.4, 0.47, 23.7,
  "Forest", "Other_Forest", "Tropical", NA, 69.4, 0.47, 32.6,

  # Savanna Fires
  "Savanna", "Savanna_Woodland", "All", "Early", 9.0, 0.29, 2.6,
  "Savanna", "Savanna_Woodland", "All", "Late", 9.0, 0.51, 4.6,
  "Savanna", "Savanna_Grassland", "All", "Early", 6.8, 0.31, 2.1,
  "Savanna", "Savanna_Grassland", "All", "Late", 6.8, 0.83, 5.6,
  "Savanna", "Other_Grassland", "All", NA, 13.4, 0.52, 7.0,

  # Organic Soil Fires
  "Organic_Soil", "Tropical_Peatland", "Tropical", NA, 41.2, 1.0, 41.2
)

print(fuel_consumption_table)
## # A tibble: 10 × 7
##   fire_category vegetation_type       climate_zone   season
##   <chr>         <chr>                 <chr>          <chr> 
## 1 Forest        Humid_Tropical_Forest Tropical_Moist <NA>  
## 2 Forest        Other_Forest          Boreal         <NA>  
## 3 Forest        Other_Forest          Temperate      <NA>  
## 4 Forest        Other_Forest          Tropical       <NA>  
## 5 Savanna       Savanna_Woodland      All            Early 
## 6 Savanna       Savanna_Woodland      All            Late  
## # ℹ 4 more rows
## # ℹ 3 more variables: MB_tDM_ha <dbl>, Cf <dbl>,
## #   MB_Cf <dbl>

# Save for use in emissions calculation
write.csv(
  fuel_consumption_table,
  "IPCC_fuel_consumption_defaults.csv",
  row.names = FALSE
)

3.4 GEE Initialisation

# Point environment to your configured python
reticulate::use_python("/opt/local/bin/python", required=T)

# Initialise Earth Engine
ee <- reticulate::import("ee")
ee$Initialize(project = "murphys-deforisk")
cat("GEE initialised\n")
## GEE initialised

# Helper: visualize EE tile in tmap/leaflet
ee_tile_url <- function(ee_image, vis_params) {
  ee_image$getMapId(vis_params)$tile_fetcher$url_format
}

# Helper: EE FeatureCollection => sf
ee_to_sf <- function(fc) {
  geojsonsf::geojson_sf(
    jsonlite::toJSON(fc$getInfo(), auto_unbox = TRUE))
}

# Load AOI boundaries from FAO GAUL
aoi_country_ee <- ee$FeatureCollection("FAO/GAUL/2015/level0")$
  filter(ee$Filter$eq("ADM0_NAME", "Honduras"))
aoi_states_ee  <- ee$FeatureCollection("FAO/GAUL/2015/level1")$
  filter(ee$Filter$eq("ADM0_NAME", "Honduras"))
aoi_country_sf <- ee_to_sf(aoi_country_ee)
aoi_states_sf  <- ee_to_sf(aoi_states_ee)

cat("AOI loaded:", aoi_country_ee$size()$getInfo(), "country,",
    aoi_states_ee$size()$getInfo(), "states\n")
## AOI loaded: 1 country, 19 states

3.4.1 Upstream Inputs from Chapters 1-2

# --- Chapter 1: Burned area (MCD64A1) ---
mcd64 <- ee$ImageCollection("MODIS/061/MCD64A1")
burned_2010 <- mcd64$
  filterBounds(aoi_country_ee$geometry())$
  filterDate("2010-01-01", "2010-12-31")

# FAOSTAT quality filter (matching Ch1)
faostat_filter <- function(image) {
  unc  <- image$select("Uncertainty")
  burn <- image$select("BurnDate")
  mask <- unc$lt(20L)$And(burn$gt(0L))
  image$updateMask(mask)
}

# Annual burned mask for 2010
annual_burned_2010 <- burned_2010$
  map(faostat_filter)$
  select("BurnDate")$max()$gt(0L)$selfMask()

# --- Chapter 2: Land cover & vegetation masks ---
lc_2010 <- ee$ImageCollection("MODIS/061/MCD12Q1")$
  select("LC_Type1")$
  filterDate("2010-01-01", "2010-12-31")$
  first()$
  clip(aoi_country_ee$geometry())

# Forest mask (IGBP 1-5)
forest_mask <- lc_2010$gte(1L)$And(lc_2010$lte(5L))

# Savanna/grassland mask (IGBP 6-10)
savanna_mask <- lc_2010$gte(6L)$And(lc_2010$lte(10L))

# Climate-based humid tropical delineation
bio <- ee$Image("WORLDCLIM/V1/BIO")
precip <- bio$select("bio12")
temp   <- bio$select("bio01")

tropical <- temp$gt(180L)
moist    <- precip$gt(1500L)
humid_tropical_mask <- tropical$And(moist)

# Forest sub-types
humid_tropical_forest <- forest_mask$And(humid_tropical_mask)
other_forest          <- forest_mask$And(humid_tropical_mask$Not())

# Pixel area helper
pixelArea_ha <- ee$Image$pixelArea()$divide(10000L)

cat("Upstream inputs loaded: MCD64A1 + MCD12Q1 + WorldClim\n")
## Upstream inputs loaded: MCD64A1 + MCD12Q1 + WorldClim

3.5 Spatial Implementation in Google Earth Engine

3.5.1 Creating Fuel Consumption Rasters

# Climate zone masks
boreal    <- temp$lt(0L)
temperate <- temp$gte(0L)$And(temp$lte(180L))

# Create fuel consumption image (M_B × C_f in t DM ha⁻¹)
fuel_consumption <- ee$Image(0)$rename("fuel_consumption")

# 1. Humid Tropical Forest (M_B × C_f = 43.1)
htf_mask <- forest_mask$And(humid_tropical_mask)
fuel_consumption <- fuel_consumption$where(htf_mask, 43.1)

# 2. Other Forest - Boreal (M_B × C_f = 18.5)
forest_boreal <- forest_mask$And(boreal)$And(humid_tropical_mask$Not())
fuel_consumption <- fuel_consumption$where(forest_boreal, 18.5)

# 3. Other Forest - Temperate (M_B × C_f = 23.7)
forest_temperate <- forest_mask$And(temperate)$And(humid_tropical_mask$Not())
fuel_consumption <- fuel_consumption$where(forest_temperate, 23.7)

# 4. Other Forest - Tropical Dry (M_B × C_f = 32.6)
forest_tropical <- forest_mask$And(tropical)$And(humid_tropical_mask$Not())
fuel_consumption <- fuel_consumption$where(forest_tropical, 32.6)

# 5. Savanna/Grassland (average = 7.0 t DM ha⁻¹)
fuel_consumption <- fuel_consumption$where(savanna_mask, 7.0)

# Apply to burned areas only
fuel_burned <- fuel_consumption$
  updateMask(annual_burned_2010)$
  clip(aoi_country_ee$geometry())

# Visualize
fuel_vis <- list(
  min = 0L, max = 50L,
  palette = c("yellow", "orange", "red", "darkred")
)

tmap::tmap_mode("view")
tmap::tm_shape(aoi_country_sf) + tmap::tm_borders(col = "white", lwd = 1.5) +
  tmap::tm_tiles(
    ee_tile_url(fuel_burned, fuel_vis),
    group = "Fuel Consumption (t DM ha⁻¹)"
  ) +
  tmap::tm_add_legend(
    type = "polygons",
    fill = c("yellow", "orange", "red", "darkred"),
    labels = c("Savanna (~7)", "Temperate Forest (~24)",
               "Tropical Forest (~33)", "Humid Tropical (~43)"),
    title = "Fuel Consumption (t DM ha⁻¹)"
  ) +
  tmap::tm_scalebar(position = c("RIGHT", "BOTTOM"), text.size = 0.5) +
  tmap::tm_basemap("Esri.WorldImagery")

3.5.2 Seasonal Stratification for Savannas

# Extract burn month from MCD64A1
extractBurnMonth <- function(mcd64_image) {
  burn_date <- mcd64_image$select("BurnDate")
  img_date  <- ee$Date(mcd64_image$get("system:time_start"))
  month     <- img_date$get("month")

  month_band <- ee$Image$constant(month)$
    updateMask(burn_date$gt(0L))$
    rename("burn_month")$
    toInt()

  burn_date$addBands(month_band)
}

# Apply to 2010 collection (quality-filtered)
mcd64_with_months <- burned_2010$
  map(faostat_filter)$
  map(extractBurnMonth)

# Composite: latest month if pixel burned multiple times
month_composite <- mcd64_with_months$
  select("burn_month")$
  max()

# Honduras dry season (Central America)
# Early dry: Feb-Mar (months 2-3)
# Late dry:  Apr-May (months 4-5)
early_season <- month_composite$gte(2L)$And(month_composite$lte(3L))
late_season  <- month_composite$gte(4L)$And(month_composite$lte(5L))

# Savanna fire subsets
savanna_fires_2010 <- annual_burned_2010$updateMask(savanna_mask)

# Woody savanna (IGBP class 8)
woody_savanna <- lc_2010$eq(8L)
# Grassland/savanna (IGBP classes 9-10)
grassland <- lc_2010$gte(9L)$And(lc_2010$lte(10L))

# Create seasonal fuel consumption
fuel_seasonal <- ee$Image(0)$rename("fuel_seasonal")
fuel_seasonal <- fuel_seasonal$
  where(woody_savanna$And(early_season), 2.6)$
  where(woody_savanna$And(late_season), 4.6)$
  where(grassland$And(early_season), 2.1)$
  where(grassland$And(late_season), 5.6)

# Mask to burned savanna areas
fuel_seasonal_burned <- fuel_seasonal$
  updateMask(savanna_fires_2010)$
  clip(aoi_country_ee$geometry())

# Visualize
tmap::tm_shape(aoi_country_sf) + tmap::tm_borders(col = "white", lwd = 1.5) +
  tmap::tm_tiles(
    ee_tile_url(fuel_seasonal_burned,
      list(min = 0L, max = 6L,
           palette = c("lightyellow", "yellow", "orange"))),
    group = "Savanna Fuel Consumption (Seasonal)"
  ) +
  tmap::tm_add_legend(
    type = "polygons",
    fill = c("lightyellow", "yellow", "orange"),
    labels = c("Early grassland (2.1)", "Early woody (2.6) / Late woody (4.6)",
               "Late grassland (5.6)"),
    title = "Savanna Fuel (t DM ha⁻¹)"
  ) +
  tmap::tm_scalebar(position = c("RIGHT", "BOTTOM"), text.size = 0.5) +
  tmap::tm_basemap("Esri.WorldImagery")

3.5.3 Calculating Total Biomass Burned

# Biomass burned per pixel (t DM) = fuel_consumption (t ha⁻¹) × area (ha)
biomass_per_pixel <- fuel_burned$multiply(pixelArea_ha)

# Forest biomass
forest_biomass <- biomass_per_pixel$
  updateMask(forest_mask)$
  reduceRegion(
    reducer   = ee$Reducer$sum(),
    geometry  = aoi_country_ee$geometry(),
    scale     = 500L,
    maxPixels = 1e9
  )$getInfo()

# Savanna biomass
savanna_biomass <- biomass_per_pixel$
  updateMask(savanna_mask)$
  reduceRegion(
    reducer   = ee$Reducer$sum(),
    geometry  = aoi_country_ee$geometry(),
    scale     = 500L,
    maxPixels = 1e9
  )$getInfo()

forest_t  <- forest_biomass[["fuel_consumption"]] %||% 0
savanna_t <- savanna_biomass[["fuel_consumption"]] %||% 0

cat("Honduras 2010 — Total Biomass Burned\n")
## Honduras 2010 — Total Biomass Burned
cat("  Forest biomass burned: ", round(forest_t / 1000, 1), "Gg DM\n")
##   Forest biomass burned:  1683 Gg DM
cat("  Savanna biomass burned:", round(savanna_t / 1000, 1), "Gg DM\n")
##   Savanna biomass burned: 815 Gg DM
cat("  Total:                 ", round((forest_t + savanna_t) / 1000, 1), "Gg DM\n")
##   Total:                  2497 Gg DM

3.5.4 Matching Vegetation Classifications to Fuel Values

# Burned area by vegetation type (from Ch2 overlay)
htf_fires  <- annual_burned_2010$updateMask(humid_tropical_forest)
of_fires   <- annual_burned_2010$updateMask(other_forest)
sav_fires  <- annual_burned_2010$updateMask(savanna_mask)

# Calculate areas (ha)
htf_area <- htf_fires$multiply(pixelArea_ha)$reduceRegion(
  reducer = ee$Reducer$sum(), geometry = aoi_country_ee$geometry(),
  scale = 500L, maxPixels = 1e9)$getInfo()
of_area <- of_fires$multiply(pixelArea_ha)$reduceRegion(
  reducer = ee$Reducer$sum(), geometry = aoi_country_ee$geometry(),
  scale = 500L, maxPixels = 1e9)$getInfo()
sav_area <- sav_fires$multiply(pixelArea_ha)$reduceRegion(
  reducer = ee$Reducer$sum(), geometry = aoi_country_ee$geometry(),
  scale = 500L, maxPixels = 1e9)$getInfo()

htf_ha <- htf_area[["BurnDate"]] %||% 0
of_ha  <- of_area[["BurnDate"]] %||% 0
sav_ha <- sav_area[["BurnDate"]] %||% 0

# Apply IPCC default fuel consumption
honduras_fuel <- tribble(
  ~fire_category, ~vegetation_type, ~area_ha, ~MB_tDM_ha, ~Cf, ~MB_Cf,
  "Forest", "Humid_Tropical_Forest", round(htf_ha, 1), 119.6, 0.36, 43.1,
  "Forest", "Other_Forest",          round(of_ha, 1),  69.4, 0.47, 32.6,
  "Savanna", "Savanna_Mixed",        round(sav_ha, 1),  7.9, 0.46,  3.6
) %>%
  mutate(
    biomass_burned_tDM = area_ha * MB_Cf,
    biomass_burned_Gg  = biomass_burned_tDM / 1000
  )

print(honduras_fuel)
## # A tibble: 3 × 8
##   fire_category vegetation_type      area_ha MB_tDM_ha    Cf
##   <chr>         <chr>                  <dbl>     <dbl> <dbl>
## 1 Forest        Humid_Tropical_Fore…  33688.     120.   0.36
## 2 Forest        Other_Forest           7163.      69.4  0.47
## 3 Savanna       Savanna_Mixed        116378.       7.9  0.46
## # ℹ 3 more variables: MB_Cf <dbl>,
## #   biomass_burned_tDM <dbl>, biomass_burned_Gg <dbl>

cat("\nHonduras 2010 — Summary\n")
## 
## Honduras 2010 — Summary
cat("  Total burned area:     ", sum(honduras_fuel$area_ha), "ha\n")
##   Total burned area:      157229 ha
cat("  Total biomass burned:  ", round(sum(honduras_fuel$biomass_burned_Gg), 1), "Gg DM\n")
##   Total biomass burned:   2104 Gg DM

3.6 Uncertainty in Fuel Consumption Estimates

3.6.1 Sources of Uncertainty

1. IPCC Default Value Uncertainty:

IPCC provides 95% confidence intervals for all default values:

Parameter	Relative Uncertainty
M_B (Forest)	±40-60%
M_B (Savanna)	±60-90%
C_f (Forest)	±20-30%
C_f (Savanna)	±30-40%
M_B×C_f (combined)	±50-70%

2. Spatial Extrapolation Uncertainty:

IPCC values derived from limited field studies
Geographic bias toward well-studied regions (North America, Australia)
Under-representation of tropical systems

3. Fire Severity Variability:

Default values represent averages across fire severity spectrum
Individual fires may have 2-3× higher/lower consumption
No severity stratification in Tier 1

4. Temporal Variability:

Fuel accumulation varies with:
Time since last fire
Precipitation anomalies (ENSO, droughts)
Forest management (fuel reduction treatments)
Tier 1 assumes constant fuel loads

3.6.2 Quantifying Combined Uncertainty

Following IPCC Error Propagation (Chapter 3, Equation 3.1):

\[ U_{total}^2 = U_A^2 + U_{MB}^2 + U_{Cf}^2 + U_{Gef}^2 \]

For Tier 1 Forest Fires:

$U_A$ = ±30% (from Chapter 1 - burned area)
$U_{MB}$ = ±50% (IPCC Table 2.4)
$U_{Cf}$ = ±25% (IPCC Table 2.4)
$U_{Gef}$ = ±30% (Chapter 4 - emission factors)

Combined uncertainty (quadrature): \[ U_{total} = \sqrt{30^2 + 50^2 + 25^2 + 30^2} = \sqrt{4350} \approx 66\% \]

For Tier 1 Savanna Fires:

$U_A$ = ±30%
$U_{MB}$ = ±80% (higher variability)
$U_{Cf}$ = ±35%
$U_{Gef}$ = ±30%

Combined uncertainty: \[ U_{total} = \sqrt{30^2 + 80^2 + 35^2 + 30^2} = \sqrt{9925} \approx 100\% \]

Interpretation: Tier 1 uncertainties are large but expected given coarse default values. Tier 2/3 approaches reduce uncertainty through country-specific data.

3.6.3 Sensitivity Analysis

library(tidyverse)

# Define parameter ranges (from IPCC confidence intervals)
# Honduras is predominantly humid tropical forest
fuel_params <- expand.grid(
  MB_forest = c(119.6 - 52.4, 119.6, 119.6 + 52.4),
  Cf_forest = c(0.36 - 0.09, 0.36, 0.36 + 0.09),
  area_ha = round(htf_ha, 0)
)

# Calculate biomass burned for each parameter combination
fuel_params <- fuel_params %>%
  mutate(
    biomass_tDM = area_ha * MB_forest * Cf_forest,
    biomass_Gg = biomass_tDM / 1000,
    scenario = paste0("MB=", round(MB_forest, 1),
                      ", Cf=", round(Cf_forest, 2))
  )

# Visualize sensitivity
ggplot(fuel_params, aes(x = MB_forest, y = Cf_forest,
                        fill = biomass_Gg)) +
  geom_tile() +
  geom_text(aes(label = round(biomass_Gg, 0)),
            color = "white", fontface = "bold") +
  scale_fill_viridis_c(name = "Biomass\nBurned (Gg)") +
  labs(
    title = "Sensitivity of Biomass Burned to Fuel Parameters",
    subtitle = paste("Honduras Humid Tropical Forest Fires (",
                     round(htf_ha, 0), "ha)"),
    x = "Fuel Load MB (t DM ha⁻¹)",
    y = "Combustion Factor Cf"
  ) +
  theme_minimal()


# Calculate coefficient of variation
cv_biomass <- sd(fuel_params$biomass_Gg) /
  mean(fuel_params$biomass_Gg) * 100
cat("Coefficient of Variation:",
    round(cv_biomass, 1), "%\n")
## Coefficient of Variation: 44.4 %

3.7 Advanced Topics

3.7.1 Country-Specific Fuel Data (Tier 2)

Moving beyond defaults:

When available, countries can use:

National forest inventory data: Standing biomass, litter depths
Fire behavior models: BEHAVE, FuelCalc, FCCS
Field measurements: Post-fire fuel consumption studies
Remote sensing: LiDAR for 3D fuel structure

Example: Honduras:

# Hypothetical: Honduras National Forest Inventory fuel loads
honduras_fuel_data <- tribble(
  ~forest_type, ~MB_tDM_ha, ~Cf, ~source,
  "Pinus caribaea (lowland pine)", 72.5, 0.42, "ICF National Inventory 2018",
  "Broadleaf humid forest",       125.3, 0.34, "SIGMOF field plots 2019",
  "Dry broadleaf / pine-oak",      55.8, 0.45, "La Mosquitia fire studies 2017"
)

# These values would replace IPCC defaults for Honduran forests
# Result: Reduced uncertainty (~±25% over ±50%)
print(honduras_fuel_data)
## # A tibble: 3 × 4
##   forest_type                   MB_tDM_ha    Cf source      
##   <chr>                             <dbl> <dbl> <chr>       
## 1 Pinus caribaea (lowland pine)      72.5  0.42 ICF Nationa…
## 2 Broadleaf humid forest            125.   0.34 SIGMOF fiel…
## 3 Dry broadleaf / pine-oak           55.8  0.45 La Mosquiti…

3.7.2 Tier 2+ Fire Severity Stratification (Pending)

Fire severity classes:

Severity	dNBR Range	C_f Multiplier
Unburned	< 0.1	0
Low	0.1 - 0.27	0.5×
Moderate	0.27 - 0.44	1.0× (default)
High	> 0.44	1.5×

# Calculate dNBR from Landsat or Sentinel-2 to adjust Cf
# Pre-fire composite (Jan-Feb 2010)
prefire <- ee$ImageCollection("LANDSAT/LT05/C02/T1_L2")$
  filterBounds(aoi_country_ee$geometry())$
  filterDate("2010-01-01", "2010-02-28")$
  median()

# Post-fire composite (Jun-Jul 2010)
postfire <- ee$ImageCollection("LANDSAT/LT05/C02/T1_L2")$
  filterBounds(aoi_country_ee$geometry())$
  filterDate("2010-06-01", "2010-07-31")$
  median()

# Calculate NBR
calculateNBR <- function(img) {
  nir  <- img$select("SR_B4")
  swir <- img$select("SR_B7")
  nir$subtract(swir)$divide(nir$add(swir))$rename("NBR")
}
nbr_pre  <- calculateNBR(prefire)
nbr_post <- calculateNBR(postfire)
dnbr     <- nbr_pre$subtract(nbr_post)

# Define severity classes
low_severity      <- dnbr$gte(0.1)$And(dnbr$lt(0.27))
moderate_severity <- dnbr$gte(0.27)$And(dnbr$lt(0.44))
high_severity     <- dnbr$gte(0.44)

# Adjust combustion factor for humid tropical forest
Cf_default <- 0.36
Cf_adjusted <- ee$Image(Cf_default)$
  where(low_severity, Cf_default * 0.5)$
  where(high_severity, Cf_default * 1.5)

# Recalculate fuel consumption
MB <- 119.6
fuel_consumption_adjusted <- ee$Image(MB)$multiply(Cf_adjusted)

3.7.3 Fuel Moisture Effects

Fuel moisture reduces combustion efficiency:

\[ C_f = C_{f,dry} \times e^{-k \times MC} \]

Where:

$C_{f,dry}$ = Combustion factor at 0% moisture
$MC$ = Moisture content (%)
$k$ = Decay constant (~0.02 for grass, ~0.01 for wood)

# Load ERA5 relative humidity or precipitation data
precip_monthly <- ee$ImageCollection("ECMWF/ERA5/MONTHLY")$
  select("total_precipitation")

# Get precipitation in month of fire
# (Would need to match with burn dates from Chapter 1)

# For demonstration: use annual average
precip_annual <- precip_monthly$
  filterDate("2010-01-01", "2010-12-31")$
  mean()

# Convert to fuel moisture proxy (simplified)
# Assume: High precip → High moisture → Low Cf
# (Actual relationship requires mechanistic model)

# This is TIER 3 level - requires significant data and validation

3.8 Integrated Emissions Calculations

3.8.1 Output Formats

Required data structure:

# Create comprehensive fuel consumption table
fuel_output <- tribble(
  ~country, ~year, ~fire_category, ~vegetation_type, ~climate_zone, ~season,
  ~area_ha, ~MB_tDM_ha, ~Cf, ~biomass_burned_Gg,

  "Honduras", 2010, "Forest", "Humid_Tropical_Forest", "Tropical_Moist", NA,
  round(htf_ha, 0), 119.6, 0.36, round(htf_ha * 43.1 / 1000, 1),

  "Honduras", 2010, "Forest", "Other_Forest", "Tropical", NA,
  round(of_ha, 0), 69.4, 0.47, round(of_ha * 32.6 / 1000, 1),

  "Honduras", 2010, "Savanna", "Savanna_Mixed", "Tropical", "Late",
  round(sav_ha, 0), 7.9, 0.46, round(sav_ha * 3.6 / 1000, 1)
)

# Save for Chapter 4
write.csv(
  fuel_output,
  "Honduras_FuelConsumption_2010.csv",
  row.names = FALSE
)

print(fuel_output)
## # A tibble: 3 × 10
##   country   year fire_category vegetation_type  climate_zone
##   <chr>    <dbl> <chr>         <chr>            <chr>       
## 1 Honduras  2010 Forest        Humid_Tropical_… Tropical_Mo…
## 2 Honduras  2010 Forest        Other_Forest     Tropical    
## 3 Honduras  2010 Savanna       Savanna_Mixed    Tropical    
## # ℹ 5 more variables: season <chr>, area_ha <dbl>,
## #   MB_tDM_ha <dbl>, Cf <dbl>, biomass_burned_Gg <dbl>

3.8.2 QC Checklist

Pre-calculation:

Vegetation classifications complete (from Chapter 2)
Climate zones assigned correctly
Seasonal stratification applied (savannas only)
IPCC fuel consumption table loaded correctly

Calculation:

Fuel values matched to vegetation types
Units consistent (t DM ha⁻¹, Gg)
Biomass = Area × M_B × C_f
No missing values (all vegetation types have fuel values)

Post-calculation:

Biomass totals reasonable (order of magnitude check)
Sum of vegetation-specific biomass = total biomass
Uncertainty quantified and documented
Results comparable to previous years (if available)

Validation:

Compare with published values (if available)
Check extreme values (outliers)
Sensitivity analysis completed
Documentation complete

3.9 Notes and Next Steps

3.9.1 Key Takeaways

Fuel consumption (M_B × C_f) varies by:
- Vegetation type: Forest (18-43 t ha⁻¹) vs. Savanna (2-7 t ha⁻¹)
- Climate zone: Boreal < Temperate < Tropical < Humid Tropical
- Season: Early (lower) vs. Late (higher) dry season for savannas
IPCC Tier 1 defaults provide globally consistent estimates but have high uncertainty (±50-100%)
Combustion factor (C_f) represents proportion of fuel consumed, ranging from 0.29 (early season savanna) to 1.0 (organic soils)
Total biomass burned calculated as: Area (ha) × M_B (t ha⁻¹) × C_f
Tier 2/3 improvements possible through country-specific data, fire severity mapping, and fuel moisture modeling

3.9.2 Chapter 4:

Apply IPCC emission factors (G_ef) by gas (CO₂, CH₄, N₂O)
Calculate trace gas emissions: Biomass Burned × G_ef
Report emissions by IPCC category (Forest, Savanna, Organic Soils)
Convert to CO₂-equivalents for aggregate reporting

Critical linkage: Each fire event carries forward:

Biomass burned (Gg DM) by vegetation type
Fire category (Forest/Savanna/Organic Soil) for emission factor selection
Climate zone (for organic soil CH₄ emission factors)
Uncertainty estimates for error propagation

--- title: "Fuel Consumption" execute: echo: true format: html: toc: true toc-location: right toc-depth: 3 toc-title: "**On this page**" highlight-style: pygments page-layout: article editor_options: markdown: wrap: 60 engine: knitr bibliography: ../references/references.bib csl: ../references/apa.csl citeproc: true --- ## Overview {.unnumbered} This chapter describes the estimation of fuel mass available for combustion (M~B~) and combustion factors (C~f~) following IPCC Tier 1 default values. These parameters, combined with burned area and vegetation classifications from Chapters 1-2, determine the total biomass burned in fire events. ### Environment Setup {.unnumbered} ```{r} #| warning: false #| message: false #| echo: true #| 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", "knitr", "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", "reticulate", "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/02-fuel-stratification/assets/aa_burn_trend.png'); background-size: contain; background-position: 50% 50%; padding-top: 80px !important; background-repeat: no-repeat; } ``` ------------------------------------------------------------ IPCC Fuel Biomass ### Component in Equation 2.27 {#sec-fuel-equation} Fuel biomass consumption represents two components of the IPCC fire emissions equation: $$ L_{fire} = A \times M_B \times C_f \times G_{ef} \times 10^{-3} $$ Where: - M~B~ = Mass of fuel available for combustion (tonnes dry matter ha⁻¹) - C~f~ = Combustion factor (dimensionless, 0-1) - M~B~ × C~f~ = Actual biomass consumed per unit area (tonnes DM ha⁻¹) Tier 1 Approach: Use IPCC default values stratified by: 1. Vegetation type (Forest, Savanna, Organic Soil) 2. Climate zone (Tropical, Temperate, Boreal) 3. Forest type (Humid Tropical vs. Other) 4. Season (Early vs. Late dry season for savannas) ### IPCC Default Values - Quick Reference {#sec-fuel-defaults} Forest Fires (IPCC 2019 Table 2.4): | Forest Type | M~B~ (t DM ha⁻¹) | C~f~ | M~B~ × C~f~ | |-------------|---------|--------|-------| | Humid Tropical Forest | 119.6 ± 52.4 | 0.36 ± 0.09 | 43.1 | | Other Forest (Boreal) | 41.0 ± 25.3 | 0.45 ± 0.15 | 18.5 | | Other Forest (Temperate) | 50.4 ± 30.8 | 0.47 ± 0.14 | 23.7 | | Other Forest (Tropical) | 69.4 ± 44.3 | 0.47 ± 0.14 | 32.6 | Savanna Fires (IPCC 2019 Table 2.4): | Vegetation Type | Season | M~B~ (t DM ha⁻¹) | C~f~ | M~B~ × C~f~ | |------------|------|---------|--------|------| | Savanna Woodland | Early | 9.0 ± 7.5 | 0.29 ± 0.09 | 2.6 | | Savanna Woodland | Late | 9.0 ± 7.5 | 0.51 ± 0.15 | 4.6 | | Savanna Grassland | Early | 6.8 ± 6.0 | 0.31 ± 0.09 | 2.1 | | Savanna Grassland | Late | 6.8 ± 6.0 | 0.83 ± 0.15 | 5.6 | | Other Grassland | \- | 13.4 ± 8.8 | 0.52 ± 0.15 | 7.0 | Organic Soil Fires (IPCC 2014 Wetlands Supplement Table 2.6): | Soil Type | M~B~ (t DM ha⁻¹) | C~f~ | M~B~ × C~f~ | |-------------------|------------------|------|-------------| | Tropical Peatland | 41.2 ± 1.4 | 1.0 | 41.2 | Note: Organic soil combustion factor = 1.0 assumes complete consumption of affected layer. ------------------------------------------------------------ ## Fuel Available for Combustion (M~B~) {#sec-fuel-available} ### Forest Fuel Loads {#sec-fuel-forest} #### Components of Forest Fuel {#sec-fuel-forest-components} Forest fuel includes multiple strata burned during fire: Surface Fuels: - Litter (freshly fallen leaves, needles) - Duff (decomposed organic layer) - Herbaceous vegetation - Woody debris (\< 7.5 cm diameter) Ground Fuels: - Organic soil horizons (O and A) - Roots (fine and coarse) - Stumps and buried wood Crown Fuels (stand-replacing fires only): - Foliage - Fine branches - Bark - Lichen and epiphytes Tier 1 Simplification: M~B~ represents average fuel consumed across fire severity gradient, from surface fires (low M~B~) to crown fires (high M~B~). #### Climate-Dependent Forest Fuel Loads {#sec-fuel-forest-climate} Boreal Forests (41.0 t DM ha⁻¹): - Slow decomposition - thick organic layers - Coniferous litter high in lignin - Permafrost limits root depth - Fire-adapted species (e.g., Pinus sylvestris) Temperate Forests (50.4 t DM ha⁻¹): - Moderate decomposition rates - Mixed deciduous/coniferous litter - Seasonal accumulation patterns - Variable fire regimes Tropical Forests (69.4 t DM ha⁻¹): - Rapid decomposition → less litter, but high biomass density - Fire-sensitive under natural conditions - Droughts enable fire spread Humid Tropical Forests (119.6 t DM ha⁻¹): - Highest fuel loads due to: - Dense canopy biomass - Multi-story structure - Epiphytes, vines, and bamboo - Deep root systems - Primarily deforestation fires (slash-and-burn) - Stand-replacing events ### Savanna Fuel Loads {#sec-fuel-savanna} #### Savanna Fuel Characteristics {#sec-fuel-savanna-chars} Dominant fuel type: Herbaceous grass (70-90% of fuel consumed) Key determinants of fuel load: 1. Annual precipitation: Grass productivity increases with rainfall 2. Season: Fuel cures (dries) during dry season 3. Fire frequency: Frequent fires -\> lower woody component 4. Grazing pressure: Reduces available fuel Spatial heterogeneity: Fuel patchy due to: - Termite mounds (fuel-free zones) - Tree shade (reduced grass growth) - Grazing concentration around water - Previous fire scars #### Climate-Dependent Savanna Fuel Loads {#sec-fuel-savanna-climate} Tropical Savannas (500-1500 mm rainfall): - High grass productivity - Distinct wet/dry seasons - Fuel curing rapid - M~B~: 6.8-9.0 t DM ha⁻¹ Temperate Grasslands (250-750 mm rainfall): - Lower productivity than tropical - Variable seasonality - Mixed grass/forb composition - M~B~: 13.4 t DM ha⁻¹ (includes crop residues) Woody Savannas vs. Open Savannas: - Woody component adds coarse fuels, but canopy reduces grass fuel - Net effect: Similar M~B~ to open savanna - Combustion factor varies more than fuel load #### Seasonal Dynamics {#sec-fuel-seasonal-dynamics} Early Dry Season: - Fuel partially cured (50-70% dry matter) - Lower combustion efficiency - Lower C~f~: 0.29-0.31 - Total consumption: 2.1-2.6 t DM ha⁻¹ Late Dry Season: - Fuel fully cured (\>90% dry matter) - Complete combustion of grass layer - Higher C~f~: 0.51-0.83 - Total consumption: 4.6-5.6 t DM ha⁻¹ ### Organic Soil Fuel Loads {#sec-fuel-organic} #### Peat Fuel Characteristics {#sec-fuel-peat-chars} Definition: Organic soils (Histosols) with \>30 cm organic layer Fuel properties: - High carbon content: 50-60% C by dry weight - Variable bulk density: 0.05-0.20 g cm⁻³ - High moisture retention: 200-1000% (dry weight basis) - Smoldering combustion: Can burn for months Fire behavior: - Surface fires: Consume litter and upper peat (10-30 cm) - Deep burns: Consume entire peat layer to mineral soil (\>100 cm) - Drainage amplifies risk: Drained peat extremely flammable #### Fuel Load Calculation {#sec-fuel-peat-calculation} IPCC Wetlands Supplement Approach: $$ M_B = D_{burn} \times BD \times 10 $$ Where: - $D_{burn}$ = Depth of burn (cm) - $BD$ = Bulk density of peat (g cm⁻³) - 10 = Conversion factor (cm = m, g cm⁻³ = t ha⁻¹) IPCC Default (Southeast Asia): - Depth of burn: 41 cm (range: 18-65 cm) - Bulk density: 0.10 g cm⁻³ (range: 0.05-0.15) - M~B~: 41.2 t DM ha⁻¹ High uncertainty: Coefficient of variation \~50% ------------------------------------------------------------ ## Combustion Factor (C~f~) {#sec-combustion-factor} ### Definition and Interpretation {#sec-cf-definition} Combustion Factor (C~f~) = Proportion of available fuel actually consumed by fire $$ C_f = \frac{\text{Mass of fuel consumed}}{\text{Mass of fuel available}} \quad (0 \leq C_f \leq 1) $$ Determinants: 1. Fuel moisture: Higher moisture = lower C~f~ 2. Fire intensity: Hotter fires → higher C~f~ 3. Fire type: Surface vs. crown fire 4. Fuel particle size: Fine fuels (\< 0.6 cm) burn more completely 5. Weather conditions: Wind, humidity, temperature ### Forest Combustion Factors {#sec-cf-forest} Low C~f~ for Humid Tropical Forests (0.36): - High ambient moisture - Rapid fuel accumulation = green fuels mixed with dead - Incomplete combustion of large woody debris - Char and partially burned material remain Moderate C~f~ for Other Forests (0.45-0.47): - Drier climates = lower fuel moisture - More complete consumption of surface fuels - Variability based on fire severity: - Surface fires: C~f~ \~ 0.20-0.40 - Crown fires: C~f~ \~ 0.60-0.80 - Default represents average across severities ### Savanna Combustion Factors {#sec-cf-savanna} Early Dry Season (0.29-0.31): - Fuel partially cured - High moisture content in live grass (30-50%) - Incomplete combustion - Patchy fire spread - Much grass left standing Late Dry Season (0.51-0.83): - Fuel fully cured (\<10% moisture) - Complete consumption of grass layer - Only woody stems and char remain - Higher C~f~ for grasslands vs. woody savannas: - Grassland: C~f~ = 0.83 (almost complete) - Woody Savanna: C~f~ = 0.51 (trees create shade/barriers) Physical explanation: $$ C_f \propto \frac{1}{\text{Fuel Moisture}} \times \text{Fire Intensity} $$ Late season = Lower moisture × Higher intensity = Higher C~f~ ### Organic Soil Combustion Factors {#sec-cf-organic} Complete Consumption (C~f~ = 1.0): - Once peat ignites, burns to completion within affected depth - Smoldering combustion eliminates unburned fuel - Char production minimal - Caveat: C~f~ = 1.0 applies to depth of burn, not total peat column - If burn depth = 41 cm, that 41 cm is fully consumed - Deeper peat layers remain intact (if fire extinguished) ------------------------------------------------------------ ## Applying Fuel Consumption Parameters {#sec-fuel-application} ### Workflow: Vegetation Type to Fuel Values {#sec-fuel-workflow} ```{r} #| eval: true #| label: fuel-lookup-table #| code-summary: "Create IPCC fuel consumption lookup table" library(tidyverse) # IPCC 2019 Table 2.4 - Default fuel consumption values fuel_consumption_table <- tribble( ~fire_category, ~vegetation_type, ~climate_zone, ~season, ~MB_tDM_ha, ~Cf, ~MB_Cf, # Forest Fires "Forest", "Humid_Tropical_Forest", "Tropical_Moist", NA, 119.6, 0.36, 43.1, "Forest", "Other_Forest", "Boreal", NA, 41.0, 0.45, 18.5, "Forest", "Other_Forest", "Temperate", NA, 50.4, 0.47, 23.7, "Forest", "Other_Forest", "Tropical", NA, 69.4, 0.47, 32.6, # Savanna Fires "Savanna", "Savanna_Woodland", "All", "Early", 9.0, 0.29, 2.6, "Savanna", "Savanna_Woodland", "All", "Late", 9.0, 0.51, 4.6, "Savanna", "Savanna_Grassland", "All", "Early", 6.8, 0.31, 2.1, "Savanna", "Savanna_Grassland", "All", "Late", 6.8, 0.83, 5.6, "Savanna", "Other_Grassland", "All", NA, 13.4, 0.52, 7.0, # Organic Soil Fires "Organic_Soil", "Tropical_Peatland", "Tropical", NA, 41.2, 1.0, 41.2 ) print(fuel_consumption_table) # Save for use in emissions calculation write.csv( fuel_consumption_table, "IPCC_fuel_consumption_defaults.csv", row.names = FALSE ) ``` ------------------------------------------------------------ ## GEE Initialisation {#sec-fuel-gee-init} ```{r} #| label: fuel-gee-setup #| warning: false #| message: false #| error: false # Point environment to your configured python reticulate::use_python("/opt/local/bin/python", required=T) # Initialise Earth Engine ee <- reticulate::import("ee") ee$Initialize(project = "murphys-deforisk") cat("GEE initialised\n") # Helper: visualize EE tile in tmap/leaflet ee_tile_url <- function(ee_image, vis_params) { ee_image$getMapId(vis_params)$tile_fetcher$url_format } # Helper: EE FeatureCollection => sf ee_to_sf <- function(fc) { geojsonsf::geojson_sf( jsonlite::toJSON(fc$getInfo(), auto_unbox = TRUE)) } # Load AOI boundaries from FAO GAUL aoi_country_ee <- ee$FeatureCollection("FAO/GAUL/2015/level0")$ filter(ee$Filter$eq("ADM0_NAME", "Honduras")) aoi_states_ee <- ee$FeatureCollection("FAO/GAUL/2015/level1")$ filter(ee$Filter$eq("ADM0_NAME", "Honduras")) aoi_country_sf <- ee_to_sf(aoi_country_ee) aoi_states_sf <- ee_to_sf(aoi_states_ee) cat("AOI loaded:", aoi_country_ee$size()$getInfo(), "country,", aoi_states_ee$size()$getInfo(), "states\n") ``` ### Upstream Inputs from Chapters 1-2 {#sec-fuel-upstream} ```{r} #| label: fuel-upstream-inputs #| code-summary: "Load burned area (Ch1) and vegetation masks (Ch2)" # --- Chapter 1: Burned area (MCD64A1) --- mcd64 <- ee$ImageCollection("MODIS/061/MCD64A1") burned_2010 <- mcd64$ filterBounds(aoi_country_ee$geometry())$ filterDate("2010-01-01", "2010-12-31") # FAOSTAT quality filter (matching Ch1) faostat_filter <- function(image) { unc <- image$select("Uncertainty") burn <- image$select("BurnDate") mask <- unc$lt(20L)$And(burn$gt(0L)) image$updateMask(mask) } # Annual burned mask for 2010 annual_burned_2010 <- burned_2010$ map(faostat_filter)$ select("BurnDate")$max()$gt(0L)$selfMask() # --- Chapter 2: Land cover & vegetation masks --- lc_2010 <- ee$ImageCollection("MODIS/061/MCD12Q1")$ select("LC_Type1")$ filterDate("2010-01-01", "2010-12-31")$ first()$ clip(aoi_country_ee$geometry()) # Forest mask (IGBP 1-5) forest_mask <- lc_2010$gte(1L)$And(lc_2010$lte(5L)) # Savanna/grassland mask (IGBP 6-10) savanna_mask <- lc_2010$gte(6L)$And(lc_2010$lte(10L)) # Climate-based humid tropical delineation bio <- ee$Image("WORLDCLIM/V1/BIO") precip <- bio$select("bio12") temp <- bio$select("bio01") tropical <- temp$gt(180L) moist <- precip$gt(1500L) humid_tropical_mask <- tropical$And(moist) # Forest sub-types humid_tropical_forest <- forest_mask$And(humid_tropical_mask) other_forest <- forest_mask$And(humid_tropical_mask$Not()) # Pixel area helper pixelArea_ha <- ee$Image$pixelArea()$divide(10000L) cat("Upstream inputs loaded: MCD64A1 + MCD12Q1 + WorldClim\n") ``` ------------------------------------------------------------ ## Spatial Implementation in Google Earth Engine {#sec-fuel-gee} ### Creating Fuel Consumption Rasters {#sec-fuel-gee-rasters} ```{r} #| label: fuel-gee-rasters #| code-summary: "Generate spatially-explicit fuel consumption maps" # Climate zone masks boreal <- temp$lt(0L) temperate <- temp$gte(0L)$And(temp$lte(180L)) # Create fuel consumption image (M_B × C_f in t DM ha⁻¹) fuel_consumption <- ee$Image(0)$rename("fuel_consumption") # 1. Humid Tropical Forest (M_B × C_f = 43.1) htf_mask <- forest_mask$And(humid_tropical_mask) fuel_consumption <- fuel_consumption$where(htf_mask, 43.1) # 2. Other Forest - Boreal (M_B × C_f = 18.5) forest_boreal <- forest_mask$And(boreal)$And(humid_tropical_mask$Not()) fuel_consumption <- fuel_consumption$where(forest_boreal, 18.5) # 3. Other Forest - Temperate (M_B × C_f = 23.7) forest_temperate <- forest_mask$And(temperate)$And(humid_tropical_mask$Not()) fuel_consumption <- fuel_consumption$where(forest_temperate, 23.7) # 4. Other Forest - Tropical Dry (M_B × C_f = 32.6) forest_tropical <- forest_mask$And(tropical)$And(humid_tropical_mask$Not()) fuel_consumption <- fuel_consumption$where(forest_tropical, 32.6) # 5. Savanna/Grassland (average = 7.0 t DM ha⁻¹) fuel_consumption <- fuel_consumption$where(savanna_mask, 7.0) # Apply to burned areas only fuel_burned <- fuel_consumption$ updateMask(annual_burned_2010)$ clip(aoi_country_ee$geometry()) # Visualize fuel_vis <- list( min = 0L, max = 50L, palette = c("yellow", "orange", "red", "darkred") ) tmap::tmap_mode("view") tmap::tm_shape(aoi_country_sf) + tmap::tm_borders(col = "white", lwd = 1.5) + tmap::tm_tiles( ee_tile_url(fuel_burned, fuel_vis), group = "Fuel Consumption (t DM ha⁻¹)" ) + tmap::tm_add_legend( type = "polygons", fill = c("yellow", "orange", "red", "darkred"), labels = c("Savanna (~7)", "Temperate Forest (~24)", "Tropical Forest (~33)", "Humid Tropical (~43)"), title = "Fuel Consumption (t DM ha⁻¹)" ) + tmap::tm_scalebar(position = c("RIGHT", "BOTTOM"), text.size = 0.5) + tmap::tm_basemap("Esri.WorldImagery") ``` ### Seasonal Stratification for Savannas {#sec-fuel-gee-seasonal} ```{r} #| label: fuel-seasonal-stratification #| code-summary: "Apply seasonal fuel consumption to savannas" # Extract burn month from MCD64A1 extractBurnMonth <- function(mcd64_image) { burn_date <- mcd64_image$select("BurnDate") img_date <- ee$Date(mcd64_image$get("system:time_start")) month <- img_date$get("month") month_band <- ee$Image$constant(month)$ updateMask(burn_date$gt(0L))$ rename("burn_month")$ toInt() burn_date$addBands(month_band) } # Apply to 2010 collection (quality-filtered) mcd64_with_months <- burned_2010$ map(faostat_filter)$ map(extractBurnMonth) # Composite: latest month if pixel burned multiple times month_composite <- mcd64_with_months$ select("burn_month")$ max() # Honduras dry season (Central America) # Early dry: Feb-Mar (months 2-3) # Late dry: Apr-May (months 4-5) early_season <- month_composite$gte(2L)$And(month_composite$lte(3L)) late_season <- month_composite$gte(4L)$And(month_composite$lte(5L)) # Savanna fire subsets savanna_fires_2010 <- annual_burned_2010$updateMask(savanna_mask) # Woody savanna (IGBP class 8) woody_savanna <- lc_2010$eq(8L) # Grassland/savanna (IGBP classes 9-10) grassland <- lc_2010$gte(9L)$And(lc_2010$lte(10L)) # Create seasonal fuel consumption fuel_seasonal <- ee$Image(0)$rename("fuel_seasonal") fuel_seasonal <- fuel_seasonal$ where(woody_savanna$And(early_season), 2.6)$ where(woody_savanna$And(late_season), 4.6)$ where(grassland$And(early_season), 2.1)$ where(grassland$And(late_season), 5.6) # Mask to burned savanna areas fuel_seasonal_burned <- fuel_seasonal$ updateMask(savanna_fires_2010)$ clip(aoi_country_ee$geometry()) # Visualize tmap::tm_shape(aoi_country_sf) + tmap::tm_borders(col = "white", lwd = 1.5) + tmap::tm_tiles( ee_tile_url(fuel_seasonal_burned, list(min = 0L, max = 6L, palette = c("lightyellow", "yellow", "orange"))), group = "Savanna Fuel Consumption (Seasonal)" ) + tmap::tm_add_legend( type = "polygons", fill = c("lightyellow", "yellow", "orange"), labels = c("Early grassland (2.1)", "Early woody (2.6) / Late woody (4.6)", "Late grassland (5.6)"), title = "Savanna Fuel (t DM ha⁻¹)" ) + tmap::tm_scalebar(position = c("RIGHT", "BOTTOM"), text.size = 0.5) + tmap::tm_basemap("Esri.WorldImagery") ``` ### Calculating Total Biomass Burned {#sec-fuel-gee-total} ```{r} #| label: fuel-total-biomass #| code-summary: "Calculate total biomass burned — Honduras 2010" # Biomass burned per pixel (t DM) = fuel_consumption (t ha⁻¹) × area (ha) biomass_per_pixel <- fuel_burned$multiply(pixelArea_ha) # Forest biomass forest_biomass <- biomass_per_pixel$ updateMask(forest_mask)$ reduceRegion( reducer = ee$Reducer$sum(), geometry = aoi_country_ee$geometry(), scale = 500L, maxPixels = 1e9 )$getInfo() # Savanna biomass savanna_biomass <- biomass_per_pixel$ updateMask(savanna_mask)$ reduceRegion( reducer = ee$Reducer$sum(), geometry = aoi_country_ee$geometry(), scale = 500L, maxPixels = 1e9 )$getInfo() forest_t <- forest_biomass[["fuel_consumption"]] %||% 0 savanna_t <- savanna_biomass[["fuel_consumption"]] %||% 0 cat("Honduras 2010 — Total Biomass Burned\n") cat(" Forest biomass burned: ", round(forest_t / 1000, 1), "Gg DM\n") cat(" Savanna biomass burned:", round(savanna_t / 1000, 1), "Gg DM\n") cat(" Total: ", round((forest_t + savanna_t) / 1000, 1), "Gg DM\n") ``` ### Matching Vegetation Classifications to Fuel Values {#sec-fuel-matching} ```{r} #| label: fuel-value-assignment #| code-summary: "Assign fuel consumption values to classified fires — Honduras 2010" # Burned area by vegetation type (from Ch2 overlay) htf_fires <- annual_burned_2010$updateMask(humid_tropical_forest) of_fires <- annual_burned_2010$updateMask(other_forest) sav_fires <- annual_burned_2010$updateMask(savanna_mask) # Calculate areas (ha) htf_area <- htf_fires$multiply(pixelArea_ha)$reduceRegion( reducer = ee$Reducer$sum(), geometry = aoi_country_ee$geometry(), scale = 500L, maxPixels = 1e9)$getInfo() of_area <- of_fires$multiply(pixelArea_ha)$reduceRegion( reducer = ee$Reducer$sum(), geometry = aoi_country_ee$geometry(), scale = 500L, maxPixels = 1e9)$getInfo() sav_area <- sav_fires$multiply(pixelArea_ha)$reduceRegion( reducer = ee$Reducer$sum(), geometry = aoi_country_ee$geometry(), scale = 500L, maxPixels = 1e9)$getInfo() htf_ha <- htf_area[["BurnDate"]] %||% 0 of_ha <- of_area[["BurnDate"]] %||% 0 sav_ha <- sav_area[["BurnDate"]] %||% 0 # Apply IPCC default fuel consumption honduras_fuel <- tribble( ~fire_category, ~vegetation_type, ~area_ha, ~MB_tDM_ha, ~Cf, ~MB_Cf, "Forest", "Humid_Tropical_Forest", round(htf_ha, 1), 119.6, 0.36, 43.1, "Forest", "Other_Forest", round(of_ha, 1), 69.4, 0.47, 32.6, "Savanna", "Savanna_Mixed", round(sav_ha, 1), 7.9, 0.46, 3.6 ) %>% mutate( biomass_burned_tDM = area_ha * MB_Cf, biomass_burned_Gg = biomass_burned_tDM / 1000 ) print(honduras_fuel) cat("\nHonduras 2010 — Summary\n") cat(" Total burned area: ", sum(honduras_fuel$area_ha), "ha\n") cat(" Total biomass burned: ", round(sum(honduras_fuel$biomass_burned_Gg), 1), "Gg DM\n") ``` ------------------------------------------------------------ ## Uncertainty in Fuel Consumption Estimates {#sec-fuel-uncertainty} ### Sources of Uncertainty {#sec-fuel-uncertainty-sources} 1\. IPCC Default Value Uncertainty: IPCC provides 95% confidence intervals for all default values: | Parameter | Relative Uncertainty | |----------------------|----------------------| | M~B~ (Forest) | ±40-60% | | M~B~ (Savanna) | ±60-90% | | C~f~ (Forest) | ±20-30% | | C~f~ (Savanna) | ±30-40% | | M~B~×C~f~ (combined) | ±50-70% | 2\. Spatial Extrapolation Uncertainty: - IPCC values derived from limited field studies - Geographic bias toward well-studied regions (North America, Australia) - Under-representation of tropical systems 3\. Fire Severity Variability: - Default values represent averages across fire severity spectrum - Individual fires may have 2-3× higher/lower consumption - No severity stratification in Tier 1 4\. Temporal Variability: - Fuel accumulation varies with: - Time since last fire - Precipitation anomalies (ENSO, droughts) - Forest management (fuel reduction treatments) - Tier 1 assumes constant fuel loads ### Quantifying Combined Uncertainty {#sec-fuel-uncertainty-quantify} Following IPCC Error Propagation (Chapter 3, Equation 3.1): $$ U_{total}^2 = U_A^2 + U_{MB}^2 + U_{Cf}^2 + U_{Gef}^2 $$ For Tier 1 Forest Fires: - $U_A$ = ±30% (from Chapter 1 - burned area) - $U_{MB}$ = ±50% (IPCC Table 2.4) - $U_{Cf}$ = ±25% (IPCC Table 2.4) - $U_{Gef}$ = ±30% (Chapter 4 - emission factors) Combined uncertainty (quadrature): $$ U_{total} = \sqrt{30^2 + 50^2 + 25^2 + 30^2} = \sqrt{4350} \approx 66\% $$ For Tier 1 Savanna Fires: - $U_A$ = ±30% - $U_{MB}$ = ±80% (higher variability) - $U_{Cf}$ = ±35% - $U_{Gef}$ = ±30% Combined uncertainty: $$ U_{total} = \sqrt{30^2 + 80^2 + 35^2 + 30^2} = \sqrt{9925} \approx 100\% $$ Interpretation: Tier 1 uncertainties are large but expected given coarse default values. Tier 2/3 approaches reduce uncertainty through country-specific data. ### Sensitivity Analysis {#sec-fuel-sensitivity} ```{r} #| eval: true #| label: fuel-sensitivity-analysis #| code-summary: "Assess sensitivity to fuel consumption parameters" library(tidyverse) # Define parameter ranges (from IPCC confidence intervals) # Honduras is predominantly humid tropical forest fuel_params <- expand.grid( MB_forest = c(119.6 - 52.4, 119.6, 119.6 + 52.4), Cf_forest = c(0.36 - 0.09, 0.36, 0.36 + 0.09), area_ha = round(htf_ha, 0) ) # Calculate biomass burned for each parameter combination fuel_params <- fuel_params %>% mutate( biomass_tDM = area_ha * MB_forest * Cf_forest, biomass_Gg = biomass_tDM / 1000, scenario = paste0("MB=", round(MB_forest, 1), ", Cf=", round(Cf_forest, 2)) ) # Visualize sensitivity ggplot(fuel_params, aes(x = MB_forest, y = Cf_forest, fill = biomass_Gg)) + geom_tile() + geom_text(aes(label = round(biomass_Gg, 0)), color = "white", fontface = "bold") + scale_fill_viridis_c(name = "Biomass\nBurned (Gg)") + labs( title = "Sensitivity of Biomass Burned to Fuel Parameters", subtitle = paste("Honduras Humid Tropical Forest Fires (", round(htf_ha, 0), "ha)"), x = "Fuel Load MB (t DM ha⁻¹)", y = "Combustion Factor Cf" ) + theme_minimal() # Calculate coefficient of variation cv_biomass <- sd(fuel_params$biomass_Gg) / mean(fuel_params$biomass_Gg) * 100 cat("Coefficient of Variation:", round(cv_biomass, 1), "%\n") ``` ------------------------------------------------------------ ## Advanced Topics {#sec-fuel-advanced} ### Country-Specific Fuel Data (Tier 2) {#sec-fuel-tier2} Moving beyond defaults: When available, countries can use: 1. National forest inventory data: Standing biomass, litter depths 2. Fire behavior models: BEHAVE, FuelCalc, FCCS 3. Field measurements: Post-fire fuel consumption studies 4. Remote sensing: LiDAR for 3D fuel structure Example: Honduras: ```{r} #| eval: true #| label: fuel-tier2-example #| code-summary: "Using country-specific fuel data" # Hypothetical: Honduras National Forest Inventory fuel loads honduras_fuel_data <- tribble( ~forest_type, ~MB_tDM_ha, ~Cf, ~source, "Pinus caribaea (lowland pine)", 72.5, 0.42, "ICF National Inventory 2018", "Broadleaf humid forest", 125.3, 0.34, "SIGMOF field plots 2019", "Dry broadleaf / pine-oak", 55.8, 0.45, "La Mosquitia fire studies 2017" ) # These values would replace IPCC defaults for Honduran forests # Result: Reduced uncertainty (~±25% over ±50%) print(honduras_fuel_data) ``` ### Tier 2+ Fire Severity Stratification (Pending) {#sec-fuel-severity} Fire severity classes: | Severity | dNBR Range | C~f~ Multiplier | |----------|-------------|-----------------| | Unburned | \< 0.1 | 0 | | Low | 0.1 - 0.27 | 0.5× | | Moderate | 0.27 - 0.44 | 1.0× (default) | | High | \> 0.44 | 1.5× | ```{r} #| eval: false #| label: fuel-severity-stratification #| code-summary: "Stratify fuel consumption by fire severity" # Calculate dNBR from Landsat or Sentinel-2 to adjust Cf # Pre-fire composite (Jan-Feb 2010) prefire <- ee$ImageCollection("LANDSAT/LT05/C02/T1_L2")$ filterBounds(aoi_country_ee$geometry())$ filterDate("2010-01-01", "2010-02-28")$ median() # Post-fire composite (Jun-Jul 2010) postfire <- ee$ImageCollection("LANDSAT/LT05/C02/T1_L2")$ filterBounds(aoi_country_ee$geometry())$ filterDate("2010-06-01", "2010-07-31")$ median() # Calculate NBR calculateNBR <- function(img) { nir <- img$select("SR_B4") swir <- img$select("SR_B7") nir$subtract(swir)$divide(nir$add(swir))$rename("NBR") } nbr_pre <- calculateNBR(prefire) nbr_post <- calculateNBR(postfire) dnbr <- nbr_pre$subtract(nbr_post) # Define severity classes low_severity <- dnbr$gte(0.1)$And(dnbr$lt(0.27)) moderate_severity <- dnbr$gte(0.27)$And(dnbr$lt(0.44)) high_severity <- dnbr$gte(0.44) # Adjust combustion factor for humid tropical forest Cf_default <- 0.36 Cf_adjusted <- ee$Image(Cf_default)$ where(low_severity, Cf_default * 0.5)$ where(high_severity, Cf_default * 1.5) # Recalculate fuel consumption MB <- 119.6 fuel_consumption_adjusted <- ee$Image(MB)$multiply(Cf_adjusted) ``` ### Fuel Moisture Effects {#sec-fuel-moisture} Fuel moisture reduces combustion efficiency: $$ C_f = C_{f,dry} \times e^{-k \times MC} $$ Where: - $C_{f,dry}$ = Combustion factor at 0% moisture - $MC$ = Moisture content (%) - $k$ = Decay constant (\~0.02 for grass, \~0.01 for wood) ```{r} #| eval: false #| label: fuel-moisture-correction #| code-summary: "Adjust Cf for fuel moisture content" # Load ERA5 relative humidity or precipitation data precip_monthly <- ee$ImageCollection("ECMWF/ERA5/MONTHLY")$ select("total_precipitation") # Get precipitation in month of fire # (Would need to match with burn dates from Chapter 1) # For demonstration: use annual average precip_annual <- precip_monthly$ filterDate("2010-01-01", "2010-12-31")$ mean() # Convert to fuel moisture proxy (simplified) # Assume: High precip → High moisture → Low Cf # (Actual relationship requires mechanistic model) # This is TIER 3 level - requires significant data and validation ``` ------------------------------------------------------------ ## Integrated Emissions Calculations {#sec-fuel-integration} ### Output Formats {#sec-fuel-to-chapter4} Required data structure: ```{r} #| eval: true #| label: fuel-output-structure #| code-summary: "Format fuel consumption data for emissions calculation" # Create comprehensive fuel consumption table fuel_output <- tribble( ~country, ~year, ~fire_category, ~vegetation_type, ~climate_zone, ~season, ~area_ha, ~MB_tDM_ha, ~Cf, ~biomass_burned_Gg, "Honduras", 2010, "Forest", "Humid_Tropical_Forest", "Tropical_Moist", NA, round(htf_ha, 0), 119.6, 0.36, round(htf_ha * 43.1 / 1000, 1), "Honduras", 2010, "Forest", "Other_Forest", "Tropical", NA, round(of_ha, 0), 69.4, 0.47, round(of_ha * 32.6 / 1000, 1), "Honduras", 2010, "Savanna", "Savanna_Mixed", "Tropical", "Late", round(sav_ha, 0), 7.9, 0.46, round(sav_ha * 3.6 / 1000, 1) ) # Save for Chapter 4 write.csv( fuel_output, "Honduras_FuelConsumption_2010.csv", row.names = FALSE ) print(fuel_output) ``` ### QC Checklist {#sec-fuel-qc-checklist} Pre-calculation: - Vegetation classifications complete (from Chapter 2) - Climate zones assigned correctly - Seasonal stratification applied (savannas only) - IPCC fuel consumption table loaded correctly Calculation: 1. Fuel values matched to vegetation types 2. Units consistent (t DM ha⁻¹, Gg) 3. Biomass = Area × M~B~ × C~f~ 4. No missing values (all vegetation types have fuel values) Post-calculation: 1. Biomass totals reasonable (order of magnitude check) 2. Sum of vegetation-specific biomass = total biomass 3. Uncertainty quantified and documented 4. Results comparable to previous years (if available) Validation: - Compare with published values (if available) - Check extreme values (outliers) - Sensitivity analysis completed - Documentation complete ------------------------------------------------------------ ## Notes and Next Steps {#sec-fuel-summary} ### Key Takeaways {#sec-fuel-takeaways} 1. Fuel consumption (M~B~ × C~f~) varies by: - Vegetation type: Forest (18-43 t ha⁻¹) vs. Savanna (2-7 t ha⁻¹) - Climate zone: Boreal \< Temperate \< Tropical \< Humid Tropical - Season: Early (lower) vs. Late (higher) dry season for savannas 2. IPCC Tier 1 defaults provide globally consistent estimates but have high uncertainty (±50-100%) 3. Combustion factor (C~f~) represents proportion of fuel consumed, ranging from 0.29 (early season savanna) to 1.0 (organic soils) 4. Total biomass burned calculated as: Area (ha) × M~B~ (t ha⁻¹) × C~f~ 5. Tier 2/3 improvements possible through country-specific data, fire severity mapping, and fuel moisture modeling ### Chapter 4: {#sec-fuel-to-chapter4-summary} 1. Apply IPCC emission factors (G~ef~) by gas (CO₂, CH₄, N₂O) 2. Calculate trace gas emissions: Biomass Burned × G~ef~ 3. Report emissions by IPCC category (Forest, Savanna, Organic Soils) 4. Convert to CO₂-equivalents for aggregate reporting Critical linkage: Each fire event carries forward: - Biomass burned (Gg DM) by vegetation type - Fire category (Forest/Savanna/Organic Soil) for emission factor selection - Climate zone (for organic soil CH₄ emission factors) - Uncertainty estimates for error propagation ------------------------------------------------------------

Overview

Environment Setup

3.0.1 Component in Equation 2.27

3.0.2 IPCC Default Values - Quick Reference

3.1 Fuel Available for Combustion (MB)

3.1.1 Forest Fuel Loads

3.1.1.1 Components of Forest Fuel

3.1.1.2 Climate-Dependent Forest Fuel Loads

3.1.2 Savanna Fuel Loads

3.1.2.1 Savanna Fuel Characteristics

3.1.2.2 Climate-Dependent Savanna Fuel Loads

3.1.2.3 Seasonal Dynamics

3.1.3 Organic Soil Fuel Loads

3.1.3.1 Peat Fuel Characteristics

3.1.3.2 Fuel Load Calculation

3.2 Combustion Factor (Cf)

3.2.1 Definition and Interpretation

3.2.2 Forest Combustion Factors

3.2.3 Savanna Combustion Factors

3.2.4 Organic Soil Combustion Factors

3.3 Applying Fuel Consumption Parameters

3.3.1 Workflow: Vegetation Type to Fuel Values

3.4 GEE Initialisation

3.4.1 Upstream Inputs from Chapters 1-2

3.5 Spatial Implementation in Google Earth Engine

3.5.1 Creating Fuel Consumption Rasters

3.5.2 Seasonal Stratification for Savannas

3.5.3 Calculating Total Biomass Burned

3.5.4 Matching Vegetation Classifications to Fuel Values

3.6 Uncertainty in Fuel Consumption Estimates

3.6.1 Sources of Uncertainty

3.6.2 Quantifying Combined Uncertainty

3.6.3 Sensitivity Analysis

3.7 Advanced Topics

3.7.1 Country-Specific Fuel Data (Tier 2)

3.7.2 Tier 2+ Fire Severity Stratification (Pending)

3.7.3 Fuel Moisture Effects

3.8 Integrated Emissions Calculations

3.8.1 Output Formats

3.8.2 QC Checklist

3.9 Notes and Next Steps

3.9.1 Key Takeaways

3.9.2 Chapter 4:

3.1 Fuel Available for Combustion (M_B)

3.2 Combustion Factor (C_f)