SAGE-IGP

SAGE-IGP (Survey Constraints on FRP-based AGricultural Fire Emissions in the Indo-Gangetic Plain): agricultural fire emissions in north India

Input Datasets

We use the following datasets:

• MODIS/MxD14A1 Fire Radiative Power, 1km
• MODIS/MCD14ML Active Fires
• VIIRS/VNP14IMGML Active Fires
• MODIS/MOD09GA Daily Surface Reflectance, 500m
• MODIS/MCD12Q1 Land Cover, 500 m

Workflow

Google Earth Engine

Preprocess input datasets at state-level

GEE/Preprocess/ >

1. MxD14A1_FRP.js: export daily state-level MODIS FRP
2. MxD14A1_FireMask.js: export a MODIS fire mask to EE assets based on the aggregate of all fire detections
3. VNP14IMGML_FRPboost.js: export daily state-level VIIRS FRP boost relative to MODIS/Aqua FRP
4. MCD14ML_SR.js: export surface reflectance values associated with each MODIS fire detection for degree of cloudiness/haziness estimation
5. MCD14ML_SR_fill.js: export surface reflectance values associated with each MODIS fire detection for degree of cloudiness/haziness estimation (back-fill for months with no fires)

Preprocess input datasets on a 0.25°x0.25° grid

GEE/GriddedFRP/ >

6. MxD14A1_FRP_Grid.js: export daily state-level MODIS FRP on a 0.25°x0.25° grid
7. VNP14IMGML_FRPboost_Grid.js: export daily state-level VIIRS FRP boost relative to MODIS/Aqua FRP on a 0.25° x 0.25° grid

R

Preprocess output tables from GEE

R/processGEE/ >

1. MCD14ML_SR_process.R: preprocess MCD14ML SR files from EE
2. MODIS_FRP_Grid_process.R: preprocess MxD14A1 FRP files from EE
3. VIIRS_FRPboost_Grid_process.R: preprocess VIIRS FRP boost files from EE

Adjusted FRP algorithm and construction of SAGE-IGP Inventory

R/adjFRP/ >

4. cloud_gap_cutoff.R: cloud/haze gap cutoff values
5. viirs_boost_coef.R: calculate VIIRS scaling factor for each state
6. adjFRP_T1.R: adjusted FRP for Tier 1 states - Punjab, Haryana
7. adjFRP_T2.R: adjusted FRP for Tier 2 states - UP, Bihar
8. adjFRP_T3.R: adjusted FRP for Tier 3 states - Rajasthan
9. adjFRP_DM.R: convert adjusted FRP to dry matter
10. adjFRP_DMgridST.R: grid state-level dry matter at 0.25°x0.25° resolution
11. adjFRP_DMgrid.R: combine gridded state-level dry matter
12. adjFRP_DMaer.R: combine gridded state-level dry matter, aerosols
13. nc_adjFRP_DM.R: make annual netCDF files for the SAGE-IGP inventory

SAGE-IGP Dataset

Liu T., L.J. Mickley, S. Singh, M. Jain, R.S. DeFries, and M.E. Marlier (2020). SAGE-IGP agricultural fire emissions in north India, https://doi.org/10.7910/DVN/JUMXOL, Harvard Dataverse, V1

Usage Notes

• Use agricultural emissions factors to convert dry matter (DM) to other chemical species, e.g. from Andreae (2019, ACP)
• Use DMaer for aerosol species and DM for all other chemical species

# ======================================
# Example R Script for Reading SAGE-IGP
# ======================================

library(ncdf4); library(raster)

# --------------------------
# Convert netCDF to raster
# --------------------------
# read SAGE-IGP netCDF for 2017
nc <- nc_open("adjFRP_Inv/nc/SAGE-IGP_daily_2017.nc")

# retrieve variables from netCDF
DMdaily <- ncvar_get(nc,"DM")

# total DM in 2017 from Sep-Dec
DMtotal <- apply(DMdaily,c(1,2),sum)

# read lat, lon
lat <- ncvar_get(nc,"lat")
lon <- ncvar_get(nc,"lon")

# extent of SAGE-IGP bounds
regionExtent <- extent(c(72,89,23,33))

# convert 'DMtotal' (a matrix) to a raster
DMtotal_ras <- raster(t(DMtotal)[rev(order(lat)),]) / 1e9 # convert from kg to Tg per grid cell
extent(DMtotal_ras) <- regionExtent # set extent of raster to regionExtent
crs(DMtotal_ras) <- crs(raster()) # EPSG:4326 - default lat/lon projection

# a simple plot of total DM in 2017
plot(DMtotal_ras)

# raster where each grid cell represents its area in sq. meters
area_m2 <- raster::area(DMtotal_ras) * 1e6

# convert Nov 1, 2017 emissions from kg to kg/m2/s
DMdaily_ras <- raster(t(DMdaily[,,62])[rev(order(lat)),])
DMdaily_ras <- DMtotal_ras / area_m2 / (24*24*60)
extent(DMdaily_ras) <- regionExtent
crs(DMdaily_ras) <- crs(raster())

# --------------------------------------
# Convert DM to other chemical species
# --------------------------------------
# for aerosol species, e.g. OC, BC, PM25, use 'DMaer'
# for all other species, use 'DM'

# retrieve variables from netCDF
DMdailyAer <- ncvar_get(nc,"DMaer")

# define emissions factors as g / kg DM, from Andreae (2019, ACP)
efs_andreae <- data.frame(OC=4.9,BC=0.42)

OCdaily <- DMdailyAer * efs_andreae$OC / 1e9 # daily OC, in Gg
BCdaily <- DMdailyAer * efs_andreae$BC / 1e9 # daily BC, in Gg

GEOS-Chem - HEMCO Compatibility

SAGE-IGP can be combined with the GFEDv4s inventory by replacing GFEDv4s with SAGE-IGP agricultural fire emissions over north India. The resulting GFEDv4s+SAGE-IGP fire emissions follows the netCDF format of the original GFEDv4s emissions in HEMCO. Please contact me via email if you're interested in running GEOS-Chem with these files.

Issues

• The original script adjFRP_T1.R used to produce Punjab/Haryana emissions in SAGE-IGP was accidentally overwritten and cannot be recovered. The updated adjFRP_T1.R yields the same mean budget of dry matter burned overall, but there may be slight differences in annual and daily variability.

Publication

Liu T., L.J. Mickley, S. Singh, M. Jain, R.S. DeFries, and M.E. Marlier (2020). Crop residue burning practices across north India inferred from household survey data: bridging gaps in satellite observations. Atmos. Environ. X 8, 100091. https://doi.org/10.1016/j.aeaoa.2020.100091