GFED¶

Global Fire Emissions Database version 5 (GFED5.1 / GFED5.1NRT — global fire emissions at 0.25° derived from satellite burned area)

Available

At a glance¶

Note on period: GFED5.1 covers 1997–2022 (1997–2000 at 1°, estimated from ATSR and TRMM VIRS active fires; 0.25° from 2001), and GFED5.1NRT continues from 2023 (the reprocessed Zenodo release covers 2023–2024; later data are updated daily on the site's SFTP). GFED4.1s is burned-area based for 1997–2016 with beta files for 2017–2023 (on 2026-10-13 the latest file in the web directory was GFED4.1s_2023_beta, updated 2024-01-26).

Overview¶

GFED estimates global fire emissions of carbon, trace gases and aerosols on a 0.25° grid by multiplying satellite burned area by fuel consumption from a biogeochemical model and by emission factors. GFED5 builds on MODIS MCD64A1 C6.1 with omission and commission corrections constrained by Landsat and Sentinel-2, small fires from active fire detections and cropland burning from GloCAB, which substantially increases burned area, and it also revises emission factors and fuel consumption. It is widely used as fire emission input for atmospheric chemistry and air quality models, in carbon budgets such as the Global Carbon Project, and in IPCC assessments.

What changes with the successor¶

GFED4.1s → GFED5 (GFED5.1 + GFED5.1NRT)¶

Transition: 2025-10

GFED5 is the largest overhaul since GFED4, replacing both the burned area input and the emission factors; global burned area rises by about 61% and carbon emissions by about 65%. Burned area was published in 2023 (ESSD, Chen et al.) and emissions in 2025 (Scientific Data, van der Werf et al.), with a VIIRS active-fire-based near-real-time extension from 2023. Because the magnitudes differ so much, the two versions must not be mixed within one study.

Sources: van der Werf et al. (2025), Scientific Data · Chen et al. (2023), ESSD · GFED Data page (GFED4s description, GFED5 recommended) · Chen et al. (2026), ESSD discussion paper (GFED5NRT)

Available variables¶

GFED5.1 daily emission files (GFED5.1_daily_YYYY-MM.nc) contain variables for 40 species following the same naming scheme. Variable names in the GFED5.1 monthly files could not be confirmed from public documentation, so the verified GFED5.0 burned area files and GFED4.1s (HDF5, with group paths) variables are listed as well.

instantaneous mean accumulated maximum minimum static — instantaneous: value at the time stamp · mean/accumulated/maximum/minimum: over the indicated interval · static: does not change with time

GFED5.1 daily emissions (GFED5.1_daily_YYYY-MM.nc, 0.25°)¶

GFED5.0 burned area (BAYYYYMM.nc, Chen et al. 2023 Zenodo, 2001–2020 at 0.25°)¶

GFED4.1s (GFED4.1s_YYYY.hdf5, monthly groups 01–12)¶

Invariant fields (GFED4.1s ancill group)¶

The full variable list is in the official documentation.

How to get the data¶

• GFED website — GFED5 gridded files (SFTP) — SFTP (server ftp.prd.dip.wur.nl, port 1022)
  • Holds GFED5.1 (1997–2022) and GFED5.1NRT (2023 onward, in reprocessed and update directories) as netCDF4 files. Public login details are at https://www.globalfiredata.org/ancill/GFED5_SFTP_info.txt (note port 1022).
• Zenodo — GFED5.1 (paper data) — HTTPS download (four zip files, about 20.5 GB)
  • GFED5.1_monthly (1.7 GB, 1997–2022 monthly burned area and emissions), GFED5.1_daily (18.4 GB, daily emissions), GFED5.1_ecosystem (2002–2022 burned area and carbon emissions by category), GFED5.1_code
• Zenodo — GFED5NRT (near-real-time extension, reprocessed) — HTTPS download
  • Daily and monthly data for 2023–2024, split into Species and Ecosystem files for each sensor combination (CMB combined, VNP Suomi NPP, VJ1 NOAA-20).
• GFED summary tables (global and regional annual emissions) — Text table download
• GFED4.1s (previous version, HDF5) — HTTPS directory download
  • Contains GFED4.1s_1997–2016.hdf5, GFED4.1s_2017–2023_beta.hdf5, Readme.pdf and ancill (emission factor tables).
• ORNL DAAC — GFED4.1 (1997–2016) — Download after Earthdata login
  • NASA archive copy with GFED4 monthly and daily burned area without small fires (HDF) and GFED4.1s annual files (1997–2015).

Example code¶

Downloads a GFED4.1s HDF5 file to compute annual global carbon emissions, then opens a GFED5.1 daily file obtained from SFTP or Zenodo with xarray. Check file names and variables against the README of the GFED5.1 release you download.

# pip install h5py numpy requests xarray netCDF4
import h5py
import numpy as np
import requests
import xarray as xr

# 1) GFED4.1s: global carbon emissions in 2015
url = "https://www.geo.vu.nl/~gwerf/GFED/GFED4/GFED4.1s_2015.hdf5"
with open("GFED4.1s_2015.hdf5", "wb") as out:
    out.write(requests.get(url, timeout=600).content)

with h5py.File("GFED4.1s_2015.hdf5", "r") as f:
    area = f["ancill/grid_cell_area"][:]                          # m2
    c = sum(f[f"emissions/{m:02d}/C"][:] for m in range(1, 13))   # g C m-2 (sum of monthly totals)
    print("2015 global carbon emissions:", (c * area).sum() / 1e15, "Pg C")

# 2) GFED5.1 daily emissions file (from SFTP or GFED5.1_daily.zip on Zenodo)
ds = xr.open_dataset("GFED5.1_daily_2020-01.nc")
co2 = ds["CO2"]                        # (time, lat, lon), 0.25°
print(co2.attrs.get("units"))          # 'g CO2 per day' — check whether per area or per cell
print(co2.sel(lat=slice(33, 39), lon=slice(124, 132)).sum(("lat", "lon")))

Practical tips¶

• GFED5 burned area is about 61% and carbon emissions about 65% higher than GFED4s (mostly small fires, omission corrections and cropland burning). Do not mix the two versions or drop GFED5 straight into model setups tuned to GFED4s.
• From 2023 (GFED5.1NRT), burned area is not observed directly but converted from VIIRS active fire counts, and the official site labels it beta. State this change in method when extending pre-2023 trends.
• The GFED5NRT combined (CMB) product is designed to fall back to the remaining instrument when Suomi NPP or NOAA-20 is missing. With Suomi NPP data delivery ending on 1 November 2026, later values will effectively be NOAA-20 only.
• 1997–2000 is at 1° and estimated from ATSR and VIRS active fires, with large uncertainty. The README warns of discontinuities in 1998 (VIRS introduced), 2001 (Terra MODIS) and 2003 (Aqua MODIS).
• Cropland burning emissions are the most uncertain component (very few field measurements of fuel consumption). State this uncertainty when analysing agricultural burning in Korea, China and elsewhere.
• GFED4.1s _beta files from 2017 onward are estimated from MODIS active fires using 2003–2016 relationships and contain only DM and C. Compute species emissions yourself as DM × emission factor (tables in the ancill folder) × fire-type share (partitioning).
• GFED4.1s HDF5 arrays run from north (89.875°) to south. Check latitude order when overlaying other datasets.
• The team notes that uncertainty grows at small scales such as individual 0.25° cells or sub-national regions. Aggregates over large regions and long periods are the safest use.

Notices and change history¶

References for your paper¶

Dataset citation¶

GFED5.1 data (Zenodo) — DOI 10.5281/zenodo.16794692

@misc{gfed51_data,
  author    = {van der Werf, Guido and Randerson, James and van Wees, Dave and Chen, Yang and Giglio, Louis and Hall, Joanne and Vernooij, Roland and Mu, Mingquan and Binte Shahid, Samiha and Barsanti, Kelley and Yokelson, Robert and Morton, Douglas},
  title     = {Global Fire Emissions Database version 5 ({GFED5})},
  year      = {2025},
  publisher = {Zenodo},
  doi       = {10.5281/zenodo.16794692}
}

GFED5NRT data (Zenodo, V5.1) — DOI 10.5281/zenodo.18702700

@misc{gfed5nrt_data,
  author    = {Chen, Yang and van der Werf, Guido R. and Mu, Mingquan and van Wees, Dave and Hall, Joanne V. and Giglio, Louis and Vernooij, Roland and Morton, Douglas C. and Xu, Li and Liu, Tianjia and Scholten, Rebecca C. and Coffield, Shane R. and Follette-Cook, Melanie B. and Ott, Lesley and Randerson, James T.},
  title     = {{GFED5NRT}: Global Fire Emissions Database Near-Real-Time Extension},
  version   = {V5.1},
  year      = {2026},
  publisher = {Zenodo},
  doi       = {10.5281/zenodo.18702700}
}

GFED4.1 (ORNL DAAC, 1997–2016) — DOI 10.3334/ORNLDAAC/1293

@misc{gfed41_ornl,
  author    = {Randerson, J. T. and van der Werf, G. R. and Giglio, L. and Collatz, G. J. and Kasibhatla, P. S.},
  title     = {Global Fire Emissions Database, Version 4.1 ({GFEDv4})},
  year      = {2018},
  publisher = {ORNL DAAC},
  address   = {Oak Ridge, Tennessee, USA},
  doi       = {10.3334/ORNLDAAC/1293}
}

Reference papers¶

GFED5 emissions (for emissions through 2022) — DOI 10.1038/s41597-025-06127-w

@article{vanderwerf2025gfed5,
  author  = {van der Werf, Guido R. and Randerson, James T. and van Wees, Dave and Chen, Yang and Giglio, Louis and Hall, Joanne and Vernooij, Roland and Mu, Mingquan and Binte Shahid, Samiha and Barsanti, Kelley C. and Yokelson, Robert and Morton, Douglas C.},
  title   = {Landscape fire emissions from the 5th version of the Global Fire Emissions Database ({GFED5})},
  journal = {Scientific Data},
  volume  = {12},
  pages   = {1870},
  year    = {2025},
  doi     = {10.1038/s41597-025-06127-w}
}

GFED5 burned area (for burned area through 2022) — DOI 10.5194/essd-15-5227-2023

@article{chen2023gfed5ba,
  author  = {Chen, Yang and Hall, Joanne and van Wees, Dave and Andela, Niels and Hantson, Stijn and Giglio, Louis and van der Werf, Guido R. and Morton, Douglas C. and Randerson, James T.},
  title   = {Multi-decadal trends and variability in burned area from the fifth version of the Global Fire Emissions Database ({GFED5})},
  journal = {Earth System Science Data},
  volume  = {15},
  number  = {11},
  pages   = {5227--5259},
  year    = {2023},
  doi     = {10.5194/essd-15-5227-2023}
}

GFED5NRT (for data from 2023 onward; ESSD discussion paper, revised version accepted) — DOI 10.5194/essd-2026-158

@article{chen2026gfed5nrt,
  author  = {Chen, Yang and van der Werf, Guido R. and Mu, Mingquan and van Wees, Dave and Hall, Joanne V. and Giglio, Louis and Vernooij, Roland and Morton, Douglas C. and Xu, Li and Liu, Tianjia and Scholten, Rebecca C. and Coffield, Shane R. and Follette-Cook, Melanie B. and Ott, Lesley and Randerson, James T.},
  title   = {Tracking recent extremes and interannual variability of global fire emissions using a near-real-time extension to the Global Fire Emissions Database},
  journal = {Earth System Science Data Discussions},
  year    = {2026},
  doi     = {10.5194/essd-2026-158}
}

GFED4s (previous version) — DOI 10.5194/essd-9-697-2017

@article{vanderwerf2017gfed4s,
  author  = {van der Werf, Guido R. and Randerson, James T. and Giglio, Louis and van Leeuwen, Thijs T. and Chen, Yang and Rogers, Brendan M. and Mu, Mingquan and van Marle, Margreet J. E. and Morton, Douglas C. and Collatz, G. James and Yokelson, Robert J. and Kasibhatla, Prasad S.},
  title   = {Global fire emissions estimates during 1997--2016},
  journal = {Earth System Science Data},
  volume  = {9},
  pages   = {697--720},
  year    = {2017},
  doi     = {10.5194/essd-9-697-2017}
}

Small-fire burned area (GFED4s method) — DOI 10.1029/2012JG002128

@article{randerson2012smallfires,
  author  = {Randerson, J. T. and Chen, Y. and van der Werf, G. R. and Rogers, B. M. and Morton, D. C.},
  title   = {Global burned area and biomass burning emissions from small fires},
  journal = {Journal of Geophysical Research: Biogeosciences},
  volume  = {117},
  number  = {G4},
  year    = {2012},
  doi     = {10.1029/2012JG002128}
}

Acknowledgements¶

Following the GFED FAQ, cite Chen et al. (2023) for burned area through 2022, van der Werf et al. (2025) for emissions through 2022, and Chen et al. (2026) for burned area or emissions after 2022; studies using emissions across both periods should cite van der Werf et al. (2025) and Chen et al. (2026). Media and presentations should credit GFED and link to the website.

Information on the datasets used in this study was compiled with the help of the Korean Climate Data Navigator (KCDN) (Sim, 2026; https://doi.org/10.5281/zenodo.23300923).

Links¶

• GFED website
• GFED Data page
• GFED FAQ (citation guidance)
• GFED current fire situation
• GFED5 burned area README (GFED5.0, Zenodo)
• GFED4s Readme (HDF5 structure and emission calculation)
• ORNL DAAC GFED4.1 user guide
• GFED5NRT code (GitHub)