MODIS ET (MOD16)¶

MODIS/Terra and Aqua Net Evapotranspiration MOD16A2(GF), MYD16A2(GF), MOD16A3GF, MYD16A3GF Version 6.1 (global 500 m actual and potential evapotranspiration from MODIS)

Being replaced

At a glance¶

Note on period: The year-end gap-filled (GF) products are the long-term record. MOD16A2GF starts in 2000 (CMR lists 2000-01-01; Terra data begin in late February 2000) and MYD16A2GF in 2002 (Aqua data from 2002-07-04), both running to December 2025 (the 2025 files were produced on 2026-02-05). The non-gap-filled forward products (MOD16A2, MYD16A2) only start on 2021-01-01 in V6.1 and, on 2026-10-13, extended to the 2026-09-22–29 composite. Yearly data exist in CMR only as GF products (MOD16A3GF, MYD16A3GF), up to 2025.

Overview¶

MOD16 is a Level 4 product that feeds MODIS leaf area index and fraction of absorbed PAR (LAI/FPAR), albedo and land cover, together with daily GMAO meteorological reanalysis, into a Penman-Monteith-based algorithm (Mu et al., 2011) to estimate actual evapotranspiration (ET), latent heat flux (LE), potential ET (PET) and potential LE (PLE) over global vegetated land at 500 m every 8 days. There are Terra (MOD16) and Aqua (MYD16) versions; the year-end gap-filled (GF) products, rebuilt after poor-quality LAI/FPAR values in the full year are replaced by temporal interpolation, are the research standard. It is widely used for water balance, drought and agricultural water use studies, but has no values over urban, barren, water and permanent snow/ice pixels.

What changes with the successor¶

MOD16 C6 → MOD16 C6.1¶

Transition: 2021-03

The science algorithm is unchanged; the input MODIS L1B calibration improved and gap handling changed. C6 left pixels with contaminated LAI/FPAR as fill values, leaving many gaps, whereas C6.1 fills them with climatological LAI/FPAR and gap-filled albedo as backup inputs. Distribution of C6 ended on 2023-07-31.

Sources: MOD16 C6.1 User's Guide · Final Notice: MODIS Version 6.0 Land Data Products Decommissioned July 31, 2023

MODIS evapotranspiration MOD16/MYD16 C6.1 → VIIRS evapotranspiration VNP16/VJ116 V2.1 (not yet released) Planned¶

To prepare for the end of MODIS, NASA released VNP16, VJ116 and VJ216 V2 — the same MOD16 algorithm applied to VIIRS on Suomi NPP, NOAA-20 and NOAA-21 — in 2025–2026, but withdrew all of them on 2026-08-17 after data issues were found, stating that they are not of science quality and should not be used. They will be reprocessed as V2.1 once the issues are fixed; no date has been set.

Sources: VIIRS Actual and Potential Evapotranspiration Data Products Unavailable (LP DAAC, 2026-08-17) · VIIRS Version 2 Actual and Potential Evapotranspiration Data Products Released (LP DAAC, 2025-07-10) · VIIRS Land — NASA Evapotranspiration and Potential Evapotranspiration

Available variables¶

Science datasets (SDS) from the LP DAAC data catalog and the C6.1 User's Guide. In the HDF-EOS2 (HDF4) files, physical value = stored value × scale_factor; in the 8-day files, stored values 32761–32767 are fill values marking non-vegetated land cover classes and must be removed first. MYD16 (Aqua) files use the same names.

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

8-day data (MOD16A2GF, MYD16A2GF year-end gap-filled / MOD16A2, MYD16A2 forward)¶

Yearly data (MOD16A3GF, MYD16A3GF, gap-filled only)¶

The full variable list is in the official documentation.

How to get the data¶

• NASA Earthdata Search (LP DAAC, Earthdata Cloud) — HTTPS download / earthaccess / Earthdata Search
  • Short names are MOD16A2GF and MYD16A2GF (8-day gap-filled), MOD16A3GF and MYD16A3GF (yearly gap-filled), and MOD16A2 and MYD16A2 (8-day forward), all version 061. Files are served over HTTPS from data.lpdaac.earthdatacloud.nasa.gov and need a free Earthdata Login. Distribution through the old LP DAAC Data Pool ended on 2025-06-30.
• AppEEARS (area and point subsetting) — Web / API — point time series and area extracts as GeoTIFF or NetCDF4
  • Lets you request only the area, period and layers you need and reproject to geographic coordinates, which is convenient for small regions.
• Direct access in AWS us-west-2 (S3) — Cloud storage (S3, from within us-west-2 only)
  • Stored under paths such as s3://lp-prod-protected/MOD16A2GF.061/ and accessed with temporary credentials.

Example code¶

Downloads the Terra 8-day gap-filled tiles around the Korean Peninsula for July 2023 with earthaccess and reads them with pyhdf (the files are HDF4, which xarray's default engines may not open). Remove the fill values (32761 and above) first.

# pip install earthaccess pyhdf numpy
import earthaccess
import numpy as np
from pyhdf.SD import SD, SDC

earthaccess.login()  # free NASA Earthdata account
results = earthaccess.search_data(
    short_name="MOD16A2GF", version="061",
    temporal=("2023-07-01", "2023-07-31"),
    bounding_box=(124, 33, 132, 39),   # (west, south, east, north)
)
files = earthaccess.download(results, "mod16")

hdf = SD(files[0], SDC.READ)
raw = hdf.select("ET_500m").get().astype("float64")
raw[raw > 32700] = np.nan          # 32761–32767: water, barren, urban, snow/ice and other fill
et = raw * 0.1                      # mm per 8 days (5–6 days in the last period of the year)
qc = hdf.select("ET_QC_500m").get()
print(files[0], np.nanmean(et))

Practical tips¶

• Use the year-end gap-filled products (MOD16A2GF, MYD16A2GF, MxD16A3GF) for research. The forward MOD16A2 is a near-real-time product that substitutes a recent 5-year LAI/FPAR climatology for cloudy inputs, and in V6.1 it only starts in 2021.
• Stored ET_500m values of 32761–32767 (65529–65535 in yearly files) are not missing data but non-vegetated land cover flags (32766 water, 32765 barren or sparsely vegetated, 32764 permanent snow/ice, 32763 wetland, 32762 urban, 32761 unclassified). Remove them before applying the scale factor or absurd values will creep in.
• Units differ. ET and PET are 8-day totals (mm); LE and PLE are daily averages (J m-2 day-1, scale 10000). The last period of each year is a 5- or 6-day total.
• Biome parameters are fixed per MODIS land cover class (IGBP 17 classes, 3-year smoothed), so land cover misclassification passes straight into ET error and regional variation within a biome is not captured. Pixels with no valid albedo during a year are computed with an albedo of 0.4.
• The meteorological input is GMAO reanalysis (0.5°×0.6° or 1°×1.25°) interpolated to 500 m, so in mountains and along coasts, where weather changes sharply, the product does not carry 500 m-scale meteorological detail.
• The first GF year (2000 for Terra, 2002 for Aqua) is less reliable because the start of the year has no observations to interpolate from.
• Data come in MODIS Sinusoidal tiles (the Korean Peninsula is mostly h27v05 and h28v05). Reproject with GDAL or rioxarray, or use AppEEARS, to get a latitude-longitude grid.
• Terra and Aqua are expected to shut down from late 2026 onward, and the VIIRS ET successor was withdrawn on 2026-08-17. If you need to extend a long time series, watch both the MODIS end-of-mission and the VIIRS V2.1 release notices.

Notices and change history¶

References for your paper¶

Dataset citation¶

MOD16A2GF V6.1 (Terra 8-day gap-filled) — DOI 10.5067/MODIS/MOD16A2GF.061

@misc{mod16a2gf_v061,
  author    = {Running, Steve and Mu, Qiaozhen and Zhao, Maosheng and Moreno, Alvaro},
  title     = {{MODIS/Terra} Net Evapotranspiration Gap-Filled 8-Day {L4} Global 500m {SIN} Grid {V061}},
  year      = {2021},
  publisher = {NASA Land Processes Distributed Active Archive Center},
  doi       = {10.5067/MODIS/MOD16A2GF.061}
}

MOD16A3GF V6.1 (Terra yearly gap-filled) — DOI 10.5067/MODIS/MOD16A3GF.061

@misc{mod16a3gf_v061,
  author    = {Running, Steve and Mu, Qiaozhen and Zhao, Maosheng and Moreno, Alvaro},
  title     = {{MODIS/Terra} Net Evapotranspiration Gap-Filled Yearly {L4} Global 500m {SIN} Grid {V061}},
  year      = {2021},
  publisher = {NASA Land Processes Distributed Active Archive Center},
  doi       = {10.5067/MODIS/MOD16A3GF.061}
}

MOD16A2 V6.1 (Terra 8-day forward) — DOI 10.5067/MODIS/MOD16A2.061

@misc{mod16a2_v061,
  author    = {Running, Steve and Mu, Qiaozhen and Zhao, Maosheng},
  title     = {{MODIS/Terra} Net Evapotranspiration 8-Day {L4} Global 500m {SIN} Grid {V061}},
  year      = {2021},
  publisher = {NASA Land Processes Distributed Active Archive Center},
  doi       = {10.5067/MODIS/MOD16A2.061}
}

Reference papers¶

MOD16 algorithm (Mu et al., 2011) — DOI 10.1016/j.rse.2011.02.019

@article{mu2011mod16,
  author  = {Mu, Qiaozhen and Zhao, Maosheng and Running, Steven W.},
  title   = {Improvements to a {MODIS} global terrestrial evapotranspiration algorithm},
  journal = {Remote Sensing of Environment},
  volume  = {115},
  number  = {8},
  pages   = {1781--1800},
  year    = {2011},
  doi     = {10.1016/j.rse.2011.02.019}
}

Acknowledgements¶

Cite the LP DAAC dataset DOI for the product you used (Terra or Aqua, 8-day or yearly, gap-filled or forward). The Aqua DOIs are 10.5067/MODIS/MYD16A2GF.061, 10.5067/MODIS/MYD16A3GF.061 and 10.5067/MODIS/MYD16A2.061. Cite Mu et al. (2011) for the algorithm.

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¶

• MOD16A2GF V061 data catalog (LP DAAC)
• MOD16 C6.1 User's Guide
• MOD16 ATBD
• MOD16 validation status (MODIS Land)
• MODIS/VIIRS Land Quality Assessment (LDOPE) known issues
• MODIS Land News (Terra and Aqua orbit drift)
• VIIRS evapotranspiration products (VIIRS Land)