TerraClimate¶

TerraClimate (global ~4 km monthly climate and climatic water balance for land, Climatology Lab)

Available

At a glance¶

Note on period: v1.1 starts in 1950, but the producers suggest skipping 1950 because of water-balance model spin-up. Yearly files on THREDDS begin in 1950 (checked 2026-10-13), while the multi-year aggregated files and the official description still say 1958 onward. Monthly climatologies for 1981–2010 and 1991–2020 and +2°C, +4°C and counterfactual scenarios (pseudo-years 1950–2025) are provided separately.

Overview¶

TerraClimate uses climatically aided interpolation — high-resolution monthly normals from WorldClim plus monthly anomalies from coarser time-varying datasets — to produce monthly temperature, precipitation, vapour pressure, solar radiation and wind speed for global land at about 4 km. These drive a simple water-balance model that yields actual and reference evapotranspiration, climatic water deficit, runoff, soil moisture, snow water equivalent and the PDSI. Version 1.0 used CRU TS4.0 and JRA-55; the current v1.1 uses ERA5 anomalies. It is popular in ecology, hydrology and agriculture when terrain-resolving monthly climate and drought indicators are needed.

What changes with the successor¶

TerraClimate v1.0 → TerraClimate v1.1¶

To get around the discontinued JRA-55 and delayed CRU updates, the time-varying inputs were switched to ERA5 monthly anomalies, and the reference ET and runoff formulations were revised. The whole record was recomputed, so v1.0 and v1.1 must not be mixed. The official page dates the v1.1 release announcement 15 February 2025 and the description of changes 'Updated 15 February 2026'.

Sources: TerraClimate (Climatology Lab) — v1.1 announcement and changes · Abatzoglou et al. (2018), Scientific Data · Google Earth Engine IDAHO_EPSCOR/TERRACLIMATE

Available variables¶

Based on the Climatology Lab variable table and the THREDDS file attributes. Each variable is stored in one file per year (TerraClimate__.nc). Some band names differ in Google Earth Engine (ppt→pr, q→ro, tmax→tmmx, tmin→tmmn, ws→vs, PDSI→pdsi).

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

Climate variables (monthly means)¶

Water-balance fluxes (monthly totals)¶

Storage and drought index (end of month)¶

The full variable list is in the official documentation.

How to get the data¶

• Climatology Lab — TerraClimate overview and download guide — Information page (links to THREDDS, direct downloads and wget scripts)
  • Version notes, change history, and links to scenario and climatology data are all on this page.
• NKN THREDDS server (yearly files) — HTTP download / OPeNDAP / NetCDF Subset Service (NCSS)
  • Files are named TerraClimate__.nc. Multi-year aggregated datasets (from 1958) are at http://thredds.northwestknowledge.net:8080/thredds/terraclimate_aggregated.html, and OPeNDAP lets you subset a region or point.
• Google Earth Engine (IDAHO_EPSCOR/TERRACLIMATE) — Earth Engine API (cloud analysis)
  • On 2026-10-13 it ran to December 2024 and its description was still v1.0-based. Band names and scale values differ from the THREDDS files.
• v1.0 (previous release) data — FTP (on request)
  • v1.0 has been moved to an FTP server; contact John Abatzoglou or Katherine Hegewisch if you need it.

Example code¶

Opens the 2025 precipitation file through the THREDDS OPeNDAP endpoint and subsets the Korean Peninsula. No account is needed. If your netCDF library raises a _FillValue type error, the official page recommends appending '#fillmismatch' to the URL.

import xarray as xr

url = ("http://thredds.northwestknowledge.net:8080/thredds/dodsC/"
       "TERRACLIMATE_ALL/data/TerraClimate_ppt_2025.nc#fillmismatch")
ds = xr.open_dataset(url)          # xarray applies scale_factor automatically
# Latitude is stored north to south, so slice from north to south
ppt = ds["ppt"].sel(lat=slice(39, 33), lon=slice(124, 132))
print(ppt.sum("time").mean().values)   # area-mean 2025 annual precipitation (mm)

Practical tips¶

• Do not join v1.0 and v1.1 into one time series. The inputs (CRU/JRA-55 → ERA5) and the PET and runoff formulations changed, so historical values all differ. State the version in your paper.
• Long-term trends are inherited from the parent datasets (such as ERA5). The producers advise against using TerraClimate directly for independent trend assessments, such as temperature trends.
• Because it combines WorldClim normals with interpolated anomalies, it cannot capture sharp anomaly gradients in mountainous or coastal areas. Validation is limited in data-sparse regions such as Antarctica.
• The water-balance model is simple — static vegetation, a single soil water holding capacity per location and a temperature-based snow model. soil, aet and q are model outputs, not observations.
• soil, swe and PDSI are end-of-month values; ppt, aet, pet, def and q are monthly totals. Do not average the monthly-total variables to get annual values.
• The official page notes that winter temperature inversions may be extrapolated unrealistically in boreal regions such as Alaska.
• The Google Earth Engine copy ran to 2024 on 2026-10-13 and carries no version label. If you need the latest year or v1.1 specifically, use the THREDDS files.

Notices and change history¶

References for your paper¶

Reference papers¶

TerraClimate description paper (v1.0) — DOI 10.1038/sdata.2017.191

@article{abatzoglou2018terraclimate,
  author  = {Abatzoglou, John T. and Dobrowski, Solomon Z. and Parks, Sean A. and Hegewisch, Katherine C.},
  title   = {{TerraClimate}, a high-resolution global dataset of monthly climate and climatic water balance from 1958--2015},
  journal = {Scientific Data},
  volume  = {5},
  pages   = {170191},
  year    = {2018},
  doi     = {10.1038/sdata.2017.191}
}

Acknowledgements¶

There is no dataset DOI; the Climatology Lab asks users to cite Abatzoglou et al. (2018). When using v1.1, also give the version (v1.1), the access date and where you obtained the data (THREDDS or Earth Engine). For the +2°C/+4°C scenarios, Qin et al. (2020, Nature Climate Change) should also be cited.

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¶

• TerraClimate (Climatology Lab)
• TerraClimate variables
• THREDDS catalogue (TerraClimate)
• Google Earth Engine catalogue
• Abatzoglou et al. (2018)