Global Cropland Extent Product at 30 meters (GCEP30)

Submitted by tadamson on Mon, 05/17/2021 - 14:45

The monitoring of global cropland extent is critical for policymaking and provides important baseline data that are used in many agricultural cropland studies related to water sustainability and food security. The USGS, in partnership with several U.S. universities and international institutes, produced the world’s first Landsat-derived, global cropland extent product at 30-meter resolution (GCEP30) for the nominal year 2015. This product represents the highest resolution freely available global cropland dataset. The cropland products are the result of a recent paradigm shift in data analysis, leveraging big data and cloud computing. Their derivation involved segmentation of the world into 74 agro-ecological zones, massive reference training and validation data, machine learning algorithms, and petabyte-scale cloud-computing on Google Earth Engine (GEE). Classification accuracies are high, with an overall map accuracy of 91.7%. For the global cropland class, the producer’s accuracy was 83.4% (errors of omission: 16.6%) and user’s accuracy 78.3% (errors of commission: 21.7%). The product offers insights into the distribution of cropland across the globe.  For the year 2015, GCEP30-derived global net cropland area (GNCA) was 1.873 billion hectares (Bha), or ~12.6% of the terrestrial area. The 10 leading cropland area countries as a percentage of the GNCA were: India (9.6% of GNCA), United States (8.95%), China (8.82%), Russia (8.32%), Brazil (3.42%), Ukraine (2.32%), Canada (2.29%), Argentina (2.05%), Indonesia (2.0%), and Nigeria (1.91%). Four countries (India, United States, China, and Russia) together encompass 36% (670 Mha) of the GNCA. GCEP30-derived country-wide statistics compare favorably with those reported in the United Nations Food and Agriculture Organization Statistical Database (FAOSTAT); the GCEP30 explained 93% of the variability in country-wide cropland data reported by FAOSTAT. The GCEP30 is downloadable at and viewable at full resolution at

Global cropland product at 30-meter spatial resolution for the year 2015 derived using Landsat satellite time-series data, machine learning algorithms, and cloud computing.

Author Name
Prasad S. Thenkabail; Pardhasaradhi G. Teluguntla; Adam J. Oliphant
Author Email;;