Predicting and Preventing Crisis in Water and Food Security in a Changing Climate through a paradigm-shift in global cropland maps and products through machine learning and cloud computing
Prasad S. Thenkabail
Adaptation and mitigation to climate change are key to ensuring global water and food security in the twenty-first century. In order to contribute to this humanity sustaining goal, the United States Geological Survey (USGS) produced the world’s first, Landsat Satellite-derived, global cropland extent product at 30m resolution (GCEP30) for the nominal year 2015 which is the highest resolution freely available global cropland dataset. The product was a result of a “paradigm-shift” in cropland production involving satellite sensor big-data analytics, machine learning, and cloud-computing on Google Earth Engine (GEE). The GCEP30 had an overall map accuracy of 91.7% with cropland class producer’s accuracy of 83.4% (errors of omission of 16.6%) and user’s accuracy of 78.3% (errors of commission of 21.7%). For the year 2015, GCEP30-derived global net cropland area was 1.873 Bha (~12.6% of the terrestrial area). The GCEP30 is downloadable at LP DAAC and viewable in full resolution here.