NASA scientists have succesfully used lasers mounted on the International Space Station to map where corn is grown in the U.S., China and France. In the future, the research team aims to map corn production around the world, which could be used to understand the harvest prospects of corn each year.
This could also help farmers and aid agencies assess food security concerns and get a sense of possible changes in management that could improve production in major corn-producing regions.
The lasers on the International Space Station are part of NASA’s Global Ecosystem Dynamics Investigation (GEDI). They send 242 rapid pulses of light down to Earth every second, which bounce off Earth’s surfaces and can be used to create three-dimensional profiles of Earth’s surface. GEDI’s primary mission is to measure tree heights and forest structure in order to estimate the amount of carbon stored in forests and mangroves. New research supported by NASA Harvest however reveals these data also can be used to map where different types of crops are being grown.
Mapping where certain crops are grown is important for estimating the overall production of the world’s major crops. But according to NASA it has been difficult to reliably map crop types from space because many plants can look the same in optical imagery.
David Lobell is an agricultural ecologist at Stanford University and helps lead crop yield studies for NASA Harvest. He and his team started using GEDI data to map corn. When fully grown, average corn stalks are about a meter taller than other crops, a difference that is detectable in GEDI profiles. Using this insight, the lidar profile data from GEDI was combined with optical imagery from the European Space Agency’s Sentinel-2 satellites. They were able to remotely map corn in three regions where there was reliable ground-based data to validate their observations: the state of Iowa in the U.S., the province of Jilin in China, and the region of Grand Est in France.
The Stanford algorithm correctly distinguished corn from other crops with an accuracy above 83 percent. The model using Sentinel-2 data alone had an overall average accuracy of 64 percent.