We have been trying for centuries to predict crop yields and trying to discover the future of crops. Now, thanks to the power of data processing and access to sources of different backgrounds (satellites, computers, climate data, etc.), we are much closer to reliable long-term predictions.
According to a recent study, led by Kaiyu Guan and published in Agricultural and Forest Meteorology, the use of machine learning systems can accurately predict wheat yield two months before the crop matures.
“We tested several machine learning approaches and integrated large-scale satellite and climatic data to arrive at a reliable and accurate prediction of wheat production across Australia,” Guan explains. “In this study, we used an analysis to identify the predictive power of weather and satellite data. We wanted to know how each one contributes. We found that climate data alone is pretty good, but satellite data provides additional information and takes prediction performance to the next level.”
By using climate and satellite data sets, researchers were able to predict wheat yield with approximately 75% about 60 days before harvest.
“Specifically, we found that satellite data can determine crop yield variability,” Guan adds. “Climate information that cannot be captured by satellite data serves as a unique contribution to wheat yield prediction throughout the season. of growth .”
Those responsible for the study affirm that the results can be used to improve predictions about future crops, with possible effects on the economy. In addition, they are optimistic that the results can be reproduced in other countries and other crops.