How can we know which soil nutrients are most important for crop production in a region, and how much gain can be had from better fertilization? In a recent collaboration with CIMMYT, we explored how large, new soil datasets collected on the ground can be combined with satellite-based measured yields to answer these questions. Read a blog post describing the work here.
In a collaboration with Corteva AgriScience we evaluate satellite estimates of maize yields with an extensive field dataset, and then examine causes of yield gaps in U.S. maize production. Read a blog post describing the work here.
Using satellite measures of crop yield, we examine how sensitivity to soil water has changed over time in the United States. Read a blog post describing the work here.
New types of training data can provide a powerful way to train machine learning models that use satellite inputs. You can read a blog post describing our recent paper here.