A maize field can look uniformly green while parts of it are already shutting down under heat and drought. Kelechi Igwe, a doctoral student in biological and agricultural engineering at Kansas State University, is building an early-warning system to find those patches before the damage is visible, and before it costs yield.

His measure is stomatal conductance, the rate at which the pores on a leaf exchange water vapour and gases with the air. Under stress the stomata begin to close, so conductance falls, but the standard way to read it, a handheld porometer clamped on one leaf at a time, is far too slow for a commercial field.

Igwe's answer is to estimate conductance from things that can be measured at scale: drone imagery, air temperature and soil moisture readings. Fed into a machine-learning model, they produce a field-wide stress map. In testing the model was right about half the time, and it picked out areas already under water stress even though they appeared healthy to the eye.

The work was presented as "Beyond the Visible: An Early Warning System for Detecting Water Stress in Maize" in Kansas State's Three-Minute Thesis competition and featured in the Graduate School's Driven to Discover series. The accuracy is modest so far, but the approach, drone imagery plus environmental sensors plus a model, is what the research is testing.

Farms.com's account notes that earlier warning of water stress lets farmers make better irrigation decisions, use water more efficiently and reduce the risk of yield loss.

Photo: Kansas State University

Source: Farms.com