What new methods can be employed to enable scientists to better understand gully erosion and how it impacts soil loss? This is what a recent study published in the Journal of Environmental Management hopes to address as a team of researchers investigated using machine learning to predict both locations of gully erosion with the goal of prevention and mitigation. This study has the potential to help scientists, farmers, and the public better understand gully erosion and how it can be prevented to help improve agriculture productivity.
For the study, the researchers used a series of machine learning models to analyze and predict potential gully sites based on approximately 200 random gully sites from existing data points within Jefferson County, Illinois. The goal of the study was to develop a stacking model to predict future gullies and address them before they negatively impact the surrounding soil. In the end, the researchers found that their stacking model method provided achieved 91.6 percent prediction accuracy compared to 86 percent for non-stacked models.
“We had conducted a previous study in the same area, but we applied only an individual machine learning model to predict gully erosion susceptibility,” said Dr. Jeongho Han, who recently defended his PhD at the University of Illinois at Urbana-Champaign and is lead author of the study. “While that study provided a baseline understanding, it had limited predictive accuracy. Furthermore, we were not able to explain how the model made predictions. This research aims to address these two key limitations.”
Going forward, this method could help farmers and land managers better understand and predict gully erosion, resulting in increased agricultural productivity.
How will machine learning help scientists better understand and predict gully erosion in the coming years and decades? Only time will tell, and this is why we science!
As always, keep doing science & keep looking up!