Communications of the ACM
Hyperspectral Sensing, AI Pave Path for Monitoring Soil Carbon
Said Sheng Wang of the University of Illinois' Agroecosystem Sustainability Center and Department of Natural Resources and Environmental Sciences, “We did a comprehensive evaluation of machine learning algorithms with a very intensive national soil
Credit: University of Illnois Urbana-Champaign ACES News
University of Illinois (U of I) researchers demonstrated that new artificial intelligence-powered algorithms based on laboratory soil hyperspectral data can estimate soil organic carbon properties as precisely as fieldwork.
"We knew some of these models worked, but the novelty is the scale and that we tried the full gamut of machine learning algorithms," said U of I's Andrew Margenot.
The researchers utilized a public soil spectral library from the U.S. Department of Agriculture's Natural Resources Conservation Service covering all soil types throughout the country.
They tested the best algorithm using simulated airborne and spaceborne hyperspectral data, and the model factored in "noise" inherent in surface spectral imagery, yielding an accurate and large-scale breakdown of soil organic carbon.
From University of Illinois College of Agricultural, Consumer & Environmental Sciences
View Full Article
Abstracts Copyright © 2022 SmithBucklin, Washington, DC, USA
No entries found