Optimizing soybean production via soil health and grain quality assessment using UAV multispectral imaging

Authors: ZHOU, JIANFENG - University Of Missouri; GHADWAL, MANOJ - University Of Missouri; REINBOTT, TIMOTHY - University Of Missouri; Sudduth, Kenneth

Submitted to: ASA-CSSA-SSSA Annual Meeting Abstracts
Publication Type: Abstract Only
Publication Acceptance Date: 11/13/2024
Publication Date: 11/13/2024
Citation: Zhou, J., Ghadwal, M., Reinbott, T.M., Sudduth, K.A. 2024. Optimizing soybean production via soil health and grain quality assessment using UAV multispectral imaging [abstract]. 2024 ASA-CSSA-SSSA International Annual Meeting, November 10-13, 2024, San Antonio, Texas. Paper No. 161493. Available: https://scisoc.confex.com/scisoc/2024am/meetingapp.cgi/Paper/161493

Interpretive Summary:

Technical Abstract: Crop growth, yield, and the nutritional content are significantly influenced by soil. Despite the consistent increase in crop yield, there is a simultaneous degradation of soil and grain quality, posing challenges to the sustainability of agriculture. Existing soil health assessment methods are laborious and challenging to implement on a large scale. To address this issue, a novel approach was undertaken in this study, aiming to develop a method for quantifying soil health and grain quality using unmanned aerial vehicles (UAVs) multispectral imaging and advanced machine learning techniques. Two soybean fields with varying soil fertility and different cropping system such as cover crops and no cover crops were considered for this research experiment. A series of weekly UAV imagery (including high-resolution RGB, near-infrared (NIR), and red-edge) were systematically collected throughout the crop growth cycle, from emergence to harvest. For soil health and fertility assessment soil samples from 120 locations were collected at two depths (0 - 3 inches and 3-6 inches). Additionally, soil moisture was monitored for both sites at multiple depths for every 5 minutes using soil moisture sensors. Yield data was obtained from the combine harvester which was equipped with a yield monitor system to generate spatial yield maps. Weather data, including growing degree days, air temperature and precipitation, was retrieved from nearby weather stations. UAV images were processed and converted into ortho mosaics and different vegetative indices like NDVI and NDRE were established from multispectral ortho mosaics. This intense aerial imaging and ground-based dataset was then used to develop different machine learning models for soil health and fertility and grain quality prediction. The study aims to provide practical recommendations for enhancing soil health, nutrient management, and grain quality.