Semester of Graduation

Summer 2024

Degree

Master of Science (MS)

Department

School of Plant, Environmental, and Soil Sciences

Document Type

Thesis

Abstract

High soil moisture from frequent excessive precipitation can reduce maize (Zea mays L.) yield and increase nitrogen (N) loss. Early-stage soybeans (Glycine max) are also susceptible to damage under these conditions. Traditional methods for assessing N status in maize and plant stands in soybean are labor-intensive and time-consuming. In this study, UAV remote sensing was used to assess maize N status under different N rates and excessive soil moisture. For soybeans, we compared a blob detection method with the YOLOv8 model to estimate plant density using ground images from a handheld camera. Additionally, YOLOv8 performance was evaluated on UAV imagery to estimate plant density and identify areas needing replanting. Experiments followed a split-plot design with four replications. Main plots were designated by the soil moisture conditions, with subplots for different N rates in maize and different seeding rates in soybean. UAV images were captured using multispectral and RGB sensors. Results showed significant effects of N rates on maize yield across site x years, with flooding effects significant at RRS but not at CRS. Flooding treatment also influenced the estimated optimum N rates. Early June NDRE readings correlated highly (0.74) with yield and aligned with N response curves. In soybean experiments, YOLOv8 outperformed the blob detection method, estimating plant densities with over 90% accuracy using aerial images. These studies underscore the value of sensing technology for decision-making in nutrient management and the potential of YOLOv8 for replanting decisions in soybean under adverse weather conditions.

Date

7-15-2024

Committee Chair

Setiyono, Tri

DOI

https://doi.org/10.31390/gradschool_theses.6006

Download

Share

COinS