Thesis Format
Integrated Article
Degree
Master of Science
Program
Geography and Environment
Supervisor
Wang, Jinfei
Abstract
Monitoring crop productivity is crucial in precision agriculture, often using biomass and yield as metrics to measure crop health and growth status. This thesis aims to predict dry above-ground biomass using Unmanned Aerial Vehicle (UAV) multispectral imagery, derived vegetation indices (VI), plant height, leaf area index (LAI), and plant nutrient content ratios. Additionally, the thesis tests the viability of VENμS satellite data as an alternative to other popular multispectral satellite data for predicting winter wheat yield. Conducted in two winter wheat fields in southwestern Ontario, Canada, the study employed Random Forest (RF) and Support Vector Regression (SVR) machine learning models with various variable combinations. The results demonstrate that the approach in biomass estimation was accurate and provided valuable insights into the applicability of biochemical parameters. Furthermore, VENμS produced promising yield prediction results, proving to be a better satellite platform compared to other publicly available satellite data for yield prediction.
Summary for Lay Audience
The main source of food for the world’s population is agriculture. As the global population grows, the strong demand for food sources and food security has emphasized the need for efficient and sustainable agricultural practices. Advances in technology have led to the development of precision agriculture, which involves applying technology and agricultural principles to strategically manage resources in all aspects of agricultural production, aiming to maximize crop performance and maintain environmental sustainability.
Remote sensing is the science of obtaining information about an object, area, or phenomenon through the analysis of data acquired by a device from a distance. In precision agriculture, this often involves using satellite imagery or imagery captured by cameras mounted on unmanned aerial vehicles (UAVs). Specialized sensors installed on these platforms can detect light emissions and reflections from the Earth’s surface that are beyond the human eye’s visible spectrum. For example, while humans cannot see near-infrared light, the sensors can detect and record the amount of near-infrared radiation reflected by plants, providing insights into their health and vigor. The reflectance data collected by these sensors can be transformed into indices that convey specific information about the plants, using formulas known as vegetation indices.
Biomass and yield are common metrics for evaluating the performance and productivity of crops. Accurately forecasting these metrics allows farmers to respond early and effectively during the growing stages to maximize harvest output. This is especially crucial for farmers in southern Ontario, where there is only one growing season each year. In this thesis, prediction models based on machine learning algorithms were developed to identify which variables are most important for accurately predicting biomass and yield. The results provide farmers with key information on the factors that most significantly influence these metrics, enabling them to monitor and manage these variables to effectively manage their crop production.
Recommended Citation
Chiu, Marco S. Mr., "Winter Wheat Biomass and Yield Estimation using Unmanned Aerial Vehicle-based and VENµS Satellite Imagery with Machine Learning Techniques" (2024). Electronic Thesis and Dissertation Repository. 10357.
https://ir.lib.uwo.ca/etd/10357
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.