Development of a Cost-Effective Machine Learning-Based Device for Predicting Nitrous Oxide Emissions in Agricultural Fields

Agriculture and Agri-Food Canada has initiated a program to reduce fertilizer N2O emissions by 30 % from 2020 to 2030 [14]. Accurate monitoring of agricultural N2O emissions is crucial for understanding their environmental impact and implementing effective mitigation strategies.
Current research mainly focuses on measuring N2O emissions from the ground. The spectroscopic-based LI-COR sensor is a typical instrument for measuring N2O emissions [19]. The instrument is expensive and typically costs a few hundred thousand to own. Therefore, there is a lack of cost-effective sensors that can predict nitrous oxide emissions from farmland, taking into account all the factors such as soil temperature, soil moisture, and nitrate level in the soil simultaneously.
This research aims to apply a machine learning technique to train a novel device using the N2O emissions data from a LI-COR sensor, as well as the amount of fertilizers in the soil, soil moisture, and soil temperature. After the device is trained, it can then be deployed to predict N2O emissions from agricultural fields using only data on nitrogen fertilizers in the soil, soil moisture, and soil temperature.

Faculty Supervisor:

Tet Yeap

Student:

Partner:

Invest Ottawa

Discipline:

Engineering

Sector:

Information and cultural industries; Management of companies and enterprises; Professional, scientific and technical services; Transportation and warehousing

University:

University of Ottawa

Program:

Accelerate