Development of a Machine Learning System for Irrigation Decision Support with Disparate Data Streams

Authors

Eric Wilkening, University of Nebraska-LincolnFollow

First Advisor

Derek M. Heeren

Date of this Version

12-2023

Citation

A thesis presented to the faculty of the Graduate College at the University of Nebraska in partial fulfillment of requirements for the degree of Master of Science

Major: Agricultural and Biological Systems Engineering

Under the supervision of Professor Derek M. Heeren

Lincoln, Nebraska, December 2023

Comments

Copyright 2023, Eric Wilkening

Abstract

In recent years, advancements in irrigation technologies have led to increased efficiency in irrigation applications, encompassing the adoption of practices that utilize data-driven irrigation scheduling and leveraging variable rate irrigation (VRI). These technological improvements have the potential to reduce water withdrawals and diversions from both groundwater and surface water sources. However, it is vital to recognize that improved application efficiency does not necessarily equate to increased water availability for future or downstream use. This is particularly crucial in the context of consumptive water use, which refers to water consumed and not returned to the local or sub-regional watershed, representing a critical consideration in water conservation.

Variable Rate Irrigation (VRI) allows for management of in-field spatial variability of water requirements. To comprehensively assess the impact of VRI on consumptive water use and pumping, this study evaluated multiple site-years of field research data. The research employed a previously developed metric known as the "consumptive use ratio" which quantifies the change in consumptive use relative to the change in irrigation water applied. The study further developed a “marginal consumptive use ratio,” which was utilized to analyze different irrigation management scenarios, including VRI with zone control prescriptions, across multiple years and field sites in Eastern and Western Nebraska.

Beyond the realm of current irrigation practices, this research explored the integration of machine learning to expedite irrigation recommendations while enhancing water usage efficiency. Leveraging two years of weather and agronomic data, machine learning algorithms were trained and tested. A novel method of this model is its recommendation of a "Latest Date" to indicate when to irrigate at a fixed application depth, allowing for flexibility across different irrigation systems.

This approach, when applied in the context of irrigation scheduling, showed promise, with the in-season validation of the model achieving a root-mean-square-error (RMSE) of 2.23. The research's future endeavors include expanding the training dataset and finer model tuning, supporting a goal of enhancing the model's practical applicability in agriculture and irrigation management. This comprehensive study bridges the gap between irrigation and the technology of machine learning for more efficient and sustainable water resource management in agriculture.

Advisor: Derek M. Heeren