Date of Award:


Document Type:


Degree Name:

Doctor of Philosophy (PhD)


Civil and Environmental Engineering

Committee Chair(s)

Sierra N. Young


Sierra N. Young


Scott B. Jones


Alfonso R. Torres


Burdette Barker


Jason Ward


Near-surface soil moisture, or the water content within the soil, is important for understanding the interactions between land and the atmosphere, and for monitoring plants in agricultural settings. However, soil moisture can be highly variable within the same field and varies considerably with time. The challenge involved with measuring soil moisture is that traditional techniques that rely on obtaining large samples are labor and time-intensive, especially for large fields. Developments in sensor technologies have allowed users to record the soil moisture regularly at the points where the sensors are installed. However, to understand how soil moisture changes across a field from sensors installed at single points, there needs to be a large number of sensors installed, which are not easily moved and are expensive. Remote sensing approaches that use imagery from satellites and drones have been used to develop soil moisture prediction models. Developing these models requires measurements from the field to validate them. However, collecting data from large fields on a regular basis is challenging. Also, remote sensing models using machine learning techniques tend to be “black box models”, or models that do not reveal any information about their inner workings and may not have any physical significance to soil moisture. To address the challenges presented here, a first-of-its-kind, cost-effective fully autonomous drone payload was developed to measure near-surface soil moisture. A new validation technique for the payload sensor measurements was developed that only relies on two pieces of data–depth of insertion and sensor signal–to obtain a calibrated moisture content. Finally, a soil moisture prediction model was developed using the soil line concept, which is a linear relationship between bare soil reflectance observed in two different wavebands, combined with machine learning models to add physical meaning to the models. The three techniques developed in this dissertation address the challenges in near-surface soil moisture measurements and represent significant progress toward automating critical data collection across large fields in agriculture.



Creative Commons License

Creative Commons Attribution-Noncommercial 4.0 License
This work is licensed under a Creative Commons Attribution-Noncommercial 4.0 License