Dissertation Defense Announcement for Mohammad Abdus Shahid Rafi – 11/26/2025 at 1 PM

Dissertation Title:
Remote Sensing Techniques for Soil Moisture Retrieval and Crop Yield Estimation: Employing Multi-Sensor Receiver Systems from UAS and ML for Precision Agriculture

When:
26 November 2025, 1:00 PM.

Where:
Simrall 228

Candidate:
Mohammad Abdus Shahid Rafi

Degree:
Doctor of Philosophy in Electrical and Computer Engineering.

Committee Members:
Dr. John E. Ball, Dr. Ali C. Gurbuz, Dr. Volkan Senyurek, and Dr. Junming Diao.

Abstract:
Accurate soil moisture (SM) measurement and crop yield estimation are critical components of modern precision agriculture (PA) management. Efficiently observing SM at high resolution on a sub-field scale can enhance irrigation planning and management, leading to improved yields and product quality while also conserving environmental resources. Yield predictions provide insights into expected production, facilitating optimized resource allocation, improved agricultural management strategies, and enhanced profitability. Unmanned Aircraft Systems (UAS) based multi-sensor receiver systems offer promising solutions in obtaining high-resolution SM measurements across large fields where satellite remote sensing falls short, as well as collecting necessary data over corn and cotton for timely yield estimation of these important crops. First, this study has developed a custom-made UAS-based passive GNSS-R (Global Navigation Satellite Systems Reflectometry) receiver system for soil moisture (SM) retrievals at the sub-field scale (higher spatial and temporal resolution) to identify relevant features and normalization techniques. From the three years of data collection over 2.31 hectares of corn and cotton fields, incorporating GNSS-R, multispectral imaging, LiDAR (Light Detection and Ranging), and in-situ SM measurements, the impact of receiver antenna characteristics, surface factors, and GNSS constellations on GNSS-R measurements are investigated. The results highlighted both the potential and the challenges of using a low-cost GNSS-R receiver system from a mid-size small UAS platform for accurate and reliable high-resolution SM measurement in PA. Next, A multi-sensory dataset was collected using multispectral cameras and LiDAR sensors mounted on UAS, along with soil moisture and temperature data from volumetric probes and environmental data from a nearby weather station. Over four years, more than 30 features were extracted weekly from five major categories, with 235 ground truth yield records from plots in the field. This subsequent study outlines the methodology for feature selection and investigates the application of machine learning (ML) techniques, including Feedforward Neural Networks (FNN), Long Short-Term Memory (LSTM), and Random Forest (RF) models, for predicting corn and cotton yields using spatiotemporal data. Using percentile root mean square error (RMSE) and mean absolute error (MAE) as performance metrics, the study found that LSTM produced lower field-wise errors compared to other models and validation, indicating superior performance in predicting yields across selected weeks. The proposed ML-based approach, validated through year-based and field-wise cross-validation methods, demonstrates the effectiveness of using UAS-collected multi-sensor data for accurate crop yield estimation in PA.