Date of Award

8-2026

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Environmental Engineering and Earth Science

Committee Chair/Advisor

Tom Owino

Committee Member

Jose O. Payero

Committee Member

A.Bulent Koc

Committee Member

Yu-Bo Wang

Committee Member

Tom Dodd

Abstract

Over-irrigation and excessive nitrogen (N) fertilization contribute to the depletion of natural resources and elevated agricultural production costs. Although the integration of automated water and N management in center-pivot systems could potentially reduce the environmental impact and production costs, this technology has mostly been unexplored for large-scale cotton production. This dissertation addresses three main objectives: (i) to develop an automated irrigation-fertigation system for a center pivot in cotton production; (ii) to evaluate the field performance of the automated system, including sensor data acquisition, spatial and temporal variability of sensor data, accuracy and application uniformity of the variable-rate N injection system across defined management zones, and a performance comparison between gyroscopic sensors and Real-Time Kinematic Global Positioning System (RTK-GPS) sensors for recording the spatial locations of the center pivot; and (iii) to simulate and compare county-scale Irrigation Water Requirements (IWRs) under Variable-Rate Irrigation (VRI) and Fixed-Rate Irrigation (FRI) strategies across 85 center-pivot irrigated cotton fields in nine counties of South Carolina, using USDA-NRCS soil survey data, daily weather records, and FAO-56 guideline.

Chapter 1 presents a Systematic Literature Review (SLR) of ground-based sensors and vegetation indices used for crop N status estimation. The results of the SLR indicated that the reflectance sensors were the most frequently used tools, while the Normalized Difference Vegetation Index (NDVI), Normalized Difference Red Edge (NDRE), and Ratio Vegetation Index (RVI) were the most applied indices. Key sensor-related sources of error included environmental effects, sensor saturation, dependence on N-rich reference strips, and temporal and crop-specific responses. The review identified critical research gaps and underscored the need for integrated, real-time sensing and decision-support systems to improve the accuracy of crop N status estimation and fertigation management.

Chapter 2 describes the development of an automated site-specific irrigation–fertigation system retrofitted onto a center-pivot for large-scale cotton production, representing the core contribution of this research. The prototype was developed and tested at the Clemson University Edisto Research and Education Center (EREC) in South Carolina during the 2023–2025 cotton growing seasons. The automation architecture consisted of distributed sensor nodes, including Soil-Water Potential (SWP), NDVI, and position sensors, coordinated by a Central Control System (CCS) through a Wireless Sensor Network (WSN) utilizing LoRa radio communication. SWP sensors installed at multiple depths across distinct irrigation zones provided continuous SWP measurements, while five canopy-mounted NDVI sensors distributed along the pivot length enabled real-time assessment of N spatial variability. Irrigation was automatically triggered when SWP readings reached a predefined threshold, while N application was actuated through individually controlled injection pumps when the recorded average NDVI within a zone fell below a baseline. The timing and quantity of N fertilizer delivery were programmed based on a cotton N accumulation capacity curve and a pulse-based injection technique. Two position sensing technologies, a gyroscope and RTK-GPS sensors, were evaluated for tracking pivot location and enabling spatially targeted irrigation/fertigation zones. The integrated framework enabled real-time, zone-specific management of water and N, effectively addressing spatial heterogeneity in soil and crop conditions while reducing reliance on uniform application practices.

Chapter 3 presents an evaluation of the automated system's field-scale engineering performance across two growing seasons. Weighted root-zone SWP thresholds were used to trigger irrigation events, while NDVI measurements referenced to a seasonal baseline and a crop N accumulation capacity approach regulated N applications. Results confirmed reliable acquisition and wireless transmission of SWP, NDVI, and positional data throughout the seasons. The automated irrigation system successfully captured spatial SWP variability and consistently triggered solenoid valves at or below the predefined –50 kPa threshold. NDVI sensors effectively characterized crop growth dynamics, including early-season variability, mid-season saturation, and late-season decline across five sensors operating under differentiated N application rates. The variable-rate N injection system demonstrated acceptable uniformity across management zones, and the innovative NDVI-based algorithm eliminated the need for N-rich reference strips. The positioning system enabled accurate identification of irrigation/fertigation zones, ensuring proper synchronization between sensor inputs and actuator responses. The RTK-GPS sensors outperformed gyroscope sensors in terms of reliability and ease of operation.

Chapter 4 extends the analysis by evaluating IWRs at a broader spatial scale through simulation. The IWRs were estimated and compared under two strategies of VRI and FRI across 85 center-pivot irrigated cotton fields in nine counties of South Carolina. County-scale IWR simulations were conducted using USDA-NRCS soil survey data and daily weather records, comparing the FAO-56 single and dual crop coefficient (Kc) methods. The dual Kc method consistently produced higher crop evapotranspiration (ETc) estimates, averaging 665 mm compared to 604 mm for the single Kc approach, and yielded higher seasonal IWR estimates of 269 mm vs. 191 mm.  Comparisons between VRI and FRI strategies revealed no significant differences in seasonal ETc and IWR, likely attributable to relatively uniform soils and frequent rainfall pattern in the study region. Despite the limited differences observed in this case, the methodology developed in this chapter provides a robust and transferable framework for evaluating irrigation management strategies across varying spatial and climatic conditions.

Author ORCID Identifier

0000-0003-2126-4219

Download

Available for download on Tuesday, August 31, 2027

Share

COinS