<link rel="stylesheet" href="styles.cf390c6bfbe7555f.css">

Understanding within-field variability and monitoring crop stress to inform site-specific input management

Loading...
Thumbnail Image

Date

Journal Title

Journal ISSN

Volume Title

Publisher

Abstract

Site-specific management of water and nutrients is central for sustaining crop productivity while limiting input losses. It requires knowing how crop status and soil conditions vary within a field and being able to measure that variation and to act on it. This thesis addresses two sides of that problem: detecting crop water stress through proximal sensing and characterizing within-field variability in nitrogen and seeding rate response in corn. The first chapter synthesizes 26 peer-reviewed studies published between 2013 and 2026, a total of 841 articles were selected following PRISMA guidelines, on proximal sensing for water stress detection in agronomic crops. Using the CIMO framework, the review traces sensing technologies deployed at leaf or canopy level detect water stress. Hyperspectral sensing dominated the literature (77%), followed by thermal sensing (27%) and fluorescence (15%). The central finding is that the shortwave infrared region, which has the strongest physical basis for direct water content detection through O-H absorption at 1450 and 1940 nm, was used in only 31% of studies. Because atmospheric water vapour does not interfere the sensor-to-canopy path, proximal sensing is the one platform on which these features can be measured directly, and also the platform on which they are least used. Spectral coverage mattered more than model complexity for detection accuracy. Reporting of the stress treatment itself was often incomplete, with twelve of the 26 studies with no exposure duration, which limits how far the reported accuracies can be generalized. The second chapter moves from sensing to management, evaluating corn yield response to variable rate nitrogen and seeding on a 22.3 ha rainfed field near Parsons, Kansas, during the 2023 growing season. The trial was conducted as part of the Data Intensive Farm Management project using a 6 × 6 Latin square design with six in-season nitrogen rates and six seeding rates. Nitrogen rate, seeding rate and their interaction all significantly affected grain yield. Tree-based machine learning models explained more yield variability (R2 = 0.52) than linear approaches (R2 = 0.28), indicating that the relationship between soil properties and yield on this field is non-linear. Cation exchange capacity was the strongest yield predictor, and its negative relationship with yield reflected the poorly drained Parsons silt loam. Zone-specific nitrogen management returned $37 ha-1 above the grower’s uniform rate of 40 kg N ha-1, improving net return in two of three management zones. Together, the two chapters approach the same problem from opposite ends. The second chapter found that soil and terrain properties explained only about half of within-field yield variation. That analysis could not measure crop status directly, and weather and canopy condition were treated as spatially uniform across the field. It identified the absence of crop status measurements as a limitation, since weather and canopy condition were spatially uniform across the field. The first chapter describes the sensing methods capable of supplying that missing information and identifies which spectral regions will be prominent. This points toward sensor-based variable rate application, where crop status measured during the season sets the rate instead of a prescription fixed before planting.

Description

Keywords

Precision agriculture, Site-specific management, Variable rate application, Water stress detection, Proximal sensing, Within-field variability

Graduation Month

December

Degree

Master of Science

Department

Department of Agronomy

Major Professor

Gaurav Jha; P.V. Vara Prasad

Date

Type

Thesis

Citation

Endorsement

Review

Supplemented By

Referenced By