Replication Data for: A Framework for Predicting Wheat Grain Quality Using Hyperspectral and Multispectral Derived Matrices with Statistical Learning Methods

This research focuses on the development and evaluation of remote sensing approaches for the assessment of wheat grain quality and productivity. Grain yield, grain nitrogen content, and grain protein content are key traits in wheat breeding and crop management; however, their measurement is often labor-intensive, time-consuming, and costly. Advances in precision agriculture and unmanned aerial vehicle (UAV) technologies provide new opportunities to estimate these traits rapidly and non-destructively through spectral information collected during the growing season. Field trials were conducted over four consecutive growing seasons in Northwest Mexico under contrasting agronomic management conditions, including different tillage systems, irrigation regimes, and nitrogen fertilization levels. Hyperspectral and multispectral imagery acquired at multiple growth stages was used to derive spectral and thermal variables associated with crop performance and grain quality attributes. Several modeling strategies combining vegetation indices, spectral bands, and thermal information were investigated using different statistical and machine learning techniques. Their predictive ability was assessed across environments and growth stages to identify robust approaches for trait estimation. The outcomes of this work contribute to the advancement of high-throughput phenotyping methodologies and support more efficient wheat breeding and nitrogen management practices through the integration of UAV-based remote sensing technologies.