An efficient and transferable remote sensing spectral index for regional corn mapping
Year: 2025
Type: article
Abstract: Accurate and timely corn mapping at regional scales supports effective agricultural management and policy development. However, traditional data-driven approaches rely heavily on extensive ground-truth samples and have limited applicability in data-scarce areas. And there is still a lack of simple and practical approaches that require few or no field observations for corn mapping. Therefore, this study presents a novel framework that integrates an improved Dual-peak Canopy Nitrogen Index (DCNI) with Green Chromatic Coordinate (GCC) analysis. The refined DCNI captures nitrogen-related spectral dynamics to distinguish corn from spectrally similar crops such as soybean and sorghum during peak growth stages. The GCC analysis identifies the optimal mapping window, mitigating the effects of cloud cover and temporal data gaps for regional corn mapping. We applied this improved index to three selected but representative corn growing regions: Nenjiang in China, Pocahontas in the United States, and Mayenne in France. Using the optimized DCNI thresholds, binary classification was performed to reduce both commission and omission errors. In the pixel-level validation for 2021, our DCNI index achieved better overall accuracy and F1 scores compared to the classical random forest model and its variants. Interannual tests from 2020 to 2022 also showed stable performance and strong agreement with official statistics, with all coefficients of determination R 2 above 0.96. We also investigate uncertainties arising from data interpolation, limited field samples, and mixed cropping patterns, and propose the integration of higher-frequency satellite observations and multisource data fusion to improve early season monitoring and broaden large-scale applicability. The proposed framework requires minimal crop samples and computational resources, providing a simple, practical alternative for regional corn mapping with robust transferability. • DCNI index can distinguish corn from other spectrally similar crops. • Integrating GCC methods to identify the DCNI-based corn mapping window. • DCNI delivers comparable accuracy to traditional corn mapping algorithms. • DCNI shows strong transferability with minimal sample requirements.
Source: Science of Remote Sensing
Authors: Muyan Han, Ji Chai
Institutions: University of Manchester, China University of Geosciences
Language: English
Subfield: Ecology
Field: Environmental Science
Domain: Physical Sciences
SDG: Zero hunger
Open Access status: gold
Estimated APC paid: $2,180
OpenAlex ID: W4414997007