Capturing harvest-induced signatures using a novel multi-temporal index for Zanthoxylum armatum mapping

• Constructed a large multi-crop time-series dataset with 56,472 labeled samples. • Identified optimal phenological stages for Z. armatum recognition. • Developed a MHZ index achieving 92.40% accuracy and 83.90% F1-score. • Produced a 2022 high-precision Z. armatum distribution map for Chongqing. Large-scale crop mapping is fundamental to cultivation management and yield assessment, yet reliable identification of woody economic crops remains challenging due to complex phenology and management-induced disturbances. Zanthoxylum armatum ( Z. armatum ), a key specialty crop in China, lacks a high-precision, region-scale distribution map, particularly in Chongqing—its origin and primary production region. Here, we construct a large phenology-aware dataset comprising 56,472 samples across Z. armatum and major competing crops, and systematically investigate its temporal discriminability in satellite time-series imagery. We identify a narrow but highly informative recognition window surrounding the harvesting stage, during which harvesting-driven pruning activities induce pronounced temporal signals that substantially enhance crop separability. Building upon this insight, we propose a novel multi-temporal, harvest-induced vegetation index for regional-scale mapping of Z. armatum (MHZ; R 835 R 665 − 1 ) that explicitly captures these phenology–management interactions. Remarkably, this single-index representation achieves performance comparable to full-spectral models, yielding an overall accuracy of 92.40% and an F1 score of 83.90%, while offering superior interpretability and computational efficiency. Applying the proposed framework, we generate the first high-precision city-wide map of Z. armatum in Chongqing for 2022, estimating a total planting area of 1105.87 km 2 , with Jiangjin District contributing 26.87%. Independent validation against official statistics demonstrates strong agreement, confirming the robustness and practical reliability of the results. Overall, this study establishes multi-temporal MHZ as a compact yet powerful phenological descriptor for Z. armatum mapping and provides a scalable, transferable framework for large-scale agricultural monitoring in complex terrain regions. The 2022 Z. armatum distribution map for Chongqing is publicly available at: 10.6084/m9.figshare.30344584