Zheng et al. (2026) Interpretable Multi-Year Winter Wheat Mapping with Sentinel-1/2 Time Series: SHAP-Based Feature Selection and Bayesian-Optimized Machine Learning

⚠️ Warning: This summary was generated from the abstract only, as the full text was not available.

Identification

• Journal: Remote Sensing
• Year: 2026
• Date: 2026-09-12
• Authors: Wenhao Zheng, Yan Xu, Lixiran Yu, Hongfei Tao, Qiao Li, 龚荫成, Yuwei Jiang, Quanjiu Wang
• DOI: 10.3390/rs18183137

Research Groups

• Institute of Remote Sensing and Digital Earth, Chinese Academy of Sciences
• Xinjiang University

Short Summary

This study presents an interpretable cross-year transfer approach for winter wheat mapping using Sentinel-1/2 time-series imagery, achieving consistent classification performance across evaluated historical years within the same irrigation district.

Objective

• Develop a model that can accurately map winter wheat areas without requiring repeated sample collection, feature selection, and parameter tuning for each year.

Study Configuration

• Spatial Scale: Tailan River Irrigation District, Xinjiang, China
• Temporal Scale: 2023-2025

Methodology and Data

• Models used: Bayesian-optimized Random Forest model
• Data sources: Sentinel-1/2 time-series imagery, NDVI, NDre1, EVI

Main Results

• The model achieved overall accuracies of 93.82% in the source-year validation (2025) and 92.13% and 91.01% in the target-year tests (2023 and 2024).
• SHAP-based selection retained 28 informative variables from 189 temporal-change features.
• May-to-June vegetation index change rate features derived from NDVI, NDre1, and EVI provided consistent phenological information.

Contributions

• This study demonstrates that a model developed from a single source year can maintain consistent classification performance across evaluated historical years within the same irrigation district.

Funding

• Not specified in the paper.

Citation

@article{Zheng2026Interpretable,
  author = {Zheng, Wenhao and Xu, Yan and Yu, Lixiran and Tao, Hongfei and Li, Qiao and 龚荫成 and Jiang, Yuwei and Wang, Quanjiu},
  title = {Interpretable Multi-Year Winter Wheat Mapping with Sentinel-1/2 Time Series: SHAP-Based Feature Selection and Bayesian-Optimized Machine Learning},
  journal = {Remote Sensing},
  year = {2026},
  doi = {10.3390/rs18183137},
  url = {https://doi.org/10.3390/rs18183137}
}