10-M mapping and analysis of major crop planting Patterns in henan province in China

To improve crop mapping in complex agricultural regions for food security, this study develops a seasonal monitoring framework, treating the growing seasons as independent classification tasks based on specific temporal windows, in Henan Province, China, using 2023 Sentinel imagery. It employs separate feature sets for summer-harvested (winter wheat) and autumn-harvested crops (maize, rice, peanut, soybean), using an XGBoost algorithm for classification and a SHAP model to overcome the black-box nature of the algorithm and identify key phenological markers. The framework demonstrated high effectiveness, achieving cross-validation consistency rates of 85.18% for winter wheat and 90.08% for maize. The resulting maps show that a winter wheat-summer maize rotation is the predominant cropping pattern in the province. The study also found that driving factors for classification differ by season: winter wheat identification depends on unique phenological signals (e.g., March NIR), whereas autumn crops are distinguished by a dual mechanism of macro-geographical latitude and specific spectral indices. This research confirms the strategy's robustness for fine-scale mapping. Bangladesh J. Bot. 55(1): 191-199, 2026 (March)