Detecting Abandoned Cropland in Monsoon-Influenced Regions Using HLS Imagery and Interpretable Machine Learning
Выявление заброшенных сельскохозяйственных угодий в регионах, находящихся под влиянием муссонов, с использованием снимков HLS и интерпретируемого машинного обучения
2025-11-24
SCID: 54.1/795v6sta
Discuss with AI
Balanced Bagging ClassifierHLS imagerySHAP analysisXGBoostabandoned cropland detection
Figures from the paper
Abstract (AI)
Abandoned cropland has been expanding due to complex socio-economic factors such as urbanization, demographic shifts, and declining agricultural profitability. As abandoned cropland simultaneously brings ecological, environmental, and social risks and benefits, quantitative monitoring is essential to assess its overall impact. Satellite image-based spatial data are suitable for identifying spectral characteristics related to crop phenology, and recent research has advanced in detecting large-scale abandoned cropland through changes in time-series spectral characteristics. However, frequent cloud covers and highly fragmented croplands, which vary across regions and climatic conditions, still pose significant challenges for satellite-based detection. This study combined Harmonized Landsat and Sentinel-2 (HLS) imagery, offering high temporal (2–3 days) and spatial (30 m) resolution, with the eXtreme Gradient Boosting (XGBoost) algorithm to capture seasonal spectral variations among rice paddy, upland fields, and abandoned croplands. An XGBoost model with a Balanced Bagging Classifier (BBC) was used to mitigate class imbalance. The model achieved an accuracy of 0.84, Cohen’s kappa 0.71, and F2 score 0.84. SHapley Additive exPlanations (SHAP) analysis identified major features such as NIR (May–June), SWIR2 (January), MCARI (September), and BSI (January–April), reflecting phenological differences among cropland types. Overall, this study establishes a robust framework for large-scale cropland monitoring that can be adapted to different regional and climatic settings.
Key Findings
1
A Balanced Bagging Classifier integrated with XGBoost mitigates class imbalance in detecting fragmented abandoned cropland under frequent cloud cover.
2
SHAP identified NIR in May–June, SWIR2 in January, MCARI in September, and BSI from January–April as major discriminative features.
3
The framework supports scalable cropland monitoring and is designed for adaptation across regional and climatic conditions.
4
The model achieved 0.84 accuracy, 0.71 Cohen’s kappa, and an F2 score of 0.84.
5
The study combines high-frequency, 30 m HLS imagery with XGBoost to distinguish rice paddies, upland fields, and abandoned cropland from seasonal spectral variation.
Research Object
abandoned cropland and other cropland types (rice paddy and upland fields) in monsoon-influenced regions
Research Subject
the spatial detection and discrimination of cropland abandonment based on seasonal spectral and phenological differences in HLS time-series imagery
Publication Details
Publication Date
2025-11-24
Journal
Publisher
ISSN
Open access PDF
Access Type
Author Information
Download PDF
Subscribe to digest