An Integrated Approach for Mapping Multi-Category Abandoned Cropland Based on Swin-Unet, Land-Use Trajectories and Phenological Features in a Double-Cropping Rice Region of Southern China
Комплексный подход к картированию заброшенных сельскохозяйственных угодий различных категорий на основе Swin-Unet, траекторий землепользования и фенологических признаков в регионе двойного выращивания риса на юге Китая
2026-09-09
SCID: 54.1/d5jru5d4
Discuss with AI
Swin-UNetblock-adaptive Otsu thresholdingland-use trajectoriesmulti-category abandoned cropland mappingphenological features
Figures from the paper
Abstract (AI)
Cropland abandonment directly undermines the regional food supply, requiring accurate monitoring to inform protection and recultivation policies. However, multi-category abandoned-cropland mapping remains challenging in southern China’s double-cropping rice regions because fragmented fields, frequent cloud/rain conditions, and complex phenology can obscure abandonment signals. Focusing on a typical double-cropping rice region in southern China, this study first used Swin-UNet to generate annual land-use maps from 2018 to 2022 and derive interannual trajectories. It then identified non-cultivated cropland within each year using Normalized Difference Vegetation Index (NDVI) data from key phenological periods and block-adaptive Otsu thresholding. Overlaying cultivation status with land-use trajectories further subdivided non-cultivated cropland into construction-occupied abandoned, naturally abandoned cropland, and fallow cropland. Finally, the spatial differentiation of each type and the potential grain-production gains from recultivating naturally abandoned cropland were analyzed. Results showed that annual classification accuracies from 2018 to 2022 ranged from 94.7 to 95.5%, supporting reliable cropland change trajectory extraction. From 2019 to 2022, total abandoned area decreased from 546.60 km2 to 486.73 km2. Naturally abandoned cropland remained dominant but declined mainly because 18.5–21.7% of it was converted back to stable cropland annually, while construction-occupied abandonment increased and fallow cropland remained relatively stable. Natural abandonment concentrated on moderate slopes near settlements and roads, with a theoretical grain-production potential equivalent to 12.88–15.14% of Maoming’s annual grain output. This study advances abandoned-cropland monitoring from area estimation to type-specific process interpretation and recultivation-priority assessment in southern double-cropping rice regions.
Key Findings
1
A Swin-UNet and phenology-based framework mapped multi-category abandoned cropland by integrating annual land-use trajectories, NDVI, and block-adaptive Otsu thresholding.
2
Annual land-use classification accuracy reached 94.7–95.5% from 2018 to 2022, enabling reliable extraction of cropland change trajectories.
3
Natural abandonment was concentrated on moderate slopes near settlements and roads, with recultivation potential equivalent to 12.88–15.14% of Maoming’s annual grain output.
4
Naturally abandoned cropland remained the dominant category but declined as 18.5–21.7% was annually converted back to stable cropland; construction-occupied abandonment increased, while fallow cropland remained relatively stable.
5
Total abandoned cropland decreased from 546.60 km² in 2019 to 486.73 km² in 2022.
Research Object
Multi-category abandoned cropland in a double-cropping rice region of southern China
Research Subject
The spatial distribution, land-use trajectories, phenological identification, type-specific dynamics, and grain-production potential of abandoned cropland
Publication Details
Publication Date
2026-09-09
Journal
Publisher
ISSN
Open access PDF
Access Type
Author Information
Download PDF
Subscribe to digest
References available in scid.ai3
A Global Review of Monitoring Cropland Abandonment Using Remote Sensing: Temporal–Spatial Patterns, Causes, Ecological Effects, and Future Prospects2025
Monitoring Cropland Abandonment in Hilly Areas with Sentinel-1 and Sentinel-2 Timeseries2022
Monitoring Abandoned Cropland in Fragmented Mountainous Landscapes Based on the ML-LandTrendr Framework2026