Spatially divergent cropland disturbance and driving mechanisms in the transboundary Heilongjiang River basin, 1990–2021
Пространственная неоднородность нарушений сельскохозяйственных угодий и механизмы их формирования в трансграничном бассейне реки Амур, 1990–2021 гг.
2026-06-18
SCID: 54.1/r8zp3xrp
Discuss with AI
Heilongjiang River BasinLandTrendr algorithmLandsat time-series imagerycropland disturbance monitoringdisturbance driver attribution
Figures from the paper
Abstract (AI)
Cropland disturbance monitoring is essential for sustainable agriculture and ecosystem conservation, yet remains challenging over large areas due to the complexity of land use change processes. This study developed a cost-effective framework to monitor cropland disturbance and analyze its driving forces in the Heilongjiang River Basin (HRB), a transboundary basin spanning China, Mongolia, and Russia. The framework integrated Landsat time-series imagery, machine learning, change-detection algorithms, and decision tree models. We processed 36,027 Landsat TM/ETM+/OLI images and used the random forest classifiers trained with annual cropland samples and spectral-temporal features to estimate annual cropland probability (1990–2021). The LandTrendr algorithm segmented the cropland probability time series to determine the spatiotemporal distribution and intensity of cropland disturbance. A decision tree model quantified contributions of different disturbance drivers. The framework enables a more comprehensive assessment of cropland stability through the integration of disturbance detection and driver attribution. Our results reveal that 128.9 × 10 4 ha of cropland were disturbed between 1990 and 2021, of which 762,900 ha (59%) were classified as severely disturbed, indicating sustained pressure on regional agroecosystems. Agricultural activities and cropland fallow accounted for for 47% and 22%of total disturbances, reflecting dominant human influences across the transboundary region. On the Chinese side (Heilongjiang and Jilin), disturbances were primarily driven by urban expansion and agricultural activities, whereas cropland fallow and abandonment dominated in the Russian side (Zabaykal'ye). These spatial divergences underscore that cropland disturbances emerge from the interplay between natural environmental conditions and region-specific policy contexts. This study clarifies the mechanisms driving cropland change in transboundary basin regions and provides a scientific basis for region-specific land management and coordinated ecological conservation to support sustainable agricultural development.
Key Findings
1
A cost-effective framework integrated Landsat time series, random forests, LandTrendr, and decision trees to detect cropland disturbances and attribute their drivers from 1990–2021.
2
A total of 128.9 × 10⁴ hectares of cropland were disturbed between 1990 and 2021; 762,900 hectares, or 59%, were severely disturbed.
3
Agricultural activities and cropland fallow accounted for 47% and 22% of total disturbances, respectively, demonstrating dominant human influence on cropland change.
4
Cropland disturbances reflected interactions between natural environmental conditions and region-specific policy contexts across the transboundary basin.
5
Disturbance drivers diverged spatially: urban expansion and agricultural activities dominated in Heilongjiang and Jilin, while fallow and abandonment dominated in Russia’s Zabaykal’ye.
6
The study processed 36,027 Landsat images and estimated annual cropland probability across the Heilongjiang River Basin using spectral-temporal features and annual cropland samples.
Research Object
Cropland in the transboundary Heilongjiang River Basin (China, Mongolia, and Russia) during 1990–2021
Research Subject
The spatial distribution, intensity, and driving mechanisms of cropland disturbance, including the contributions of agricultural activities, fallowing, abandonment, and urban expansion
Publication Details
Publication Date
2026-06-18
Journal
Publisher
ISSN
Open access PDF
Access Type
Author Information
Download PDF
Subscribe to digest