The susceptibility assessment dataset for freeze-thaw disasters in Northeast China is based on on-site investigation data of freeze-thaw disasters in Northeast China. Five machine learning models (GAM, GAM, GBM, RF, ANN) and their weighted integration methods are used to construct a 1000 meter resolution freeze-thaw disaster susceptibility assessment product for Northeast China. The data expresses the spatial susceptibility level of regional freeze-thaw disasters in grid form, which can be used in fields such as permafrost disaster monitoring, risk assessment, ecological environment research, and regional planning.
The levels of freeze-thaw disasters are divided into the following 5 categories:
1: Extremely low
2: Low
3: Medium
4: High
5: Extremely high
Based on on-site investigations and geographical data of spatial environmental factors.
Sample construction: The positive sample (disaster occurrence point) comes from on-site investigations in Northeast China, recording the location, type, and environmental background of freeze-thaw disasters. Negative samples (non disaster points) are randomly sampled at a certain proportion in areas where disasters have not occurred, ensuring uniform spatial coverage and avoiding model bias towards areas with dense disaster points.
Classification features: Extract features such as terrain, climate, soil and surface, hydrology, etc. based on environmental geographic data.
Algorithm execution: Train five types of models separately using Python, evaluate model performance using k-fold cross validation (k=5 or 10), and generate primary products by weighted averaging of the predicted results of the five models.
Post processing: Divide the continuous probability values output by the integrated model into five levels, including the natural breakpoint method (Jenks), convert the predicted results into a 1000 m resolution grid, project them into WGS84 (EPSG: 4326), and output them as GeoTIFF
The 1000 meter resolution freeze-thaw disaster prone product developed for machine learning integration in Northeast China has good data quality and can be used for various spatial analysis tasks to identify high-risk areas and spatial clusters of freeze-thaw disasters. It is overlaid with terrain, climate, land use and other data to analyze driving factors.
| # | number | name | type |
| 1 | 2022FY100700 | Survey of Permafrost Conditions and Freeze-Thaw Damage in the High-Latitude Regions of Northeast China | Basic Resource Survey Project |
This work is licensed under
CC BY 4.0 (Creative Commons Attribution 4.0 International License).
CSTR:11738.11.NCDC.NIEER.DB7284.2026
10.12072/ncdc.nieer.db7284.2026
