AUTHOR=Wang Shuo , Tuerhong Aihemaitijiang , Maimaitituersun Nueraili , Ning Zuo-Jun 

TITLE=Variations in maximum freezing depth in Northeast China from 1975 to 2024 using a machine learning model

JOURNAL=Frontiers in Soil Science

VOLUME=Volume 5 - 2025

YEAR=2025

URL=https://www.frontiersin.org/journals/soil-science/articles/10.3389/fsoil.2025.1642004

DOI=10.3389/fsoil.2025.1642004

ISSN=2673-8619

ABSTRACT=A freezing depth prediction model was constructed using machine learning, incorporating comprehensive data from ground meteorological monitoring stations and remote sensing reanalysis data. The maximum freezing depth (MFD) of seasonally frozen ground (SFG) in Northeast China was systematically analyzed from 1975 to 2024. The simulation results from the machine learning model (MLM) indicated that the MFD of SFG in Northeast China displayed a decreasing trend over the past 50 years, with an average rate of change of -8.54 cm per decade. The average maximum freezing depths (AMFDs) in Northeast China for each decade were: 136.71 cm (1975−1984), 131.96 cm (1985−1994), 123.07 cm (1995−2004), 110.82 cm (2005−2014), and 104.58 cm (2015−2024). The area occupied by each AMFD interval in Northeast China over the past 50 years increased in regions with freezing depths <160 cm. The area with freezing depths >160 cm displayed a decreasing trend. The results not only reveal the impact of climate change on freezing depths, but also provide a scientific basis for environmental management and ecological protection in frozen ground areas. Changes in freezing depth directly affect many sectors such as agriculture, construction, and transportation, making accurate prediction essential for developing climate adaptation strategies. Considering the lack of data regarding the MFD of SFG in Northeast China for the past 50 years, the MLM provided an effective method for predicting changes in MFD using meteorological data and remote sensing reanalysis data.