Soil temperature (ST) anomaly memory describes the persistence of ST anomalies and their influence on subsequent ST and atmospheric anomalies.Due to the limited availability of long-term ST observational data,reanalysis datasets are widely used in numerical simulations and statistical diagnostic analyses.However,the extent to which reanalysis data affect the persistence of ST anomalies remains uncertain,directly impacting the reliability of conclusions drawn from numerical and statistical studies on the influence of antecedent ST anomalies on atmospheric processes.Therefore,evaluating the memory of ST anomalies in reanalysis datasets is essential.This study evaluates the memory of ST anomalies at a monthly timescale using ERA-Interim,ERA5,and GLDAS reanalysis datasets from 1979 to 2019,alongside observational data from China.The ST anomaly memory is quantified using the autocorrelation method while considering the vertical propagation of ST anomaly signals across different soil layers.Results indicate that all three reanalysis datasets effectively reproduce the spatial distribution of multi-year average shallow ST.However,notable differences exist in ST memory.High-memory regions in ERA5 and ERA-Interim are primarily located in areas with an average annual precipitation of 400 to 800 mm,where ST anomalies persist for approximately 8—10 months.In contrast,GLDAS exhibits a distinctly different spatial pattern,with ST memory values significantly higher in the western part of China than in the eastern part of China.The intermonthly variations of ST memory show strong spatial consistency across different months in all datasets.Additionally,ST anomalies tend to propagate into deeper soil layers over time.ERA-Interim and ERA5 indicate longer ST anomaly persistence in the first soil layer in Shanxi,Shaanxi,and Henan Provinces,while in GLDAS,ST anomalies persist longer in western China.Compared with observational data,GLDAS and ERA5 better represent the persistence characteristics of ST anomalies in the first soil layer but fail to accurately capture the persistence characteristics of ST anomalies across the entire soil column.The ability of reanalysis datasets to reproduce ST anomaly persistence varies significantly by region and season.These findings highlight differences in ST anomaly persistence among reanalysis datasets,contributing to a better understanding of the climatic effects of antecedent ST anomalies.Moreover,discrepancies in soil stratification among different reanalysis datasets complicate direct comparisons of ST memory.Therefore,when using reanalysis datasets to study the influence of antecedent ST anomalies on subsequent climate variability,it is crucial to evaluate ST anomaly persistence to ensure the reliability of research conclusions.This study enhances the understanding of uncertainties in land surface-atmosphere interactions and the climatic impacts of antecedent land surface anomalies.