Abstract: Using five soil moisture data sets of, including the station observation during 1992-2013 over Huaihe River Basin(HRB), the spatio-temporal characteristics of soil moisture variation and its memory are analyzed, meanwhile, the remote-sensed and reanalysis soil moisture dataset are also evaluated against the observation. Based on the station observation, it is found that the observed soil moisture is relatively dry in the northern HRB, and wet in the southern HRB, with stronger inter-annual variability in western HRB and weaker variability in eastern part of the basin. Generally, the latest reanalysis soil moisture, from European Centre for Medium-Range Weather Forecasts (ERA5-Land), agrees well with the station observation in both spatial distribution and temporal variation, and the community land surface model (CLM4) can well reproduce the observed inter-annual variation of soil moisture when forced with observed meteorological variables. The longer memory of soil moisture can be found with increased time scale in surface soil layer, but this enhancement of memory with time scale becomes weak in deeper soil layer, like 40-50 cm. The soil moisture memory also shows strong seasonal variation, with longer memory in late autumn and early winter, and weak memory in summer, and the observed seasonal cycle of soil moisture memory can be well reproduced by both the standalone CLM4 simulation and the IAP atmospheric general circulation model coupled with CLM4. The spatial difference of soil moisture memory can be found for ERA5-Land, CLM4 and IAP-AGCM4.1, with longer memory in the northwest part of HRB and weaker memory in the southern HRB, and the regional contrast of soil moisture memory is the most significant in autumn.