This paper describes the construction of a 0.25 & DEG; x 0.25 & DEG; daily soil temperature (TS) data set at six soil depths (i.e., 0, 5, 10, 15, 20, and 40 cm) based on a homogenized TS at more than 2,000 meteorological stations across mainland of China for 1960-2017. First, at each station, the observed TS is split into a daily climatology of 1981-2010 and a daily anomaly for 1960-2017. Then, they are respectively interpolated to 0.25 & DEG; x 0.25 & DEG; horizontal resolution by using the thin-plain spline and the angular distance weight interpolation methods. Finally, a gridded multilayer TS data set is constructed from the sum of the above two gridded products. Latitude, longitude, and high-resolution digital elevation models are explicitly incorporated into the interpolation processes. Selectivity tests indicate that the optimal interpolation distance for the TS anomaly field is 500 km, where the gridded data set may have maximum coverage of the entire Chinese mainland. The results show that the gridded TS climatology is slightly lower than the station observations. The differences between gridded and the station-based TS varies & PLUSMN;3 & DEG;C, with relatively large in SW and TIBET. For the gridded anomaly fields, the cross-validation shows that the mean absolute errors between gridded and station-based TS are generally less than 0.6 & DEG;C in the YZ, SW, and SC, but they are above 1.0 & DEG;C in the NW and TIBET due to the sparse station distribution and large topographic relief. From 1961 to 2016, the seasonal variability of the gridded TS is generally consistent with that of the surface air temperature. The gridded TS over the majority of land displays significantly positive trends in both summer and winter (p > 95%), indicating an overall warming land surface in China. The long-term gridded TS data not only preserves the spatial features represented in the station observations, but also provides a spatio-temporal continuous long-term product of mainland China. Thus, it may be used in future studies, for example, evaluation of land surface model simulations, validation of satellite retrievals, land-atmosphere interactions, and climate change.
