This study focus on development of the estimation models for yearly bare soil temperature (Ts) according to soil depth (z) and examined the difference between them [dT(z1-z2)=Ts(z1)-Ts(z2)] using data collected from 23th May 2023 to 23th May 2024 in Sandy Loam. Principle results are as follows: (1) The deeper of the soil depth, the higher of the annual maximum, average and mininum soil temperature, where as the smaller of the annual range. (2) The deeper of the soil depth, the smaller of standard deviation of the annual soil temperature. (3) The differential of soil temperature to soil depth (dT/dz) for annual maximum, average and minimum soil temperature was 0.015°C, 0.049°C and 0.110°C/cm, respectively. (4) The cation process of the models were based on the error analysis criteria with the view point of accuracy, precision and association. (5) The estimation models for Ts at the depth according to Julian date were developed as a sine functions (MTs10: model for 10 cm, MTs20: model for 20 cm, MTs30: model for 30 cm) all judged 『Fit well』. (6) The estimation models for dT between the Ts of the soil depth according to Julian date were developed as a sine functions [(MdTs(10-20): model for 10-20 cm, MdTs(20-30): model for 20-30 cm, MdTs(10-30): model for 10-30 cm)] all judged 『Fit』. In summary, this study successfully developed estimation models for Ts and dT, employing sine function regression approaches. The models were rigorously evaluated for the change of soil temperature using error analysis criteria, providing valuable insights into annual soil temperature dynamics according to soil depth in Sandy Loam.