Engineering projects, such as deep tunnel and nuclear waste repository, may involve very deep overburden. As the depth at the project site increases, the underground temperature tends to elevate. This study aims to develop a reasonable methodology for estimating underground rock temperature under deep overburden. This paper proposes a methodology of modular analysis for estimating deep underground rock temperature. The proposed methodology involves the finite-element method and nonlinear optimization method. The proposed analysis assumes that the distribution of underground temperature is solely due to two-dimension thermal conduction. After verification, numerical experiments demonstrate that the distribution of underground temperature significantly depends on the thermal conductivity of underground rock. Parametric study illustrates that the anisotropy of thermal conductivity and the topography of ground surface also affect the distributions of temperature and thermal gradient. This study also discusses the sources of error for estimating the underground temperature.