Abstract
Accurate estimation of soil thermal properties is essential for various environmental and engineering applications. However, these parameters are often not directly accessible through measurements due to the complex and heterogeneous nature of soil. This study presents an inverse modeling framework to estimate these coefficients using time-resolved boundary temperature data. Synthetic data generated from the solution of the forward heat transfer problem with known parameters were used to assess the method under controlled conditions. The inverse problem was formulated as a minimization task and solved using a gradient-based optimization approach, with gradients computed via the adjoint problem. The Thomas algorithm was employed for stable discretization. Numerical experiments demonstrated reliable convergence to the exact values under various initial assumptions, highlighting the robustness and accuracy of the method. These findings underscore the method’s potential in geotechnical engineering, soil physics, and environmental heat transfer modeling.