<p>This study proposes an indirect method to estimate the ground effective thermal conductivity in the formation of a depth over several tens to one hundred meters by averaging probability-weights of all soil/rock types. Thermal response test results are used as the in-situ measurements for determining a set of the individual effective thermal conductivities in the least square. The probability values of all soil/rock types are estimated at any location and depth through the indicator kriging interpolation with the adjacent water well data. In this study, 76 test data were used for analysis and the individual effective thermal conductivities of 8 soil/rock types, i.e., clay, sand, gravel, volcanic ash, Quaternary volcanic rock, Neogene fine rock, Neogene coarse rock, and bedrock were determined. For comparison, this study demonstrated a conventional method to estimate the ground effective thermal conductivity as a thickness-weighted average. The conventional method requires the direct geo-information such as geologic columns, which are not often obtained especially in the deep zones several meters deep. Also in this study, only 25 test data included the geologic columns. Thus, the conventional method resulted in the insufficient agreements of ground effective thermal conductivity between the estimates and the measurements. On the other hand, the analysis of the proposed method was performed in all 76 test data by using the probability estimates through the kriging system. As a result, all individual effective thermal conductivity values were reasonably obtained: the values of clay, sand and volcanic ash were almost equal to the reference values in texts. The value of gravel was relatively large probably because of the natural convection flows along the borehole during heating. The values of soft Neogene rocks were smaller than those of hard Quaternary volcanic rock and bedrock. The estimation errors of the ground effective thermal conductivity were close to the possible error by testing itself, except for the results over 3 W/(m･K). The K-fold cross validation (K= 10) also indicated the stability of the analysis results. This means the potential use of the method not only in the study area but also in other regions. This proposed method contributes to various heat transport problems in practice by providing the estimates of the ground effective thermal conductivity at any location. The method is also efficiently used to construct a database of the ground effective thermal conductivity, as shown in a case study of Sapporo City.</p>