Abstract
Temperature data in the Yanaidzu-Nishiyama geothermal field have been analyzed statistically in order to reveal general trends of temperature profiles and to estimate deep level temperatures from shallow level temperature profiles. Applied regression functions are (1)T=T0+b·√d+c·d(3-coefficient power function), (2)T=T0+b·√d-d0+c·(d-d0)(4-coefficient power function), (3) T = T∞ [1- exp { - b (d-d0) }] (exponential function), and (4) T = T0 + b·1n (d-d0) (logarithmic function), where d is depth, T is temperature, and b, c, d0 and T0 (and T∞) are constants. Equation (2) is the best for regression of the total data of each well, but the worst for estimation of deep level temperatures. In contrast, Equation (4) is the best for the estimation, though the advantage is small for Equations (1) and (3). This fact suggests that these three functions are applicable for reducing a general trend when geostatistics is applied for underground temperatures. It is possible to estimate temperatures ahead 500 m and 1000 m within errors of 50°C and 100°C, respectively, when Equation (4) is applied for temperature data more than 500 m in depth.
