3D-Temperature Estimation Using Geostatistical Methods

Abstract

The temperature field down to a depth of 5000 m below sea level has been estimated for Germany and Upper Austria in a consistent manner with universal kriging in 3D space. Conventional temperature mapping interpolates the temperature between sample locations at certain depths. This way, no consistency between different depth levels is ensured. Information from shallower levels gets lost. Kriging provides the best fit to measured temperatures. Another major advantage of kriging over conventional mapping algorithms is the prediction of the uncertainty associated with the temperature estimate. Any temperature estimate would lose much of its meaning without considering the uncertainty. 3D kriging resolves most of the limitations of conventional mapping techniques, however, like any statistical estimation it does not take into consideration varying thermal conductivities or the thermal convection of fluids. The 3D subsurface temperature model is used as an example to demonstrate what has to be considered when applying the 3D kriging method. Details are also given on current research that focuses on handling various data qualities in the kriging system.