Modeling of Seismological Temperature Forecast at Depths of the Upper Crust

The accuracy of deep temperature forecast based on seismic wave velocity data and model thermograms was investigated depending on the distance to the forecast point. For this purpose, the results of seismic tomography along the sublatitudinal profile of the Northern Tien Shan subsurface area were used, as well as the temperature model previously constructed for this profile to a depth of 27 km. An assessment of the temperature forecast accuracy using artificial neural networks showed that at distances to the forecast point of upto 16 km, the discrepancies between the forecast and model temperature values based on longitudinal and transverse wave velocities, as well as their combination, are 7.4, 5.7 and 4.6%, respectively. With a 4-fold increase in the distance to the forecast point, they increase by 2-3 times. In general, it can be concluded that neural network temperature forecast of the earth’sinterior based on seismic wave velocity data can be performed with acceptable accuracy at large distances from the initial data measurement points and serve as a “seismological geothermometer”.

temperature, seismic wave velocity, well, forecast, artificial neural network, geothermometer