Subsurface temperature prediction in geothermal field with neural network using 3d mt data inversion and borehole temperature data

This research is looking for a well-temperature data link (borehole) with resistivity results of 3D inversion MT data to get the best interpretation results. The temperature data of borehole that has been tied with resistivity values will be used in temperature spread prediction by using Neural Network (NN) with Matlab software. The vector used is the coordinates of the position (x, y, and z), the resistivity gradient, and the resistivity values with the temperature target in that position. This is used because the relationship between TOR (Top of Reservoir) is very related to BOC (Best of Conductor). Weights on each network obtained from the NN training will be used to estimate the temperature at the next well drilling (temperature vs depth simulation). At the training stage, it must be ensured that the best data on the MT measurement is closest to the well so that the weight value can represent the geothermal field. The weight of the Neural Network will be used to predict the temperature in 3D so that we can give a temperature spread on a geothermal field. 3D temperature modeling results from Neural Network will be used as one of the considerations in further drilling because it can help determine the up-flow area, which is the target area (sweet zone) and outflow along with other geophysical data, thereby reducing the risk in developing geothermal fields.