THE USE OF LANDSAT SPACE IMAGES AND WATER CONTENT DATA IN NEURAL NETWORK FORECASTING OF PROMISING DRILLING SITES IN THE YURUBCHEN-TOKHOMO PETROLEUM ACCUMULATION ZONE

V. V. Dostovalov

The paper considers integrated multifactor neural network analysis of 2D and 3D data with a modified training algorithm applied in the «GeolEdit» program complex (a case study of the Yurubchen-Tokhomo petroleum accumulation zone). The author analyze the possibility to use additional criteria in case of extending the territory under study toward north-east (Kuyumbinsky and Tersko-Kamovsky sites added). The author also considers the application of neotectonic data derived from Landsat space images and data on the total water content of productive intervals in neural network forecasting of promising deep drilling sites. A segment of the neural network forecast of deep well location (Kuyumbinsky site) is given. The data on the inflow of deep drilling wells including those with the horizontal borehole end are analyzed. The author proves the efficiency of the neural network forecasting under conditions of heterogeneous carbonate reservoirs for sites extensively covered by drilling.

Key words: neural network, hydrocarbon localization, Yurubchen-Tokhomo zone, Kuyumbinsky site, Landsat space images.

Download PDF