Abstract:
Geostatistical methods and machine learning models offers predictive insights in the spatial domain which reduce the risk and uncertainty across the scope of exploration and development projects. To accommodate the needs of addressing the most complicated subsurface reservoir systems be they carbonate, clastic or other lithology, this tool needs to be optimized. Knowledge of the geophysical properties of the subsurface as a function of various petrophysical parameters like radioactivity (GR), Density (RHOB), Resistivity (RESD), Velocity (Vp, Vs), etc. is an important consideration in Geoscience exploration. Over a spatial area, one can measure these properties using the limited control points by relying on machine learning techniques to employ them and predict over unknown/unobserved, Several ML techniques are currently used such purposes. The geoscience exploration industry is being forced to re-evaluate techniques for carrying out of new exploration programs, as a large number of easy mineral resources targets been already found. In order to locate future targets, a holistic multidisciplinary approach is required to identify trends across multiple characteristic features, including geology, geophysics, and geochemistry. PRIME is an Integrated platform for solving Petrophysical and Rock Physical challenges for today’s exploration and development activities. It currently employs quantitative methods to predict fluid, rock, and pressure properties. In this thesis, I developed the Geostatistical tool by enhancing its predictability potential in the subsurface and compared the result with the true value with its variance explained metrics. The developed method has scope for further improvement
