انجمن ملی ژئوفیزیک ایران مجله ژئوفیزیک ایران 2008-0336 20 3 2026 07 23 A data-driven approach to reservoir characterization: machine learning and seismic Attribute integration in the penobscot field, Nova Scotia basin A data-driven approach to reservoir characterization: machine learning and seismic Attribute integration in the penobscot field, Nova Scotia basin 153 165 225947 10.30499/ijg.2025.524352.1697 FA Khlieeq Ul Zaman M.Sc Student, Department of Earth Sciences, University of Sargodha, Pakistan Rani Ummay Farwa M.Sc Student, Department of Earth Sciences, University of Sargodha, Pakistan Syed Haroon Ali Assistant Professor, Department of Earth Sciences, University of Sargodha, Pakistan 0000-0002-8619-7005 Amjad Ali Assistant Professor, College of Mechanical and Energy Engineering, Beijing University of Technology, Beijing, China Fahad Ali Assistant Professor, Department of Geology, Bacha Khan University, Charsadda, Khyber Pakhtunkhwa, Pakistan Muhammad Iqbal Hajana Assistant Professor, Department of Earth and Environmental Sciences, Bahria University, Islamabad, Pakistan Journal Article 2025 05 18 <span>Petrophysical analysis and advanced attributes of machine learning are used to evaluate exploratory wells, B-41 and L-30 of Penobscot Basin, Nova Scotia, Canada. The main objective of this paper is to evaluate the petroleum system and the prospects and leads. Well B-41 and L-30 reached their TD at 3483m and 4360m respectively, both wells were declared dry and plugged abandoned. The petrophysical studies include Bulk density and neutron porosity cross plots, in both wells, neutron and porosity cross plots show the almost linear trend of values, showing clay and sand lithology. Porosity values of B-41 and L-30 are 8-10% and 10-12%, respectively. Shale volume of B-41 is 37-44% and for L-30 is 23-32%; however, both wells show a fair porosity, but water saturation is high, so it is not a favorable condition for hydrocarbons to accumulate. For using attributes of machine learning 11 sets of 2D seismic lines and 1 set of 3D seismic surveys were used an advance technique of machine learning known as SOM (Self Organizing Map) is used, which is computational data analysis technique which enables mapping of nonlinear data to lower dimension and also at different frequencies, for this analysis the frequencies of 11, 18, 26 Hz are used. Machine learning enables efficient and accurate predictions even with limited data, providing a more practical and streamlined alternative to conventional reservoir simulation techniques.</span> <span>Petrophysical analysis and advanced attributes of machine learning are used to evaluate exploratory wells, B-41 and L-30 of Penobscot Basin, Nova Scotia, Canada. The main objective of this paper is to evaluate the petroleum system and the prospects and leads. Well B-41 and L-30 reached their TD at 3483m and 4360m respectively, both wells were declared dry and plugged abandoned. The petrophysical studies include Bulk density and neutron porosity cross plots, in both wells, neutron and porosity cross plots show the almost linear trend of values, showing clay and sand lithology. Porosity values of B-41 and L-30 are 8-10% and 10-12%, respectively. Shale volume of B-41 is 37-44% and for L-30 is 23-32%; however, both wells show a fair porosity, but water saturation is high, so it is not a favorable condition for hydrocarbons to accumulate. For using attributes of machine learning 11 sets of 2D seismic lines and 1 set of 3D seismic surveys were used an advance technique of machine learning known as SOM (Self Organizing Map) is used, which is computational data analysis technique which enables mapping of nonlinear data to lower dimension and also at different frequencies, for this analysis the frequencies of 11, 18, 26 Hz are used. Machine learning enables efficient and accurate predictions even with limited data, providing a more practical and streamlined alternative to conventional reservoir simulation techniques.</span> https://www.ijgeophysics.ir/article_225947_539caed5eb8be8890e921d62413a0d71.pdf