Loading...
Machine learning web application to automate the interpretation of the origins and post-generation processes of natural gases, A
McBride, Ian C.
McBride, Ian C.
Citations
Altmetric:
Advisor
Editor
Date
Date Issued
2023
Date Submitted
Keywords
Collections
Research Projects
Organizational Units
Journal Issue
Embargo Expires
Abstract
The origins and post-generation effects of natural gas, mixtures of both hydrocarbons (methane or C1, ethane or C2, propane or C3, and butanes or i-C4/n-C4) and non-hydrocarbons (CO2, N2, and He), have traditionally been interpreted using comparisons of molecular and isotopic combinations in empirically defined gas fields of binary genetic diagrams. These genetic diagrams were supplanted by larger datasets that redefined the empirical boundaries, and spurred the development of a web-based machine learning tool to automate the process based on predetermined manually assigned labels of interpretations in a global geochemical dataset. This research focuses on the further development of this web-based application to look at the incorporation of more input features and their effect on prediction accuracy. Geologic habitat, and the carbon isotope of ethane were added to a Random Forest (RF) prediction model, improving accuracy and flexibility for the user. A layered random forest model based on the confidence intervals of origin utilizes the data further to predict the source rock type for thermogenic gases with a level of confidence. The web-based tool was designed with a dynamic format, making it easily accessible for a variety of screen sizes on most modern electronic devices.
Associated Publications
Rights
Copyright of the original work is retained by the author.