| Abstract |
Developing an understanding of a geothermal system with the use of more than one geophysical dataset can help de-risk drilling prospects and help with reservoir management. This research aims to constrain the structure of the Montserrat Geothermal system (MGS) located in the western part of Montserrat, an island in the Eastern Caribbean. This study utilizes existing data obtained from three separate geophysical surveys: a magnetotellurics survey, a high-resolution seismic tomography survey and a gravity survey, to jointly interpret the structural features within the MGS. Inverse modelling of the geophysical datasets produces three dimensional subsurface models of resistivity, seismic velocity and density. These physical property estimates can then be used to help determine the spatial distribution of subsurface materials in the reservoir such as clay, unaltered rock and fluid saturated reservoir rocks. To jointly interpret the datasets, a machine learning application, the Fuzzy c- means clustering algorithm (FCM), was used in this study. The FCM method uses a distance measure and a membership function to cluster the multiparameter geophysical dataset (each point in the dataset is associated with a density contrasts, resistivity and velocity perturbation value). Utilizing two validity indices, four clusters were determined as the optimal number of clusters within the multiparameter geophysical dataset. Each cluster was interpreted based on the variations in each geophysical property. Cluster 1, 2 and 3 were interpreted as the unaltered region, clay cap and geothermal reservoir of the MGS respectively. While further investigation is needed, cluster 4 was interpreted as an intrusion possibly associated with the extinct volcanic center, Centre Hills. The advantage of the cluster analysis is that each structure is now assigned more than one geophysical parameter that can be used to refine previous calculations on petrophysical parameters, such as saturation, permeability and porosity. |