PCA and NMFk to constrain the geologic controls of hydrothermal processes
Abstract
Faults, fractures, and other geologic structures affect subsurface permeability and are commonly critical controls of hydrothermal fluid flow. Items of particular interest are (1) identification of the specific aspects of the geologic architecture that host production-grade fluid flow and (2) developing methods to distinguish these locations from the majority of the rock volume that does not transmit hydrothermal fluids. Such information can contribute to efficient exploration and development and facilitate sustainable resource management. In two published studies, we analyzed the 3D geologic structure of the Brady geothermal field in western Nevada. We used two unsupervised machine learning (ML) techniques: principal component analysis (PCA) and non-negative matrix factorization with k-means clustering (NMFk) to identify geologic attributes that are most closely related to production wells. Both ML techniques can distinguish productive from non-productive wells based on the geological data. Both techniques also show that structurally significant faults, and the dense area of faulting in the fault step-over are important controls on geothermal production. We compare the methodologies and results of these two studies and discuss the implications for understanding fault and fracture controlled hydrothermal systems in general.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2021
- Bibcode:
- 2021AGUFMIN42B..04S