Stanford Geothermal WorkshopFebruary 9-11, 2026

Microearthquakes (MEQs), also known as microseismic events, are weak seismicity with magnitudes below 3.0 that do not present any risk to nearby communities or infrastructures. Their occurrence is induced and thus inherent to geothermal field operations, playing an important role in monitoring changes in reservoir stress fields. Events typically occur when conditions in pre-existing faults or fracture networks change due to fluid extraction and brine injection, causing stress buildup that eventually reaches a threshold causing rock failure or slippage, releasing stored strain energy as MEQs. Mapping and analyzing the spatial distribution of these events can contribute to the development of injection management strategies especially when integrated with injection tracer test data. The same is also used to confirm the viability of an area as a drilling target. Spatial clustering of MEQs is employed to uncover patterns and trends, ultimately deriving valuable insights related to reservoir structures and dynamics. Currently, conventional methods like grid-based clustering calculate event density within fixed grids and has been proven to be effective in identifying anomalies or dense clusters. However, this approach struggles to distinguish nested clusters or account for the depth component of the events during clustering. To address these limitations, unsupervised machine learning algorithms have gained prominence in identifying significant patterns of arbitrary shapes within large datasets. In this study, the OPTICS (Ordering Points To Identify the Clustering Structure) algorithm, a density-based clustering method, is applied to a 3D spatial dataset of MEQ events from the Tiwi Geothermal Field between 2012 and 2017. This research aims to examine the impact of input parameters on clustering solutions and identify regions of interest for detailed analysis.

Topic: Geophysics