Ongoing Work:

Thermal moonquakes have been correlated to the lunar day-night cycle and are attributed to stresses induced by solar illumination, but knowledge on their precise source mechanism has been limited due to a lack of systematic cataloging caused by poor data quality. 

We use frequency-based waveform processing algorithms and stochastic gradient descent to obtain moonquake parameters and locations to assess what surface processes may be responsible for these seismic phenomena. 

EQSpec: A Python package for automatic seismic detection within noisy datasets using spectrograms

Time-domain seismic detection methods struggle with noisy data. This Python package uses Convolutional Neural Networks, limited training data, and spectrograms to create seismic catalogs for messy datasets. This methodology has been demonstrated to work in a variety of environments including the arctic, ocean-bottom, and even the Moon!

Package is currently in closed Beta testing, please let me know if you'd like to be involved. 

Scientific return for planetary missions is fundamentally constrained by power requirements for continuous data telemetry. What if a lander could be trained to automatically figure out the difference between signal and noise and only send back the information we cared about?

We use a limited training set of earthquakes from a single Earth seismic station and use Convolutional Neural Networks to obtain the first high-quality seismic catalog of thermal moonquakes within the Apollo 17 Lunar Seismic Profiling Experiment!

Long-duration, tremor-like signals are often detected at producing geothermal fields and associated with subsurface properties. We're interested in detecting and locating these signals to understand the causal mechanisms and enhance field productivity. 

Past Projects:

Variations in the ambient seismic noise cross-correlation were used to calculate velocity changes for the Sierra Negra volcano using the Moving Window Cross-Spectral technique. We found a -0.27% decrease in velocity 17 days before the eruption, which we attribute to dilatation from a M 4.8 earthquake and degassing occurring after magma intrusion. 

Earth Observations (EO) have revolutionized Earth science in that data is available at unprecedented temporal and spatial scales. Manually sorting through the abundance of EO data to gather information on a specific phenomenon is often time-consuming and events may be missed. This project done as part of the NASA Frontier Development Lab explored the use of a featurizer and similarity search to find specific atmospheric phenomena of interest in unlabeled data. 

We used ambient seismic noise to compute Rayleigh and Love-wave dispersion maps for 5-25 seconds period for the Saudi Arabian shield using over 100 broadband seismic stations. The dispersion maps were inverted to shear-wave velocity using the Neighborhood Algorithm through the Dinver software package. 

Fluid injection for geothermal production has the potential to produce subsidence and microseismicity that can incur heavy financial cost or hazard. We used seismic ambient noise to obtain time-dependent measurements of shear velocity within the geothermal reservoirs of Rotokawa and Ngatamariki, two producing geothermal fields in the Taupo Volcanic Zone located in the central North Island of New Zealand. 

Harrat Rahat is a volcanic field located in west-central Saudi Arabia and is the site of the most recent eruption in the country. Tomography studies can be used to infer material properties such as partial melt, and are instrumental for volcanic hazard assessment. We use ambient seismic noise to compute Rayleigh and Love surface-wave maps and invert to a pseudo-3-D model of shear-wave velocity and determine the subsurface structure of the volcanic field. 

Finding subsurface faults at geothermal fields is a challenging and expensive process as reflection and refraction studies are often hindered by energy scattering and attenuation in the near-surface layers. We used ambient high-frequency noise and the Refraction Microtremor method to determine shallow subsurface structure and a fault location at the Ngatamariki geothermal field.