Undergraduate summer interns get involved with PNSN research
As summer winds down, we’d like to highlight some of the contributions of our undergraduate summer interns. The interns came from many parts of the country via different internship programs. The variety of contributions they made give a peek into the many different corners of hazard-related science that PNSN and affiliated UW, UO and USGS researchers are involved in.
Natalie Culhane, Western Washington University
I’m a rising senior at Western Washington University, and I spent the past couple of months in a Pathways internship with the USGS’s Earthquake Science Center. Under research civil engineer Alex Grant, I built a model that maps the runout length of landslides caused by seismic events. Previous scientists have spent countless hours digitally mapping slides that occurred just after large earthquakes, and the USGS has (publicly available!) landslide inventories for 44 of these global events, including ones from Puerto Rico, Hawaii, and Nepal. However, none of these inventories include information about the landslides’ runout lengths, and many have thousands of mapped slides. My summer task was to find a way to automatically trace these lengths using coding and mapping software. My GIS model accomplishes this, with an accuracy of 96%. Eventually, my team’s goal is to use this new data to find relationships between runout lengths and hill slope, slide volume, etc. If we can find these relationships, we may be able to make rough predictions on how far a slope will spill out after an earthquake, which would be great help in disaster preparedness for earthquake- and landslide-prone areas.
Automatically mapped runout lengths of landslides caused by the 2007 M6 earthquake in Aysén Fjord, Chile.
Francesca Skene, University of Washington
I am an undergraduate at UW dual majoring in Physics and Earth Sciences and have been an undergraduate research assistant working with Prof. Marine Denolle and PNSN researchers. My project uses seismic data to find the flow direction, velocity, and location of surface events, e.g. avalanches, rock slides, glacier quakes etc., on the Cascade Volcanoes. The data that I use is 1548 distinct events picked and labeled as surface events by the Pacific Northwest Seismic Network. I organize all nearby signals for each event, look at the differences in frequency content to estimate velocity and direction of the flow, and use station locations and pick times to locate the events with a grid search approach. I’ve also been determining the count of events per month from 2002 to 2021 at each volcano, taking into account the bias that the number of active stations has on this statistic. The results may contribute towards warning systems in the future and help to identify unstable slopes on the mountain.
Alaura Custard, University of Kansas
This summer, I worked with Ian Stone and Erin Wirth of the USGS in Seattle, WA. I was looking at the Tualatin Basin and how seismic waves interact with the structures of the basin. The Tualatin Basin, just outside of Portland, OR, is a sedimentary basin that is over 5 km deep. The sediment layers in the basin can cause amplified ground shaking in the event of a nearby earthquake. In June 2021, the USGS deployed seven seismometers across the region, with stations inside and outside the basin. Using the data collected from these stations, I was able to determine the frequencies at which seismic waves will have the highest amplitude in the basin. Once I found these frequencies, I looked at large global earthquakes and the waveforms recorded by the stations deployed. Using the amplitude of S-waves inside and outside the basin, I was able to find levels of amplification at each station. The results of this study imply there are specific frequencies that produce amplified shaking within the basin. These frequencies should be considered during earthquake hazard analysis, as buildings with the same natural frequencies will experience greater shaking in the event of an earthquake.
Tualatin basin depth map showing location of seismic stations (triangles) and calculated resonant frequencies (numbers) and their amplification factors (size of circles).
Nicholas Smoczyk, University of Minnesota Duluth
I am a junior Environmental Science major at University of Minnesota Duluth, URG2 project summer intern. My project groups seismic events based on waveform similarity to identify repeating seismic events from five Pacific Northwest volcanoes: Mt. Baker, Mt. Hood, Newberry Volcano, Mt. Rainier, and Mt. St. Helens. Identifying repeating events and their information, like location, can show how each volcano has changed over time, as well as how each volcano responds to environmental factors like rainfall.
I do this by using the REDPy (Repeating Earthquake Detector in Python) catalogs for these volcanoes, which have already identified many repeating events. Then I use EQcorrscan, a template matching python package, to enrich the catalog with new detections and to backfill to earlier dates before the REDPy catalog begins.
Repeating events from REDPy at station CC.HOOD (Mt. Hood), each waveform happens at a different time but has a high enough similarity to be grouped into the same cluster.
Amanda Syamsul, University of Washington
I’m an undergraduate senior at the University of Washington majoring in Geophysics. Over the past two years, I’ve been studying global surface load induced earthquakes with Professor Brad Lipovsky.
Loads on the surface of the Earth modify earthquake occurrence, however, previous studies of this phenomenon have been limited in geographical and temporal extent. Here, we systematically searched global datasets for a relationship between surface mass loading and earthquake occurrence. Surface loading can be described as mass change in the Earth’s water, ice sheets, and solid earth, which can occur during events such as landslides, glacial isostatic adjustment, or monsoon season. Using a non-parametric, Bayesian analysis we confirmed that surface loading during periods with an earthquake differs from the background in a statistically significant way. This result is independent of whether we filter the earthquake catalog to remove earthquakes that might have been triggered by prior earthquakes. We found that the highest relative probability of earthquake occurrences were at locations undergoing extreme surface mass loading and unloading. To better understand this, we analyzed the focal mechanisms (descriptions of slip sometimes called “beachballs”) of events we categorize as Surface Load Induced Earthquakes (SLIQs). We examined the vertical component of the slip direction vector, the one most sensitive to changes in mass loads, and concluded that there is moderately strong evidence that the symmetry of the surface load response is correlated with faulting style. The existence of surface load induced events suggest novel interactions between cascading hazards, such as earthquakes and landslides.
Earthquakes from April 2002 to January 2022 with magnitude > 5.4 analyzed in this study. The small gray dots represent all the earthquakes in our initial catalog of 12573 events. The colored circles show those events likely induced by significant changes in mass loading (relative conditional probability > 1.74). Circle size denotes magnitude.