Research

Currently, I work with Dr. Rogier Windhorst’s cosmology research group at ASU. I have also worked on research in planetary science and astrobiology, both of which I am interested in building upon as future projects. My main area of interest for future research is the physics of impact cratering and impact seismology. This page includes descriptions, thought processes/work flows, and data/figures for various research projects I have worked on, both formal research and research adjacent class projects. If you are interested in any of this work and want more details or have questions, please email me at semmons3@asu.edu

Galaxies and Cosmology

Galaxy Inside Out Growth (2025-present)

This is my current project. Getting started took awhile due to both my limited knowledge of the methods used, and some technical issues. That being said, we are making progress now and I have learned a lot while working on this project.

For this project, I am using the Python package PySersic to fit a sample of several hundred galaxies from JWST images. This is being done for each galaxy twice - once using a multi-band fit of 8 NIRCam filters and again using a single-band fit of H-α maps from NIRISS. The resulting outputs, primarily the Sérsic index and effective radius, can then be compared between the two sets to determine where in the galaxies star formation is still happening and where it has stopped - specifically to study inside out growth.

Partway through the project; after fitting the galaxies there are a few that are extreme outliers - these seemed to be mostly, if not entirely, poor fits that may need to be improved. For example, some of the models appear to have a very faint ring extending several times the galaxy’s radius beyond it which is not present in the JWST image.

An example of those very faint rings (may need to look closely)

In an attempt to fix this, I reran the continuum fits with the effective radius prior set as a truncated Gaussian distribution with a mean of 10, standard deviation of 3, and bounds of 1 and 35; all in pixels; rather than leaving it as PySersic’s autogenerated prior. After rerunning the fits, they are generally much better, and the project has been able to continue on.

At the current point (9/25/2026), the preliminary results are very much unexpected. So I am attempting several things to figure out if those results are real or if they are an artifact. The main potential issue is that the resolution of our continuum imaging is higher than that of the Hα maps. I am injecting noise into the continuum images to bring both to the same signal to noise ratio, then I will rerun the continuum fits and see if the unexpected results are still present.

Planetary Science

PsycheESE (2025-2026)

This project started as the capstone for my astrophysics major. Despite that, it is the first (and currently only) research I have done that is published in some way, specifically as an iPoster at the American Astronomical Society 248th meeting in June 2026. A link to the poster can be found here and a GitHub repository with a guide to replicate our methodology, example files, and pdfs of our class presentations and report is here. The research consisted of seismolgy simulations that tested the idea of using artificial impacts to study asteroid interiors. Due to the nature of the capstone project, this was specifically tested with models of (16) Psyche.

The project lasted 2 semesters, the first was dedicated to conceptualization, study, and planning. For this, the class somewhat followed the process used for planning at NASA and similar organizations - with each group writing and presenting a System Requirements Review (SRR), Preliminary Design Review (PDR), and Critical Design Review (CDR), as well as maintaining subsystem Interface Control Documents (ICDs) throughout both semesters. Each group was required to have a Team Lead, a Science Lead, and a Lead for each subsystem who was responsible for ensuring that subsystem was complete and functional (though all group members were expected to contribute to all subsystems). These roles were decided on by the group. I was my group’s Science Lead - responsible for ensuring our project’s output actually answered our science question and was heavily involved with several of the subsystems where I helped to provide accurate input parameters for the Model and Simulation subsystems and reasonable tests for the simulation’s accuracy.

The second semester started with a Delta CDR, focusing on any changes that had occurred over winter break. The rest of the semester was dedicated to actually doing the project. After setting up our computer we iterated on our simulations, slowly ramping up the size and complexity as we found and fixed errors or other issues.

As for the research itself, we used the simulation software SeisSol to study the feasability of using artificial impacts and seismometers to study the interior of the asteroid (16) Psyche. We created three simplified models of the asteroid, a homogeneous rocky model, a two layer rock and metal model, and a “blobby” rock and metal model. Our simulated impactor was based on the Small Carry-On Impactor (SCI) from JAXA’s Hayabusa2 mission and we used the noise floor of the VBB from the Seismic Experiment for Interior Structure (SEIS) on NASA’s InSight mission on Mars.

The three models

For the other inputs, we approximated the impact’s seismic moment tensor as an explosion (see the Nishiyama et al., 2021 citation in the poster), which for us looks like:

\[M=\sqrt{\frac{2}{3}}\begin{bmatrix} M_0\quad 0 \quad 0 \\ 0\quad M_0 \quad 0 \\ 0\quad 0\quad M_0 \end{bmatrix}\] \[M_0=\left(\frac{E_I}{4.5\times10^{-9}}\right)^{\frac{1}{1.24}}\]

We also had to estimate the sheer modulus and Lamé’s First Parameter for the rock and metal materials in the models and used a binary search algorithm to find the time and normalization constant. There were some fun errors that were found and corrected as we worked, such as having the seismic efficiency being 8 orders of magntiude smaller than anything ever measured. It turns out that there were two issues causing this. First that we were including the seismic efficiency in our seismic moment tensor calculations as it was included in the Nishiyama et al., 2021 calculations but was not necessary for us due to the different simulation software. Fixing this helped, but the seismic efficiency was still 3 orders of magnitude short of what we expected. After several more simulations and experimenting with inputs, it turned out that because we had our simulated seismic source on the surface of the asteroid a large portion of the energy was being sent out into space instead of into the asteroid; the fix was to shrink the size of the source as small as we reasonably could and change the position to be very slightly below the 3D model’s surface.

Our outputs included both animations to visualize the seismic waves traveling through the asteroid and seismograms (note, the animations are sped up 10 times):

Homogeneous Model

2-Layer Model

“Blobby” Model (note that due to the random distribution of the blobs, this cross-section only intersects 2 of them)

In the end, we did determine that this would be possible with current technology, but there were several limitations primarily relating to time and computer ram constraints on the size of our simulations. For more detail, take a look at the poster and GitHub repository linked earlier, I have also included a Python script used to read the data output by the simulations on the Programming page of this website.

Tests of Seismic Reduced Impact-Induced Seismic Resurfacing on Rubble Pile Asteroids (Upcoming)

This is my planned term project for my GLG 404 class. I have not started yet, but will present the idea to the class on Thursday, October 8th. If it is approved I will then begin working on it. For the time being, I will explain the motivations and plan, and update this section as I work through the project.

As far as why I want to do this research; the Nishiyama et al., 2021 paper that was cited in the PsycheESE project found that the amount of seismic resurfacing as a result of the SCI impact on Ryugu was significantly less than expected - the furthest any of the boulders they measured moved was less than a meter. They propose several possible explanations as to why, and I have a few more hypotheses of my own. This includes breaking and/or compression of material, regolith behaving non-elastically in microgravity, and my own idea that maybe the seismic moment tensor approxation of an explosion is not valid for small impacts like the SCI. I have some ideas to test these, both experimentally and in simulations. So, I am going to perform those experiments, though perhaps not as rigorously as I would like due to the limited equipment available. If I have the time and resources I will also run the seismic simulations as a stretch goal.

The plan for the experiments is to drop a ball bearing into a box of coarse sand. I will have small items of known mass scattered around it to approximate Ryugu’s boulders. I can measure the movement of the “boulders” to replicate the Nishiyama et al., 2021 methodology, as well as take measurements of the crater’s depth and diameter and how much the sand at the bottom was compressed. If I am able to run the simulations, I will use the methodolgy developed in the PsycheESE project and place my recievers at the locations of Ryugu’s boulders to test how much seismic energy is transferred to them and can test different seismic moment tensors to see if one more closely matches what was observed on Ryugu and/or my sandbox.

Crater Counting on Charon (2023)

Though for a class, this is arguably my first research project; taking place during my second semester at ASU when I took SES 123 Earth, Solar System, and Universe; Lab. The concept is fairly well explained by the title, I counted the craters on Pluto’s largest moon, Charon. As I was very inexperienced and did not know what kind of tools may exist for this task, I did it by hand, it was incredibly tedious but also kind of fun. My results confirmed existing hypotheses, that Charon has (or relatively recently had) some level of geologic activity - likely cryovolcanism - which resurfaced some parts of the surface.

My process was to take the following mosaic of New Horizons images from the USGS astrogeology website and open it in photoshop.

I then applied a grid over it that was scaled such that each square was approximately 32 square kilometers (not accounting for deformations due to the map projections). Then I went through each grid square and counted the number of visible craters - due to the resolution the smallest were approximately 1.5 kilometers in diameter. The number of craters in these grid squares could then be used for analysis. The figures and images from this project are included below:

This is a heat map showing the locations of the craters I counted. The white outlines are craters that overlapped more than one grid square, the black areas are those that were unable to be counted due to images either not existing or being too low resolution. Note that due to an error when saving the image, some of the grid lines are faint or invisible.

These bar graphs show the number of craters and how they are distributed. An interesting detail is that the Northern Hemisphere, being better imaged, has nearly 4 times as many grid squares included than the Southern Hemisphere (1028 vs 262). Despite this, I counted only 20 more craters in the Northern Hemisphere.

Astrobiology

Diagnosing Life (2026)

This project was also completed for a class; SES 311 Astrobiology. It was the final project for the class and not just a group project, but one where the group was all the students in the class. The class was largely discussion based and this project came out of a debate on different ways of defining life. One student suggested that since it was so difficult to make a solid definition, what if we used a set of diagnostic criteria like the DSM used in psychology and mental health. Our professor, Dr. Sara Walker, decided to have us create these diagnostic criteria, apply them to a number of possible and/or disproven astrobiological life detections as case studies, and write a scientific paper-styled report on it as our final project. Each criterion and each case study had 2-3 students assigned to work on it (the criterion groups and case study groups were largely not the same), then we wrote the paper together.

The sections I was assigned were the “Life can Create” criterion and the possible detection of DMS/DMDS on exoplanet K2-18b. In addition to my assigned sections, I wrote both the abstract and Appendix B. I also contributed to the Limitations and Next Steps section and the Conclusion, as well as handling much of the formatting, works cited, and some general editing. The details of the methodology and results can be found in the final paper here (warning: 40 pages including appendix and references). Instead, I’m going to use this space to discuss how I would like to improve upon this project, as myself and several of the other more involved students agreed that we really liked the concept, but our execution needed significant work.

The biggest issue we agreed on was that there was a lot of overlap between the criteria, including my own. I think a lot of this comes from our limited timeframe - only about a month for the entire project - but also that we worked on each criterion in a small group with little overlap or communication between them until we tried to apply the codes and write the paper. The other students and I agreed that if we were to build upon or redo this project in the future, we would spend a lot more time ensuring the criteria are unique.

Beyond just cleaning up the criteria/codes, one specific improvement I would implement would be to separate the criteria into two groups, one would work like the existing ones do, the other would be for things “that are strongly indicative of life’s presence when they are present, but are not found in all life…” (this paper, page 32). This group of criteria could be simple yes or no questions, a yes response indicates nearly certain detection of life, while a no just means we can move on to the next question or section. For example, “Do we detect a technosignature?” if yes, then we have (probably) found life, if no, then we move on to the other criteria since life can easily exist without technology. Another improvement would be to differentiate between answers of “no” and “don’t know” for criteria. In other words, some way to quantify the uncertainty in the output. There are two possible ways to go about this described in the Limitations and Next Steps section of the paper.

In all, I think these would dramatically improve the reliability and consistency of the results. That being said, the current version still does an exellent job of showing how little information we have to verify if a potential biosignature is actually evifence of life, assuming of course that the detection itself can be confirmed.