The goal of this project is to successfully detect dust devil signals within infrasound data gathered in the Mojave Desert in Nevada. Dust devils, or dust-loaded convective vortices, feature a low pressure center, which reads on infrasound sensors as a characteristic heart-beat shaped dip and peak signal. Leveraging the specific signal that dust devils create, we plan to use waveform cross correlation to try and detect them automatically, trialing the procedure first on generated synthetic data to appraise the accuracy and limitations of the algorithm in actually detecting dust devils. The larger goal of this project is to better understand the factors controlling dust devil formation and size, with the purpose of using that information to better plan Martian missions to take advantage of the solar panel clearing that dust devils provide.
I'm doing alright and my project is clipping along, sorry this post has taken me so long to make. I have to run everything I post about my project through Sandia's review and release system so that added quite some time to the process.
Week 1 Blog
I would like to focus my time at this placement on skill-building. My coding skills and signal processing knowledge are at the moment very limited, and already the reading for this project have gone beyond what I’ve learned in Earth Physics so far.
I would like to get my work space set up and begin working on the actual project by the end of the first third of this internship. This week has seen many upon many technical and security challenges, so working past those will be my primary goal for these first few weeks. Getting my computer to work, getting it to work off-site, accessing my intern page on Sandia, getting Python to run on my computer, and setting up my Python environment for the project are all things that need to get done before I can do further work. I would also like to do most of the preliminary reading during this first third, which i have already started to tackle during my IT woes.
I would like to get more comfortable with the workflow of this project dur- ing this middle section. I’m hoping that some of the trickier concepts that this project involves, mainly the specifics of correlation and convolution will become clearer as I get more hands on experience with the code. I’m not sure how far into the project’s objectives I’ll be able to get by week 8, but I’m hoping i will be in a comfortable place to prepare to wrap up and start planning for AGU in the fall. Due to Sandia’s security protocols, I will need to have my AGU abstract ready to be checked and approved for release by July 1, so that is something concrete to have finished by this middle section.
I’m not sure what concrete things I want to get done in this final section. I would like to leave this internship feeling prepared/knowledgeable on what I need to do in order to successfully present at AGU in the fall. By this point I want to feel comfortable running the cross correlation codes and potentially be able to write my own code when needed. This summer will mostly involve using Python so I’m hoping to become a more proficient coder, especially with modules that I haven’t used much like Obspy and Scipy.
Week 2 Blog
We will be using infrasound data collected in the Mojave Desert in Nevada over a period of several years. The dataset is unusually complete and exten- sive compared to most infrasound collections, and is fairly recent. I’m not sure how many researchers have already worked with this data, or even if it’s open-source. I would say that this is a pretty strong data set, although I don’t think that its sensors have a geometric distribution, and infrasound itself seems more finicky to me than seismic data, being more localized and more prone to obscuring noise. I believe that the data has already had its most preliminary processing done for me, or at least had the procedure auto- mated. The course of the project will be using a cross correlation algorithm to detect and identify dust devil signals/events, and from there evaluating the accuracy of the procedure.
Skill : ’Identify and use a range of relevant bibliographic and virtual sources related to your research’
To be proficient in this skill means to be comfortable and familiar with the literature relevant to your research area. I am definitely not up to date with the research surrounding seismology or infrasound research, but I’ve already read a few background papers for this project. Being proficient with using the literature is key to being able to inform the direction of your research. Knowing what others have already done gives insight into potential questions to investigate, and keeps us from reinventing the wheel by repeating someone else’s research with no alteration.
Week 3 Blog
I thought that the elevator pitch assignment was pretty easy. I’m not always the clearest communicator, but I think that science is really only useful if you can explain it to other people. I really value science communication that is as straightforward and intelligible as possible, and I have some very limited experience with this in explaining my science classes to my parents and my liberal arts major friends. I think that having a description of my research in everyday language will be useful whenever I need to talk to someone about myself. Most people are curious about other people’s careers and interests, and having a way to describe my project without a brick wall of vocab words makes for more pleasant conversations. Beyond just social interaction, being able to sell yourself in a succinct and engaging way is key to making professional connections, and even if I’m not very charismatic, having an idea for how to describe my field in an approachable way I’m sure will help me. i think that science communication is an essential skill to have in order to be a competent researcher, and I hope to work on improving mine during this internship.
Week 4 Blog
With the 4th of July there hasn’t been a whole lot to write home about project-wise. I did get to start really digging into coding this week which has been both good and bad. I’m really pleased with what I’ve been able to do, the PANDAS data formatting that I had to learn as I went along during the school year has come in handy. I managed to read in csv files, create csv files from my own dataframes, separate out weirdly formatted NOAA data entries, learn some snazzy new matplotlib plots, and transcribe a really hefty function out of excel. My biggest current frustrations are mostly with the present state of my own skills and the small stumbling blocks with coding that I’ve been running into. I’ve also been having a hard time keeping up with IRIS assignments for a few reasons, chiefly that I can’t access my per- sonal accounts when onsite at Sandia so I have to do them while at home, and knowing that they’re building up has been a bit stressful.
Dust devils aren’t a major force on Earth, but understanding their character- istics could be crucial to future missions on Mars. Dust devils are essentially whirlpools of air that move dust around, and while not as powerful as ter- restrial storms, the dust devils on Mars can have far reaching effects. On Mars, dust and wind action is the main driver eroding potential evidence of past life, and the dust that the vortices kick up can affect how much sunlight reaches the planet’s surface, and thus affect its climate. Dust devils can also knock the dust off of solar panels, prolonging the lives of the Martian Rovers like one dust devil did for the Spirit Rover in 2005. To study Martian dust devils and better prolong the lives of expensive instruments on Mars’ surface, we can look for them in similar environments on Earth, such as in the Mojave Desert. Pressure sensors, specifically Hyperion infrasound pressure sensors, due to a quirk in their operation, register dust devils as a distinctive signal, and we can create computer programs to search through their recorded data for that signature using cross-correlation, which is essentially an automated way to detect a signal based on how similar it is to a defined template or shape.