Interns are a regular part of the IMPACT team. Again and again, interns have brought significant value to IMPACT projects with their fresh perspectives, enthusiasm, and willingness to learn. Their unique outlooks challenge established patterns and can lead to creative problem-solving and innovative solutions. These IMPACT interns also infuse teams with a surge of energy, fostering a dynamic and collaborative atmosphere that invigorates project discussions and drives productivity. This summer was no exception.
Visualizing Flight Paths
Anabelle Brodsky, a rising junior majoring in computer science at Boston University, interned with the Airborne Data Management Group (ADMG) within IMPACT. She supported their efforts to develop new features for the Catalog of Archived Suborbital Earth Science Investigations (CASEI), a comprehensive search and discovery interface for accessing information about NASA suborbital research. Her “Take Flight with CASEI” project focused on converting aircraft meteorological navigational datasets into visualized flight paths that could be displayed on terrain maps and ultimately integrated into CASEI’s metadata collection.
USRA Earth Scientist and ADMG team member Elijah Walker mentored Anabelle as she analyzed datasets from three airborne campaigns: the Investigation of Microphysics and Precipitation for Atlantic Coast-Threatening Snowstorms (IMPACTS), Fire Influence on Regional to Global Environments and Air Quality (FIRE-AQ), and the Long Island Sound Tropospheric Ozone Study (LISTOS). Anabelle downloaded navigation datasets for each campaign from Earthdata Search and transferred the data granules into code that mapped each flight path on a basic two-dimensional map. After honing this process, she drafted a more advanced version of the visualization code that could display aircraft flight paths on a terrain map that more closely matched those currently featured in CASEI.
Anabelle shared that she had a positive experience working with the ADMG team and CASEI:
This summer I was mentored by a team filled with intelligent and hard-working people. I always felt comfortable asking questions, and I gained skills I would not have learned in the classroom. Working with ADMG showed me that a diverse team is best suited for driving innovation and accomplishing amazing things.
In the final week of her internship, Anabelle met with Heidi Mok, a Development Seed developer who supports CASEI operations, and ADMG team lead Stephanie Wingo to discuss how her work would soon contribute towards a new feature for CASEI users to explore. Using Figma, a collaborative web design application, they prototyped possible implementations of flight path visualizations on CASEI’s campaign pages.
Check out Anabelle’s Linkedin profile here.
Opening Up the Smallsat Data Explorer
Noam Mayerfeld and Shrey Gupta were tasked with enhancing the Smallsat Data Explorer (SDX) tool. In response to the need for improved data discoverability and analysis, the two developed a data processing service that will enhance the way end-users interact with the expansive data archive within the SDX. This processing service empowers users with the ability to explore the data before downloading it. The service design allows users to select from an assortment of curated data processing algorithms. This open access to algorithms ensures that users, regardless of their technical expertise, can harness the power of complex analytical tools.
In describing the architecture, Shrey noted,
What’s nice about the workflow orchestrator is that it handles load. Because time is not an issue, we can control how much space we’re using (from a compute perspective) without worrying about downtime. We used Airflow with Lambda, a serverless compute instance, that will run the algorithms for up to 15 minutes asynchronously.
Commercial Smallsat Data Acquisition (CSDA) Program data are restricted to authenticated science users as defined by vendor specific end user license agreements (EULAs). The processing service generates derived data products which may be accessible to unauthenticated users, leaving the original data inaccessible. Consequently, the service dramatically broadens the accessibility and utility of CSDA data, catering to a wide range of individuals who can derive value from it.
Noam describes the value he got out of the internship:
My passion lies in building complex systems and programs that will make a real visible difference in the world. Earth Science research can play a major role in understanding, and maintaining, the planet we call our own, so I was privileged to play a small role in that effort.
Using Machine Learning to Track Abandoned Oil Wells
Tuong Phung, a senior at the Massachusetts Institute of Technology (MIT), interned with the Machine Learning team within IMPACT. Under mentor Muthukumaran (Kumar) Ramasubramanian’s guidance, he built on previous work by Lilly Thomas and used machine learning and high-resolution satellite data to detect and track abandoned oil wells. The overall goal of this project is to develop a system that can monitor wells over time and identify when active wells are abandoned.
Abandoned oil wells are a significant, but unpredictable, source of methane emissions. Identifying and tracking them enables researchers to better understand and ultimately mitigate their environmental impact. As part of his project, Tuong used Detectron2, a popular open-source research platform for object detection and segmentation to train a model on satellite images of active oil wells. He also analyzed high-resolution satellite data using Rasterio, a Python library for working with geospatial raster data and pyproj, another Python library used for cartographic projections and coordinate transformations.
Asked about his interest in this field of research, Tuong responds:
As a researcher, I am strongly interested in the applications of AI to problems in science and engineering. This project was a great opportunity to use machine learning to address an important problem related to climate change: the identification and tracking of abandoned oil wells to better understand and ultimately mitigate their environmental impact.
Global methane emissions are a recognized concern and monitoring them is a high priority for researchers. Data from the MethaneSAT mission (currently scheduled for a January 2024 launch) will be used in conjunction with the time series analysis system Tuong developed. By combining the data from MethaneSAT with the detection and tracking of abandoned oil wells, researchers will be able to identify regions of interest and gain a more comprehensive understanding of methane emissions and their sources.
Check out Tuong’s Linkedin profile here.