Discipline: Computer Sciences and Information Management
Subcategory: Geosciences and Earth Sciences
Session: 2
Room: Exhibit Hall
Frieda Farias - Columbia University
Co-Author(s): Bernard C. Chang, The University of Texas at Austin, Austin, TX; Maša Prodanović, The University of Texas at Austin, Austin, TX.
Throughout the past few years, researchers around the world have been using the University of Texas at Austin’s Digital Rocks Portal to store their porous material datasets. Continued access to the Digital Rocks Portal is very important as it allows researchers to reproduce experiments and reuse old data for projects such as image analysis and machine learning training. Currently, the Digital Rocks Portal has around 150 published projects and citations from more than 200 publications[1]. The lack of standardization for data formatting in the portal has made it difficult for researchers to use each other’s datasets, as every lab has its own practices for storing data. Previously, the portal’s current managers had already begun researching different file formats and creating programs for data processing. This summer, we wrote Python scripts to create an improved image processing interface for tasks such as visualization and pore space characterization. Our scripts were then used to organize the data on the portal and were combined to create the Digital Porous Media (DPM) Tools[2] Python module on GitHub. This will allow other researchers to process and interpret their own datasets more easily, so that the portal and its community as a whole can grow, furthering the use of scientific computing in fields such as geology and petroleum engineering.1. Chang, B., Digital Rocks Portal, 20222. B. Chang and F. Farias, “Digital Porous Media (DPM) Tools.” University of Texas at Austin, 2022. [Online]. Available: https://github.com/BC-Chang/dpm_tools
Funder Acknowledgement(s): The National Science Foundation Award #2150390
Faculty Advisor: Dr. Maša Prodanović, masha@utexas.edu
Role: I wrote a Python script to gather data about the most popular file formats on the Digital Rocks Portal and then analyzed the data to determine that TIFF files were the best file type for the data on the portal. I also wrote several data conversion scripts, and added everything to the Digital Porous Media (DPM) Tools Python module io functions (https://github.com/BC-Chang/dpm_tools/tree/master/dpm_tools/io).