Originally published in our August 2022 newsletter (Issue 16)
As a software engineer in the MSD community, Travis focuses on enabling efficient, scalable, and reproducible modeling workflows for a broad range of research teams and experiments. This can range from writing highly parallelizable code for pre- and post-processing large unwieldly datasets to updating existing models with modern tools for portability and ease of use and creating websites that facilitate science communication and reproducible methodology.
Travis Thurber is a software engineer in the Earth Systems Predictability and Resilience group at Pacific Northwest National Laboratory. Prior to this role, Travis worked on medical record software at Epic Systems, provider- and patient-facing healthcare applications at Seattle Children’s Hospital, and various technology startups. Travis earned a master’s degree from the King Abdullah University of Science and Technology and a bachelor’s degree from Columbia University, both in Mechanical Engineering with a focus on computational fluid dynamics.
As a software engineer in the MSD community, Travis’ role focuses on enabling efficient, scalable, and reproducible modeling workflows for a broad range of research teams and experiments. This can range from writing highly parallelizable code for pre- and post-processing large unwieldly datasets, to updating existing models with modern tools for portability and ease of use, to creating websites that facilitate science communication and reproducible methodology. Any given day can see Travis pulled in multiple directions for tasks big and small; read on for a few examples!
As the current technical leader of the mosartwmpy large-scale water routing and management model, Travis is responsible for ensuring the model is performative, documented, and easy to use. Such considerations are critical in enabling a research team to consider thousands of model runs in an uncertainty quantification experiment. To facilitate ease-of-use and enable early career data scientists to contribute more easily, Travis translated mosartwmpy from the Fortran language into Python. During this translation, it was important not to lose out on the speed and performance offered by Fortran. To accomplish this, Travis implemented the numba library for harnessing the power of fast math in C within Python computational loops.
In an ongoing exploratory analysis experiment studying water availability in the Colorado basin, Travis developed a data transformation tool to reformat StateMod model outputs from a schema-less raw text format to a highly compressible, columnar data format known as parquet. The parquet schema is designed to enable running complex queries over multi-file datasets quickly and in parallel. While such a transformation is not particularly novel in the software engineering community, applying these techniques to a large ensemble exploratory analysis experiment enabled the research team to efficiently mine insights from the immense dataset; a task that would be nearly intractable with the original data format.
Finally, since he has a background in web app development, Travis is often called upon to assist in efforts of science communication. A recent example is the creation of an automated deployment pipeline using GitHub Actions for the Addressing Uncertainty eBook. This pipeline enables the authors to focus solely on developing and updating content, instead of spending time compiling restructured text or manually updating web servers. Instead, just click a button and moments later a new version of the book is minted, published, and live!
Highlighted Articles:
Thurber, T., Vernon, C., Sun, N., Turner, S., Yoon, J., & Voisin, N. (2021). mosartwmpy: A Python implementation of the MOSART-WM coupled hydrologic routing and water management model. Journal of Open Source Software, 6(PNNL-SA-161232).
Smith, A. D., Stürmer, B., Thurber, T., & Vernon, C. R. (2021). diyepw: A Python package for Do-It-Yourself EnergyPlus weather file generation. Journal of Open Source Software, 6(64), 3313.
Reed, P.M., Hadjimichael, A., Malek, K., Karimi, T., Vernon, C.R., Srikrishnan, V., Gupta, R.S., Gold, D.F., Lee, B., Keller, K., Thurber, T.B, & Rice, J.S. (2022). Addressing Uncertainty in Multisector Dynamics Research [Book]. Zenodo. https://doi.org/10.5281/zenodo.6110623