# Selected Publications

This page provides an overview of essential themes in my scholarship in machine learning for science, planetary plasma transport, and education. You can find a full list of works at Google Scholar.

### Machine Learning for Scientific Insight

These works create and review methods for data intensive research and provide paths toward integrating data science into planetary science and space physics.

• Azari, A. R., Biersteker, J. B., Dewey, R. M., Doran, G., Forsberg, E. J., Harris, C. D. K., Kerner, H. R., Skinner, K. A., Smith, A. W., Amini, R., Cambioni, S., Da Poian, V., Garton, T. M., Himes, M. D., Millholland, S., and Ruhunusiri, S. [+31 co-signatories] (2020) “Integrating Machine Learning for Planetary Science: Perspectives for the Next Decade”. Submitted to the NRC Planetary and Astrobiology Decadal Survey. Full text availiable via arXiv.

This paper represents a community of 16 co-authors and 32 co-signers and our recommendations for machine learning in planetary science to funding agencies and the planetary community. It summarizes current efforts in the field in machine learning applied to planetary science, and compares this to other fields in the natural sciences. It concludes with recommendations in the next decade for supporting a data-rich future for planetary science.

• Azari, A. R., Lockhart, J. W., Liemohn, M. W., and Jia, X. (2020). “Incorporating Physical Knowledge into Machine Learning for Planetary Space Physics”. Access at Frontiers in Astronomy and Space Sciences .

This work details a process to include physical knowledge in planetary and space physics machine learning models, before demonstrating that this inclusion increases the interpretability of models for scientific knowledge gain. Within the Machine Learning in Heliophysics collection.

• Liemohn, M. W., Shane, D., Azari, A. R., et al. (2021) “RMSE is not enough: Guidelines to robust data-model comparisons for magnetospheric physics”. Accesible at Journal of Atmospheric and Solar-Terrestrial Physics.

This work summarizes data-model comparisons including event metrics and argues for increased data inclusive comparisons in physics models.

• Liemohn, M. W., Azari, A. R., Ganushkina, N. Y., and Rastatter, L. (2020) “The STONE curve: A ROC-derived model performance assessment tool”. Accesible at Earth and Space Science.

This work describes a new assessment tool for continuous data-model comparisons most similar to the ROC curve for classifcation analyses.

### Planetary Systems Plasma Transport - Terrestrial Planets

• Azari, A. R., Mitchell, D., Xu, S. (in prep) “The Draped Magnetic Field of Mars: Influence of Crustal Fields & Solar Wind”.

This work details a large-scale data-driven evaluation of Mars’ magnetic field response to the solar wind at low altitudes with MAVEN spacecraft data. This work was presented at AGU 2020.

• Dewey, R. M., Slavin, J., Raines, J., Azari, A. R., and Sun, W. (2020) “MESSENGER observations of flow braking and flux pileup of dipolarizations in Mercury’s magnetotail: Evidence for current wedge formation”. Access at the Journal of Geophysical Research: Space Physics.

This paper performed an automated identification of plasma transport in Mercury’s magnetic environment to compare this system to that of Earth’s. In analyzing transport events, this work found that a small fraction of events reached the nightside surface, which has implications for surface space weathering.

### Planetary Systems Plasma Transport - Outer Planets

• Azari, A. R., Jia, X., Liemohn, M. W., et al. (2019). “Are Interchange Injections Organized by Rotation Longitude?” Access at Journal of Geophysical Research: Space Physics.

This work demonstrates the limited influence of the upper atmosphere on interchange through systematically evaluating previous statistical studies over different sampling regimes.

• Azari, A. R., Liemohn, M. W., Jia, X. et al. (2018). “Interchange Injections at Saturn: Statistical Survey of Energetic H$^{+}$ Sudden Flux Intensifications”. Access at Journal of Geophysical Research: Space Physics.

This paper developed a new standardized, and automated, physics based identification method for interchange most similar to logistic regression, and used this to understand the statistical distributions of events. A figure in this paper was featured in the journal for its novel use of visualizations from other data-rich fields (genomics).

• Paranicas, C., Thomsen, M., Kollmann, P., Azari, A. R., et al. (2020). “Inflow speed analysis of interchange injections in Saturn’s magnetosphere”. Access at the Journal of Geophysical Research: Space Physics.

This work estimates the speed at which plasma is traveling toward Saturn, an essential number required to calculate system wide transport.

• Rymer, A…[46 total including Azari, A. R.] (2020). “Neptune and Triton: A Flagship for Everyone”. Submitted to the Planetary Science and Astrobiology Decadal Survey 2023-2032. Access online.

This paper represents the outer planets community’s motivation for sending a mission to the Neptune system in the near future. Neptune, to date, has never had an orbiting spacecraft. A mission to Neptune would advance cross-disciplinary planetary science, including: geologic studies for Triton, considering Neptune as a class of discovered exoplanets, and plasma transport studies. For plasma transport, Neptune has large non-dipolar magnetic fields, complicating our currently understood systems. Such a mission would represent the next several decades, given the time scale required for these missions, of scientific progress for these planetary systems, and understanding mass transport at the icy giants.

### Inclusive Computational Geoscience Education

• Azari, A. R. (2020) “A Data-Driven Understanding of Plasma Transport in Saturn’s Magnetic Environment” Ph.D. Thesis. University of Michigan. Access on the DeepBlue Thesis Database.

The appendix of this work discusses the implementation of Jupyter notebooks to provide an approachable experience for students into computation in a course for undergraduate and graduate students. We found that Jupyter notebooks naturally provided instructors to embrace a growth mindset style of learning by allowing students to attempt, and retry solutions. It also allowed instructors to grade computational assignments for partial credit, reinforcing a growth minded approach to student learning by the instructional team. This work was presented at AGU in 2019. Brian Swiger, Alex Shane, Drs. Liemohn and Huang-Saad, and myself are currently in the process of finalizing our assessment for submission to the Journal of Geoscience Education.