Research

Research Areas

Numerical Analysis (broad), Computational Fluid Dynamics (specialty), Machine Learning (new area)

Numerical Methods for fully evolutionary, coupled porous media flow with pollutant transport, resulting in PhD dissertation and six publications to date.Methodologies:  temporal finite difference methods, spatial finite element method, and time filters. Numerical experiments conducted with MATLAB and FreeFEM++.

Application of Machine Learning to problems in Atmospheric Science: In Spring 2020, completed 15-week NSF-sponsored CyberTraining in Big Data + High-Performance Computing + Atmospheric Sciences. Completed research project with publication on application of machine learning with feature importance analysis on environmental sounding data of supercell storms. Conducted summer research in 2021 with undergraduate students on the intersection of computational fluid dynamics and machine learning.Methodologies: Machine learning algorithms (research focused on Random Forest Classification and Convolutional Neural Network Models), high performance computing, parallel programming. Work completed using Python and MATLAB.

List of Publications

  1. Coffer, B.; Kubacki, M.; Wen, Y.; Zhang, T.; Barajas, C.; Gobbert, M. Brice. Machine Learning with Feature Importance Analysis for Tornado Prediction from Environmental Sounding Data. PAMM, 20, 1, (2021). https://doi.org/10.1002/pamm.202000112.
  2. Ervin, V.; Kubacki, M.; Layton, W.; Moraiti, M.; Si, Z.; Trenchea, C. Partitioned penalty methods for the transport equation in the evolutionary Stokes-Darcy-transport problem. Numer. Methods Partial Differential Eq. 2018; 35: 349-374. https://doi.org/10.1002/num.22303.
  3. Kubacki, M.; Tran, H. Non-Iterative Partitioned Methods for Uncoupling Evolutionary Groundwater-Surface Water Flows. Fluids 2017, 2(3). https://www.mdpi.com/222960.
  4. Ervin, V.J.; Kubacki, M.; Layton, W.; Moraiti, M.; Si, Z.; Trenchea, C. On Limiting Behavior of Contaminant Transport Models in Coupled Surface and Groundwater Flows. Axioms 2015, 4, 518-529. https://www.mdpi.com/116846.
  5. Kubacki, M. and Moraiti, M. Analysis of a Second-Order, Unconditionally Stable, Partitioned Method for the Evolutionary Stokes-Darcy Model. Int. J. Numer. Anal. Mod., 12 (2015), pp. 704-730. http://www.math.ualberta.ca/ijnam/Volume-12-2015/No-4-15/2015-04-06.pdf .
  6. Jiang, N.; Kubacki, M.; Layton, W.; Moraiti, M.; Tran, H. A Crank-Nicolson Leapfrog stabilization: Unconditional stability and two applications, Journal of Computational and Applied Mathematics, Volume 281, June 2015, Pages 263-276, ISSN 0377-0427. https://www.sciencedirect.com/science/article/pii/S0377042714004336.
  7. Kubacki, M. Higher-Order, Strongly Stable Methods for Uncoupling Groundwater-Surface Water Flow (Doctoral dissertation). University of Pittsburgh D-Scholarship Database, http://d-scholarship.pitt.edu/21894/ (2014).
  8. Kubacki, M. Uncoupling evolutionary groundwater-surface water flows using the Crank-Nicolson Leapfrog method. Numer. Methods Partial Differential Eq., 29:1192-1216, 2013. https://onlinelibrary.wiley.com/doi/abs/10.1002/num.21751.
  • Technical Report
  • Brice Coffer, Michaela J. Kubacki, Yixin Wen, Ting Zhang, Carlos Barajas, and Matthias K. Gobbert. Using Machine Learning Techniques for Supercell Tornado Prediction with Environmental Sounding Data. Technical Report HPCF-2020-18, UMBC High Performance Computing Facility, University of Maryland, Baltimore County, 2020. http://hpcf-files.umbc.edu/research/papers/CT2020Team8.pdf

Links to Scholar Profiles