Alex Connolly
1038-26 Mudd Hall
520 West 120th Strreet
New York, NY 10027
Alex Connolly is a Postdoctoral Research Scientist at Columbia University in the lab of Prof. Pierre Gentine and a member of the M2LInES project and an affiliate of the NSF Science and Technology Center, LEAP. Alex studies the dynamics of the atmosphere, particularly boundary-layer processes and primarily using computational methods such as deep learning and simulation. A current focus is using data-driven methods to extract information from high resolution atmospheric simulations that can improve coarser, climate models which suffer stubborn biases related to boundary-layer and cloud processes, e.g. the stratocumulus-to-cumulus transition. Previously, Alex focused on applications of equivariant deep neural networks, trained in python, but implemented in a parallel C++ model to more accurrately predict 3-dimensional, turbulent momentum fluxes in the stably stratified atmospheric boundary layer in simulations which remain numerically stable, even in novel configurations, due to new methods for deep learning parameterizations that respect the vertical anisotropy of the atmosphere as well as other geometric and self-similarity constraints. Alex graduated from University of California Berkeley where he was advised by Prof. Fotini K. Chow and researched various land-surface and complex terrain effects on winds and thermodynamics of the atmospheric surface and boundary layers including buoyancy driven slope flows and topographic cold air pools that exacerbate concentrations of air pollutants.
