During my time at Argonne National Laboratory, my work was split between three projects: gap-filling meteorological data using statistical and machine learning techniques, predicting soil greenhouse gas fluxes using satellite imagery, and reducing the runtime for a Generalized Extreme Value (GEV) distribution strong scaling analysis  of extreme weather events in using the Crux high-performance computing environment.

To address gap precipitation data, I developed multiple extreme gradient boost models, that can employ both within-site and between-site methods to estimate precipitation events and the amount of precipitation at each site.

To predict soil greenhouse gas fluxes,  I utilized machine learning techniques driven by vegetation indices and site-specific information to accurately predict CO2 soil emissions. However, I was unsuccessful in finding a link between satellite imagery and N2O, and CH4 emissions. 

Lastly, I performed GEV analysis of extreme weather events utilizing variable quantities of CPUs, threads, and processors. This analysis aimed to show that varying CPUs and threads yielded similar run times, while varying the number of processors used could drastically change the running time of the analysis, allowing for an optimized prediction of extreme value distribution estimation.