Severe storms & machine learning
A 30-year ERA5 framework for estimating tornado, severe-hail, and severe-wind occurrence across the Ohio River Basin.
Ohio River Basin · 1996–2025Submitted to AIES
Scientist. Researcher. Educator.Pinkrah Nana Ofosuhene · Atmospheric scientist
Atmospheric questions.
Data-driven answers.
I am an M.S. researcher in Geosciences at Western Kentucky University, working with the NSF Kentucky EPSCoR CLIMBS project. My research connects severe-storm environments, climate variability, and probabilistic hazard prediction.
My work spans the Ohio River Basin, the wider United States, and Ghana—from large-scale reanalysis and climate-model workflows to station observations and regional weather simulations.
I work with Python, xarray, MetPy, scikit-learn, and XGBoost, using NCAR’s Derecho and Casper systems and TACC’s Stampede3 for atmospheric data processing.
Explore my research approach ↗Severe-storm climatologyClimate data scienceNumerical weather predictionHydroclimatology
Selected work
Research in focus.
Compound drought & heat waves
A 75-year observational analysis of compound drought–heat-wave occurrence, severity, and sensitivity to event definitions.
Contiguous United States · 1951–2025Research in progress
Tourism climate suitability in Ghana
Station-based seasonality, ERA5 evaluation, long-term trends, and the relationship between climate suitability and visitor arrivals.
Ghana · 1981–2023Manuscript in preparation
Recent work
On the research desk.
2026 · Manuscript2026 · Research25 Sep 2026 · Upcoming
Severe-weather machine learning submitted to AIES
107 ERA5 diagnostics, two ensemble models, and independent testing for three hazards.
A broader view of Ghana’s tourism climate
Updated station seasonality, ERA5 evaluation, trends, and tourism-arrival context.
Preparing for Kentucky’s SuperCollider meeting
Severe-weather case studies for the CLIMBS research showcase in Lexington.