Feb 2016 – May 2016
Data Science Researcher
Universidad Nacional de Colombia
Colombia
Research at CALAIRE (Air Quality Laboratory, Facultad de Minas) under Prof. José Fernando Jiménez, on forecasting critical air pollution episodes in the Aburrá Valley (Medellín, Colombia).
- Developed a feedforward neural network in Python to forecast next-day PM2.5 concentrations and flag critical episodes (daily average above 35 µg/m³).
- Derived mixing layer height from vertical temperature and humidity profiles measured by the microwave radiometer (Radiometrics MP-3000A) operated by SIATA, the valley's early warning system, improving knowledge of the atmospheric boundary layer (ABL) and using it as a model input.
- Combined boundary layer data with surface meteorology (temperature, humidity, solar radiation, precipitation, wind) and previous-day PM2.5 from the valley's air quality network.
- Contributed to the research line published in DYNA (2019): "Artificial neural network applied for the forecast of critical PM2.5 events in the Aburra Valley".