Authors: Xiaohe Yu*, UTD, David Lary, UTD, Lakitha Omal Harindha Wijeratne, UTD
Keywords: Airborne Particulates, Weather RADAR, ECMWF, Machine Learning
Session Type: Virtual Paper
Start / End Time: 9:35 AM / 10:50 AM
Room: Virtual 25
Presentation File: No File Uploaded
Airborne particulates such as pm2.5 and pollen could have significant health outcomes and impact environmental air quality. Accurate in-situ observation devices are expensive, and labor intensive for installation. Information on the spatial and temporal distribution of airborne particulates can be gleaned from weather RADARs. We seek to make use of the RADAR data in conjunction with in-situation observations and machine learning to estimate the airborne particulates in this study. Airborne particulates vary in size, composition and origin as well as their spatial and temporal distribution, machine learning approaches are particularly useful in this study because they allows us to learn from the limited observations available to us. The characteristic spatial and temporal scales of airborne particulates are explored using semi variograms. These features are then combined with the data from NEXRAD and in-situ observations to train and validate a suite of machine learning models.