AGU Fall Meeting 2021

The Use of Machine Learning to Infer the Transport Influence of US West Coast WildFires on the US East Coast Planetary Boundary Layer

, , , and

According to California’s Office of Emergency Services, during the early part of September 2020, California experienced unprecedented wildfires due to a combination of dry weather, heat waves, and lightning strikes. The wildfires created extreme devastation in California and surrounding areas. It was also observed that smoke resulting from the wildfires was transported across the United States. Understanding the potential effects that transported wildfire smoke can have on air quality is of vital concern to the health of the individuals at risk living some distance from these events. In this study, we performed an analysis of the effects of the September 2020 California wildfires in the United States on the Planetary Boundary Layer (PBL) along the East Coast, both from multiple ground-based ceilometer Lidar observations and with the WRF-CHEM model output for the timeframe of September 9, 2020 to September 19, 2020. We conducted two model simulations, one with daily MODIS and VIIRS fire emission input data and the other without the emissions. We also simultaneously streamed LIDAR aerosol backscatter data from three LufftCHM15K ceilometers located on the East Coast in Bristol, Pennsylvania [40.10N,74.85W], Baltimore, Maryland [39.27N,76.73W] and Blacksburg, Virginia [37.23N,80.41W] for this same timeframe and generated daily images of aerosol height concentrations. Employing a holistic edge detection machine learning algorithm, we identified the transport of smoke that reached the East Coast from West Coast wildfires on Sept. 17, 2020 and the planetary boundary layer heights for each of the ceilometers. We further observed, for the ceilometer data for Catonsville, the descent of the smoke into the PBL on Sept. 19, 2020. We find similar increases in the PBL height for the WRF-CHEM model simulation with the fire emission data. To our knowledge, this study is the first to use machine learning to demonstrate that the long-distance transport of wildfire smoke can penetrate into the PBL and subsequently increase particulate matter and impact air quality


American Geophysical Union

UMBC ebiquity