GH016-11
Using Space/Time Data Fusion to Estimate PM2.5 Concentrations and Quantify the Acute Health Impacts of Smoke Exposure During the 2017 California Wildfires

Tuesday, 15 December 2020: 07:30
Virtual
Stephanie E Cleland1, J Jason West1, Yiqin Jia2, Steve Reid2, Sean M Raffuse3, Susan O'Neill4 and Marc L Serre1, (1)University of North Carolina at Chapel Hill, Chapel Hill, NC, United States, (2)Bay Area Air Quality Management District, San Francisco, CA, United States, (3)University of California Davis, Davis, CA, United States, (4)US Forest Service, Seattle, WA, United States
Abstract:
Exposure to wildfire smoke causes a range of adverse health outcomes. This health risk in combination with a predicted increase in the frequency and severity of wildfires due to climate change suggest the importance of accurately estimating smoke concentrations and quantifying the health impacts of smoke exposure. While chemical transport models (CTMs) and the spatial interpolation of observations are often used to assess smoke exposure, geostatistical methods can combine surface observations with modeled and satellite-derived concentrations to produce more accurate exposure estimates during wildfires.

Here we estimate ground-level PM2.5 during the October 2017 California wildfires, using the Constant Air Quality Model Performance (CAMP) and Bayesian Maximum Entropy (BME) methods to bias-correct and fuse together three concentration datasets: permanent and temporary monitoring stations, a CTM, and satellite observations. Four different BME space/time kriging and data fusion methods were evaluated for accuracy. We then use the most accurate PM2.5 estimations in a risk assessment to calculate the excess respiratory, cardiovascular, and asthma hospital admissions attributable to exposure to fire-originated PM2.5.

All BME methods produce more accurate estimates than the standalone CTM and satellite products, emphasizing the importance of combining multiple datasets to estimate smoke exposure. Performing a non-linear bias-correction on the modeled and satellite-derived concentrations, via CAMP, notably improves accuracy. Adding temporary station data to the BME estimation increases the R2 by 35%. The data fusion of observations with the CAMP-corrected CTM provides the best overall PM2.5 estimate (R2=0.73), especially in station-scarce regions. Including satellite data does not improve performance. Using these ground-level PM2.5 estimations, we estimate approximately 60,000 people were exposed to very unhealthy air (daily average PM2.5 ≥150.5 µg/m3) and 15.3 million people were exposed to concentrations greater than the EPA’s 24-hour PM2.5 standard, 35 µg/m3. We further estimate that smoke exposure during the fires caused 260 (95% CI: 124, 435), 73 (95% CI: -11, 171), and 28 (95% CI: 19, 85) excess respiratory, cardiovascular, and asthma hospital admissions, respectively.

<< Previous Abstract | Next Abstract >>