College of Engineering | Apply to CSU | Disclaimer | Equal Opportunity Statement | Privacy | Search CSU

Blending Model Output with Satellite-Based and In-Situ Observations to Produce High-Resolution Estimates of Population Exposure to Wildfire Smoke

You must be on the CSU network—either physically or using VPN—to watch this or any of the videos on this site.

September 8, 2016
Will Lassman
Hosted by Jeff Pierce (advisor), Emily Fischer, Russ Schumacher, Sheryl Magzamen (Environmental and Radiological Health Sciences) Gabriele Pfister (UCAR)


In the western US, emissions from wildfires and prescribed fire have been associated with degradation of regional air quality. While atmospheric aerosol particles with aerodynamic diameters less than 2.5 μm (PM2.5) have known impacts on human health, there is uncertainty in how particle composition, concentrations, and exposure duration impact the associated health response. Due to changes in climate and land-management, wildfires have increased in frequency and severity, and this trend is expected to continue. Consequently, wildfires are expected to become an increasingly important source of PM2.5 in the western US; therefore, there is a need to develop a quantitative understanding of wildfire-smoke-specific health effects. A necessary step in this process is to determine who was exposed to wildfire smoke, the concentration of the smoke during exposure, and the duration of the exposure. Three different tools are commonly used to assess exposure to wildfire smoke: in-situ measurements, satellite-based observations, and chemical-transport model (CTM) simulations, and each of exposure-estimation these tools have associated strengths and weakness.

In this thesis, we investigate the utility of blending these tools together to produce highly accurate estimates of smoke exposure during the Washington 2012 fire season for use in health studies. For blending, we use a ridge regression model, as well as a geographically weighted ridge regression model. We evaluate the performance of the three individual exposure-estimate techniques and the two blended techniques using Leave-One-Out Cross-Validation. Due to the number of in-situ monitors present during this time period, we find that predictions based on in-situ monitors were more accurate for this particular fire season than the CTM simulations and satellite-based observations, so blending provided only marginal improvements above the in-situ observations. However, we show that in hypothetical cases with fewer surface monitors, the two blending techniques can produce substantial improvement over any of the individual tools.