A Novel Ensemble Design for Probabilistic Predictions of Fine Particulate Matter Over the Contiguous United States (CONUS)

Abstract

This study examines the benefit of using a dynamical ensemble for 48 hr deterministic and probabilistic predictions of near‐surface fine particulate matter (PM2.5) over the contiguous United States (CONUS). Our ensemble design captures three key sources of uncertainties in PM2.5 modeling including meteorology, emissions, and secondary organic aerosol (SOA) formation. Twenty‐four ensemble members were simulated using the Community Multiscale Air Quality (CMAQ) model during January, April, July, and October 2016. The raw ensemble mean performed better than most of the ensemble members but underestimated the observed PM2.5 over the CONUS with the largest underestimation over the western CONUS owing to negative PM2.5 bias in nearly all the members. To improve the ensemble performance, we calibrated the raw ensemble using model output statistics (MOS) and variance deficit methods. The calibrated ensemble captured the diurnal and day‐to‐day variability in observed PM2.5 very well and exhibited almost zero mean bias. The mean bias in the calibrated ensemble was reduced by 90–100% in the western CONUS and by 40–100% in other parts of the CONUS, compared to the raw ensemble in all months. The corresponding reduction in root‐mean‐square error (RMSE) was 13–40%. The calibrated ensemble also showed 30% improvement in the RMSE and spread matching compared to the raw ensemble. We have also shown that a nine‐member ensemble based on combinations of three meteorological and three anthropogenic emission scenarios can be used as a smaller subset of the full ensemble when sufficient computational resources are not available in the operational setting.

Development and Evaluation of a North America Ensemble Wildfire Air Quality Forecast: Initial Application to the 2020 Western United States “Gigafire”

“…In addition, diurnal and day‐to‐day variations of wildfire behavior due to fuel aridity and availability, fire weather, fire containment activities and combustion stage can limit model forecasting performance during large wildfires (Saide et al., 2015). Besides, a variety of input data sets, such as meteorological fields and chemical transports (Garcia‐Menendez et al., 2013; F. Li et al., 2019; Y. Li et al., 2020) and plume rise schemes (Briggs, 1969; Freitas et al., 2007; Y. Li et al., 2023; Paugam et al., 2016; Sofiev et al., 2012; Stein et al., 2009; Vernon et al., 2018; Zhu et al., 2018), are implemented differently in each model and can also impact the AOD and PM 2.5 forecasting performance (Delle Monache & Stull, 2003; Kumar et al., 2020).…”

“…Li et al, 2020) and plume rise schemes (Briggs, 1969;Freitas et al, 2007;Y. Li et al, 2023;Paugam et al, 2016;Sofiev et al, 2012;Stein et al, 2009;Vernon et al, 2018;Zhu et al, 2018), are implemented differently in each model and can also impact the AOD and PM 2.5 forecasting performance (Delle Monache & Stull, 2003;Kumar et al, 2020).…”

“…Many operational forecasting systems have been developed to predict the dispersion and transformation of trace gases and aerosols, with the primary goal of mitigating the health effects of poor air quality (Campbell et al., 2022; Lee et al., 2017; L. Pan et al., 2014; Tang et al., 2015). However, the accuracy of deterministic forecasts from a single model is subject to uncertainties in emission and meteorological input data, as well as the physical and chemical processes of the dispersion or chemical transport models (Delle Monache & Stull, 2003; Kumar et al., 2020; Y. Li et al., 2020). One of the effective ways to improve forecasting performance is the ensemble approach, which can provide probabilistic forecasts by calculating the mean from either multiple models or varying inputs in a single model (Delle Monache, Deng, et al., 2006; Delle Monache, Nipen, et al., 2006; Delle Monache et al., 2008, 2020; Delle Monache & Stull, 2003; Y. Li et al., 2020; Petersen et al., 2019; Solazzo et al., 2012; Xian et al., 2019).…”

Chemical Weather and Chemical Climate

“…For both species, the correlation coefficient of the median is higher than the correlation coefficient derived from each of the nine models. The ensemble approach can be improved by attributing a weight to each model that is determined by an optimization procedure like a least‐square error minimization of the ensemble forecasts during a training period covering a few weeks (Pagowski et al., 2005) or via ensemble‐calibration using the variance‐deficit and model output statistics methods (Kumar, Alessandrini, et al., 2020). Large computational costs of a multi‐model ensemble can make their implementation in operations challenging, but this problem can be partially addressed using an analog‐based method of generating air quality ensemble (Delle Monache et al., 2020).…”