Wind-Blown Dust Modeling Using a Backward-Lagrangian Particle Dispersion Model

Abstract

Presented here is a new dust modeling framework that uses a backward-Lagrangian particle dispersion model coupled with a dust emission model, both driven by meteorological data from the Weather Research and Forecasting (WRF) Model. This new modeling framework was tested for the spring of 2010 at multiple sites across northern Utah. Initial model results for March–April 2010 showed that the model was able to replicate the 27–28 April 2010 dust event; however, it was unable to reproduce a significant wind-blown dust event on 30 March 2010. During this event, the model significantly underestimated PM2.5 concentrations (4.7 vs 38.7 μg m−3) along the Wasatch Front. The backward-Lagrangian approach presented here allowed for the easy identification of dust source regions with misrepresented land cover and soil types, which required an update to WRF. In addition, changes were also applied to the dust emission model to better account for dust emitted from dry lake basins. These updates significantly improved dust model simulations, with the modeled PM2.5 comparing much more favorably to observations (average of 30.3 μg m−3). In addition, these updates also improved the timing of the frontal passage within WRF. The dust model was also applied in a forecasting setting, with the model able to replicate the magnitude of a large dust event, albeit with a 2-h lag. These results suggest that the dust modeling framework presented here has potential to replicate past dust events, identify source regions of dust, and be used for short-term forecasting applications.

Implications of a shrinking Great Salt Lake for dust on snow deposition in the Wasatch Mountains, UT, as informed by a source to sink case study from the 13–14 April 2017 dust event

“…Overall, the WRF-STILT dust modeling framework was able to capture the timing and duration of the dust event ( figure 2(a)). Modeled dust emissions in figure 2(b) indicated that the Great Salt Lake Desert (GSLD), exposed portions of the GSL lake bed, Delta, and Lake Sevier were the largest emitters of dust during this event, which is consistent with results from previous work (Hahnenberger and Nicoll 2012, Steenburgh et al 2012, Mallia et al 2017. The majority of dust contributions towards ASP, according to WRF-STILT, indicated that the GSLD was the biggest contributor of dust (46%), followed by Lake Sevier (21%), Delta (13%), and other source regions across the eastern Great Basin (11%) ( figure 2(c)).…”

“…In an effort to quantify the impacts of wind-blown dust events along the Wasatch Front for the mid-April dust event, we used a backward Lagrangian modeling framework that was recently developed specifically for dust simulations (Mallia et al 2017). This model was able to successfully capture the timing and magnitude of two major dust events across northern Utah during the spring of 2010, when compared to a number of air quality stations along the Wasatch Front.…”

“…The simulations also incorporated a dry deposition scheme that depends upon aerosol size and the land cover (Zhang et al 2001). The dry deposition was carried out along each STILT backward trajectory, forward in time at 2 min time increments, identical to methodology described in (Mallia et al 2017). Dry deposition was calculated as Figure 1.…”

“…Several modifications were made to the default version of the FENGSHA model so that dust can be emitted from dry lake beds (playa) (Mallia et al 2017), which is a major source of dust across the eastern Great Basin (Hahnenberger andNicoll 2012, Steenburgh et al 2012). These changes included adding a new soil type category within the FENGSHA dust emission model, with its own unique friction velocity threshold.…”

Bayesian inverse estimation of urban CO2 emissions: Results from a synthetic data simulation over Salt Lake City, UT

“…All components are first given as hourly standard error (2 nd column) and, if applicable, correlation is applied between same-day observations using a decay time-scale parameter (3 rd column). Component errors are then aggregated to express observational error in enhancements that represent an average for a given afternoon (4 th column error components are not expected to be random within a given afternoon), while R transWIND is given decaying correlation with increasing time between observations, assuming a correlation timescale of 2.8 hours as in Mallia et al (2017). In this study we assume error to be constant across sites, and do not account for spatial correlation of measurements between sites; however, it should be noted that in a real-data application, errors are likely to be correlated between towers, especially if separation distances between sites are small (as they are in this study area).…”