Uncertainties in emission processing steps with SMOKE and EPS2

Generally emission models, SMOKE and EPS2, follow the same processing steps; 1) data reading to importing the EI in various formats, 2) chemical speciation to speciate the lumped VOC and NOx emissions into specific model species in the chemical mechanism such as CB-IV and SAPRC99 employed in the AQMs, 3) spatial allocation to assign the county-based emissions onto each grid cell in the model domain of interest, and 4) temporal allocation to temporally resolve peak ozone day and annual average day emissions into hourly emissions or to reshape hourly emissions into generalized time coordinate in the model system (i.e. UTC).  During the procedures except data importing the emission models refer to internal files, so called cross-reference table and profiles (i.e. surrogates for spatial allocation, spilt factors for chemical speciation and hourly activity factors for temporal allocation). (MCNC, 2002; U.S. EPA, 1992; Hogrefe et al., 2003).  In addition to the Texas EI, TCEQ has developed Texas-specific cross-reference table and profiles to process the EI in EPS2.  For example, TCEQ has developed a set of surrogate data and VOC split factors for gridding and chemical speciation in the region (Funk et al., 2002; TNRCC, 2002).  However, unlike EPS2, SMOKE uses default nationwide cross-reference and profile data provided by U.S. EPA.  These emission processing systems may present different AQM-ready emission inputs depending on the uses of different internal database for each step of processing as well as the EIs used.  

After TEIPS, two emission modeling systems, SMOKE and EPS2, were compared by processing the Texas EI available for the HGA air quality studies, focusing on the effects of differences in spatial surrogates, chemical speciation and temporal allocation data employed in the systems to identify uncertainties at the emissions processing steps.  For each step of the EI processing, EPS2 uses the Texas EI specific cross-reference and profile data developed by TCEQ, and SMOKE uses the U.S. EPA cross-reference and profile data.  In order to estimate the impacts on photochemical model simulations, SMOKE/BEIS3 and EPS2/GloBEIS3 were compared with model-ready emissions inputs and CMAQ results.

VOC/NOx emission ratios

Difference in VOC/NOx emission ratio (mole C/mole) before plume rise of major point sources were presented. (left: EPS/GloBEIS3, right: SMOKE/BEIS3)

CMAQ results

Ozone simulations of CMAQ using different emissions from EPS2 and SMOKE were presented.  The hourly ozone simulation results were compared to those observed at 19 CAMS sites inside the domain during the period of Aug. 23 ~ 31, 2000.

Stationary measurements

As an example, simulated concentrations for the selected species were compared to those measureed at Clinton site (data from TCEQ's web site). 

Ozone predictions

When the simulated 1-hr and 8-hr ambient ozone concentrations exceed 120 ppb and 80 ppb, respectively, the daily occurrences in the domain were counted with number of cells (4x4 km).  For both 1-hr and 8-hr ozone concentrations, EPS2/GloBEIS3 presents more daily frequencies exceeding the standards.  However, the area of occurrence has changed by case.   The daily maximum simulation ozone concentrations are also shown for the two emissions.