For more than a decade, ARL research related to atmospheric pollution modeling has consciously followed a path involving both Lagrangian and Eulerian models. The former follow a puff of pollution as it moves downwind from some selected source; they are well suited for relating sources to downwind effects. The latter takes a gridded view, with sources combined across grid cells, and with application well suited to the analysis of complex chemistry where emissions from various sources are interacting. These techniques should yield similar answers, but because their applications are generally quite different it is rare that opportunities for comparisons arise. A recent model intercomparison, performed by the European community, resulted in a first verification that the two approaches were consistent with each other. Current projects within ARL are serving to use these complementary modeling approaches in combination to create more powerful analysis capabilities.

The ARL Lagrangian HYSPLIT model has been used for many years and by many user groups world-wide to track long-range transport of effluent from volcanoes, major forest fires, and large scale dust events. The Eulerian Community Multiscale Air Quality (CMAQ) model is a mesoscale numerical model simulating regional and local photochemical ozone smog, fine particulate matter, acid/nutrient/mercury deposition, and air toxics. Recently both models have been used by NOAA/NWS for air quality forecasting of smoke from wildfires and regional ozone. Research is now being conducted to use both models in combination for air quality forecasting and assessment, with HYSPLIT tracking and quantifying large-scale transport of primary pollutants from large wildfires and dust storm events occurring outside of the continental U.S. gridded CMAQ domain. The HYSPLIT model will be capable of transporting pollutant mass of primary gases and particles to the boundaries of the CMAQ model domain and defining the spatial and temporal variations of boundary mass fluxes while sources within the U.S. model domain will be accounted for by the CMAQ model processes.

The capabilities and strengths of the CMAQ and HYSPLIT modeling frameworks are being utilized in the same study to examine the impacts of recent emission reductions of nitrogen oxides (NOx) at major point sources on ozone concentrations in the eastern United States. Photochemical simulations by the CMAQ model on a large domain with a 12-km grid cell size have been performed to generate ozone and other pollutant species concentrations for all days from June through August of 2002 and 2004. Separate model simulations were conducted with existing 'base case' emissions for 2002 and with 'revised emissions case', reflecting the NOx emission reductions from the electric generating units implemented during 2004. By taking advantage of the recent development of software tools linking the CMAQ and HYSPLIT modeling systems, the same meteorological fields used in the CMAQ simulations are being applied to exercise the HYSPLIT trajectory model in an effort to determine back-trajectories or forward trajectories from selected sites situated throughout the modeling region. In this study, a HYSPLIT back-trajectory follows the same path taken by pollutants in the CMAQ simulations in an upwind manner in space and time from a particular location in an effort to determine the origin of the air parcel containing the pollutant concentrations impacting a site (see Figure 1). Another new CMAQ/HYSPLIT modeling tool utilizes the spatial coordinates of each HYSPLIT back-trajectory to probe CMAQ's 3-dimensional concentration field in an effort to extract ozone and other pollutant values along each back-trajectory path within the gridded domain. In particular, model results from cases with back-trajectories passing through notable emission source regions, such as the Ohio River Valley, will be examined to assess the ozone concentration differences between the base case and reduced emission simulations. With the combined application of both modeling systems and the new computer software/visualization tools linking components of CMAQ and HYSPLIT, the results from this investigation are expected to provide beneficial quantitative information about the effectiveness of NOx emission changes implemented in particular source areas on reducing ozone concentrations in downwind areas of the eastern U.S.

In addition, the feasibility of developing an urban hybrid modeling system is being investigated. In this system, grid-based modeling of meteorology and atmospheric chemistry would be conducted by models such as WRF and CMAQ with horizontal grid cells of 1-5 km size. The resulting fine-scale urban meteorological fields would then be used to drive multiple HYSPLIT trajectories from selected source locations within the urban area to define a probabilistic envelope of trajectory ensembles and resulting dispersed pollutant concentrations. The combination of CMAQ's deterministic concentration fields and HYSPLIT's ensemble probabilistic concentration fields can then be used to better describe atmospheric dispersion within an urban area with a robust modeling framework. The combined CMAQ-HYSPLIT modeling system can be a powerful tool for understanding the processes affecting atmospheric transport, dispersion, transformation, and removal of pollutants and for predicting the concentration distributions over multiple scales, spanning local-to-urban-regional-to-continental in a consistent manner.

Figure 1. Map displays the path taken by each back-trajectory started from a different site (denoted by stars) in the CMAQ modeling domain. The wind fields used in this HYSPLIT back-trajectory application were supplied by CMAQ's MCIP (Meteorology-Chemistry Interface Processor) meteorological data sets.