84th AMS Annual Meeting

Wednesday, 14 January 2004: 1:45 PM
Diagnostic evaluation of gas-phase photochemical dynamics in the U.S. EPA Community Multiscale Air Quality modeling system (CMAQ): testing chemical indicators in the model and with observations at the 1999 special chemistry sites in Nashville, TN and Atla
Room 612
J. R. Arnold, NOAA/ARL and USEPA/ORD/NERL, Seattle, WA; and R. L. Dennis
Work with observations and models carried out for the multiyear, multisite Southern Oxidants Study has been interpreted to suggest that cities in the southeastern U.S. will require controls on regional sources of NOX rather than on urban sources of VOCs to meet the U.S. [O3] standard. A number of chemical species ratios have been developed in recent years by Kleinman, Milford et al., Sillman, and Tonnesen et al. in hopes of indicating whether NOX or VOC control in a given area should be preferred. These chemical indicators have been tested to varying degrees in models and with some observations by Kleinman et al., Olszyna et al., Arnold et al., and others.

We ran several probes of photochemical dynamics and indicator species ratios against specialized observations and model results to help delineate photochemical production rates and other process information in the atmosphere and in the model for July 1999 when the SOS99 and SEARCH99 special chemistry sites were operational in Nashville, TN and Atlanta, GA. These sites provided high precision measurements for important species not routinely measured including NO2, a key component in several of the proposed indicators. Model simulations were made on a continental 32km domain, an 8km southeast U.S. domain, and two 2km domains centered on Nashville and Atlanta, and all were performed with two chemical mechanisms, Carter’s SAPRC99 and Gery et al.’s CB4.

Some of these probes we developed through process-oriented studies using theoretical assumptions, model-derived explanations, and results from instrumented models ranging from 1-D box models to the full 3-D photochemical modeling system performed by Tonnesen et al. Table 1 shows a listing of diagnostic elements that can be used to probe the photochemical dynamics of O3 production (P(O3)).

Table 1. Diagnostic probes of photochemical dynamics.

___________________________________________________________

Individual Component Aspects

· HOX initiation

· HOX termination

· termination pathway competition

· air mass aging

Process Aspects

· HO production

· HOX propagation

· HO propagation efficiency, Pr(HO), or the average fraction of HO recreated in the photochemical cycle of OH reactions

· HO chain length, or the average number of times a new radical cycles through the photochemical system before termination

· NOX chain length, or the average number of times NO is recreated and cycled before termination

· O3 production efficiency, P(O3) per NOZ, where NOZ=NOY - NOX

Response Surface Aspects

· location of the [O3] ridgeline in the response space

· system state in the response space relative to the [O3] ridgeline

· slope of the radical-limited response space region

· slope of the NOX-limited response space region

___________________________________________________________

Here we report results from tests using one probe from each of the first two categories, air mass aging using the ratio [NOZ]/[NOY], and O3 production efficiency using the relationship between [O3] and [NOZ], and a probe of two response surface aspects, location of the [O3] ridgeline in the response space and the photochemical system state in the response space relative to that ridgeline. Further, in order to ensure that rate and process information from the atmosphere or the model is not confounded by inappropriate averaging over these diurnally changing photochemical dynamics, we use one of the response surface probes, [O3]/[NOX], as a chemical filter to separately select NOX-sensitive hours in the observations and the simulations. We also report results from our analysis of other suggested indicator ratios, including [HCHO]/[NO2] and [HO2H]/[NOZ], in the observations and the models. Model results are compared against the observations across grid spacings and for each chemical mechanism to help determine whether smaller grids or newer, more complete chemical representations, for example, might be preferred.

The diagnostic probes of photochemical dynamics are shown to provide reliable process information chiefly about the aggregate P(O3) rate and about some details of NOX cycling in determining that rate both in the atmosphere and in the model. We note that these probes give consistent information about the model processes and are particularly useful when used together. For example, the [O3] response surface indicator [O3]/[NOX] can broadly characterize a photochemical system: low values of the ratio tend to be associated with early morning periods when [NOX] is high and [O3] low, and higher values of the ratio are associated with hours late in the photochemical day when conditions have reversed. However, the values of this ratio are not uniquely determined and could be ambiguous with regard to the response surface ridgeline if the system O3 response were not also filtered by an indicator of air mass aging, the fraction NOZ/NOY, for example. In a similar way, information about the production efficiency of O3 per NOX converted is sharpened significantly when filtered by the response surface indicator to be centered in the region of strongest NOX limitation.

Application of these probes to CMAQ for the SOS99 and SEARCH99 cases shows that the model is for the most part simulating well the atmosphere’s photochemical dynamics at these sites in July 1999 when not confounded by problems in the emissions inventories or the meteorology simulations. We identify some such confounding problems and suggest possible solutions. Moreover, NOX aging, P(O3) efficiency, and transition across the [O3] response surface in the model are all largely concordant with these dynamics in the atmosphere. Taken together, these tests applied in the NOX-limited region of the [O3] response surface substantially increase our confidence that the model is performing correctly and demonstrate the power of these diagnostic probes for helping to elucidate issues of production dynamics in the models and in the observations.

Supplementary URL: