ROG/NOx ratio

For the single VOC - NOx or VOC - CO - NOx experiments, the model is able to simulate the A([03]-[N0]) to within 25% or better in most cases, which is better than the 30% seen in previous mechanism evaluations with the older chamber data (Carter and Lurmann, 1990, 1991 Gery et al., 1989, Carter, 2000). However, there are indications of non-negligible biases in model simulations of certain classes of experiments. The cleaner conditions and the relatively lower magnitude of the chamber effects may make the nm-to-run scatter in the model performance may be less than in the simulations of the previous data, and this tends to make relatively small biases in the model performance more evident. There are, for example, definite biases in the model to under predict O3 formation and NO oxidation in the surrogate - NOx experiments carried out at lower ROG/NOx ratios. These cases are discussed further elsewhere (Carter, 2004b). 

The two large open circles on Figure 1 show the ROG and NOx levels chosen to serve as the base case in the incremental reactivity experiments that are currently underway in this chamber to assess ozone impacts of various different types of VOCs (Carter, 2004c). The 25-30 ppb NOx levels were chosen to be representative of pollution episodes of interest in California, based on input provided by the staff of the California Air Resources Board, which is funding most of the current reactivity studies. The ROG/NOx ratios were chosen to represent two sets of conditions of NOx availability relevant to VOC reactivity. The lower ROG/NOx ratio was chosen to represent the relatively higher NOx conditions of maximum incremental reactivity (MIR) where ozone is most sensitive to VOCs, to approximate the conditions used to derive the widely-used MIR ozone reactivity scale (Carter, 1994). The higher ROG/NOx ratio was chosen to be one-half that yielding maximum ozone levels, and is used to represent conditions where ozone is NOx-limited, but not so NOx-limited that VOC reactivity is irrelevant. These are referred to in the subsequent discussion as the MIR and MOIR/2 base eases, respeetively. 

The ability of the SAPRC-99 mechanism to simulate ozone formation in representative MIR and MOIR/2 base ease experiments is shown on Figure 2. It can be seen that good reproducibility in ozone formation is observed in the replicate experiments, but that the model consistently under prediets O3 formation in the MIR experiments, though it gives much better predictions of ozone in the experiment at the higher ROG/NOx ratio. This tendency to under... 

Figure 3. Plots of die bias for the SAPRC-99 mechanism to vmder predict NO oxidation and O3 formation against normalized ROG/NOx ratios for ambient surrogate -NOx experiments in three environmental chambers.

Figure 4. Plots of the mechanism under prediction bias for model simulations UCR EPA ambient surrogate - NOx experiments against normalized ROG/NOx ratios for the SAPRC-99 and Carbon Bond mechanisms. The SAPRC-99 calculations show effects of varying the MONO offgasing chamber effect parameters and using a lower OH + NO2 rate constant.

Although the UCR EPA chamber has only been in operation for a relatively short time, it has already obtained usefiil information concerning the performance of current mechanism in predicting the effects of VOCs and NOx on ozone formation. As discussed here and also in our companion paper (Carter, 2004a), the SAPRC-99 mechanism predicts O3 formation reasonably well in low NOx experiments, and in ambient simulation experiments at the high ROG/NOx levels where maximum ozone formation potentials are achieved. However, die new data indicate problems with the mechanisms that were not previously realized. The SAPRC-99 mechanism consistently under predicts O3 formation in the lower ROG/NOx experiments where O3 formation is most sensitive to VOCs, and the problem is even worse for CB4. Other experiments indicate that there are problems with the formulation wifli die current aromatics photooxidation mechanisms. It is possible that the problems with the under prediction at low ROG/NOx ratios may be caused by problems with the aromatics mechanisms. Experimental and mechanism development work to investigate and hopefully resolve these problems is underway... 

---