Figure 2. The median values and 75th percentile of the total oxidant production rate derived from Jacob (1999) (dotted line and bars) and from the OBM (dashed line and bars) are plotted as a function of the NO x concentration. Also shown is the ozone production rate calculated by the ID model (solid line). (b) Same as (a) but plotte against ANOx.
as 30 ppbv/h. Because the data for OBM are from 12 different sites in the KaoPing area that include upwind, urban center and downwind sites, the scattered distribution of P(Ox) vs NOx is expected. P(Ox) can also be calculated from the Jacob equation. Values of P(Ox) from this method agree very well with those of the OBM approach. There are some outliers estimated by the OBM with low values less than 2 ppbv/h. As discussed earlier, they are likely the result of fresh emissions.
Also shown in Fig. 2 is the 1D-model-simulated P(Ox) versus NOx. The median values of P(Ox) obtained by the OBM and the Jacob equation are about 60% and 40% smaller than model-simulated values, respectively.
We have compared our values of P(Ox) from the OBM with those derived by Frost et al. (1998). Unfortunately, our values are at much higher concentrations of NOx than those of Frost et al. (1998), as theirs are aircraft measurements while ours are from surface stations in urban and suburban areas in the morning between 910 a.m. and 11 a.m.-12 noon. It is well known that the top of the mixed layer usually grows in the morning and does not reach the top of the boundary layer of the previous day 1.5 km) until around noon. As a result, surface stations tend to be significantly influenced by fresh emissions in the morning. This is evident from the observed decreasing concentrations of CO, NOx and NMHCs in the morning at nearly all stations. Part of the decrease of NOx and NMHCs is due to photochemical consumption; the decreases of CO is due to the dilution of fresh emissions when the mixed layer grows in thickness.
P(Ox) simulated by our 1D model is very similar to that of Frost et al. (1998) below a few ppbv of NOx. Above a few ppbv of NOx, P(Ox) in the model of Frost et al. (1998) levels off and starts to decrease, while P(Ox) in our model continues to increase with NOx (Fig. 2). One reason for the difference is that their NMHC mix has significantly higher isoprene which does not increase with NOx. Another reason is probably that we have a more reactive mix of NMHCs. In this context, we note that Kleinman et al. (2002a) compared P(Ox) vs NOx in five urban areas in the US and found that the peak P(Ox) occurred at a NOx concentration of about a few to 20 ppbv for those cities. At highly polluted sites, P(Ox) was larger than 30 ppbv/h and decreasing values of P(Ox) with increasing NOx were not observed, consistent with results of our 1D model and the OBM.
From the discussion on the ozone production efficiency, it is clear that P(Ox) should be related to the consumed NOx rather than NOx itself. In Fig. 2, P(Ox) is plotted against ANOx, the consumed NOx between 9-10 a.m. and 11 a.m.-12 noon. As expected, the correlation of P(Ox) with ANOx improves significantly because ANOx is a better measure of the photochemical activity. Values of P(Ox) from the OBM agree well with those from the 1D model at ANOx greater than 3 ppbv. Below 3 ppbv of ANOx, values of P(Ox) from the OBM are about 50% lower. Considering the uncertainties involved in all three methods, the discrepancies are surprisingly small. In fact, the high level of agreement between the 1D model and Jacob's approach is likely to be fortuitous, as the latter is the gross photochemical production of Ox, while the former is the net Ox photochemical production. The OBM values are expected to be less than those of the other two approaches, because the OBM also includes the loss of Ox due to processes such as heterogeneous loss, dispersion and surface deposition of O3 and NO2. The effect of these processes is expected to be more significant at low production rates of Ox, consistent with the feature in Fig. 2 that P(Ox) from the OBM starts to become lower than for the other two approaches at about 5 ppbv of ANOx.
The agreement of P(Ox) from Jacob (1999) with the 1D model is even better. Given the independent nature of the three methods in evaluating P(Ox), the significant level of agreement among them provides strong support for the validity of all three methods.
A widely used approach to formulating the ozone control strategy is the so-called empirical kinetic modeling approach (EKMA) or ozone isopleth diagrams (e.g. Milford et al., 1989; Gery and Crouse, 1990; Altshuler et al., 1995; Seinfeld and Pandis, 1998). The EKMA diagram usually depicts daily peak O3 mixing ratios as a function of initial concentrations or emission rates of O3 precursors. An example is given in Fig. 3, which shows peak O3 concentrations calculated for various reduction scenarios in NOx and NMHC emissions in southern Taiwan (Chang, G.-H., private communication, 2004). One can easily see that reduction of NMHC emission while holding the NOx emission constant (i.e. sliding to the left along the top horizontal edge) is quite effective in lowering the ozone level. On the other hand, reducing NOx emission while holding NMHC emission constant (i.e. sliding down along the right vertical edge) will raise the ozone level.
Was this article helpful?