Ozone carried over or transported aloft is only important if it
affects surface concentrations. In rural areas with low emissions,
the aloft ozone can mix to the ground when the mixing layer rises
in the morning and cause surface concentrations to increase. In
urban areas or urban plumes, the aloft contribution is more complicated
since the resultant surface ozone concentrations depend on how
the ozone and precursors aloft react with the fresh surface emissions.
The potential contributions in both urban and rural areas are
explored below.
Ozone aloft can contribute to surface concentrations when the
mixing layer rises during the day and the ozone aloft is mixed
to the surface. To confirm their understanding of this effect,
Zhang et al. (1996) modeled the effect on surface concentration
for the morning of July 14 at Reston, VA. They used the rawinsondes
at Sterling (Dulles) to estimate the increase in mixing height
during the day and the morning aircraft spiral data at Manassas
to characterize the ozone carried over aloft. They estimated the
surface ozone concentrations due solely to the mixing down of
the aloft ozone and compared the results to the surface concentrations
at Reston, VA. The model predictions and measurements are shown
on Figure 9-1. The aloft concentrations
were able to explain the ozone seen at the surface through roughly
1200 EST. After that time, some new ozone formation would have
been required to match the observed concentrations.
The aloft ozone was shown to contribute about 70 ppb to the surface
concentrations by midday. This amount is approximately equal to
the average ozone concentration from the surface through about
1200 m msl in the morning (0730 EST) spiral. This average was
made up of about 40 ppb in the surface layer through 300 m msl
and about 80 ppb from 300 m through 1200 m msl.
An analysis was performed to estimate the impact of aloft ozone
on surface ozone concentrations. A photochemical box model with
two layers (surface and aloft) was run for different cases based
on surface initial conditions similar to those measured on August
1, 1995 at the Bronx, NY PAMS site. The box model was run for
0500-1700 EST. The model was not intended to simulate actual conditions,
but was designed to represent the effects of typical urban emissions
in the morning traveling over a rural inland area in the afternoon.
The mixing height was assumed to increase during the day at the
rate measured on August 1 by the New Brunswick, NJ (Rutgers)
radar profiler. The model inputs and the chemical mechanisms used
were varied to assess the model response and sensitivity to a
variety of conditions.
Two different chemical mechanisms, Carbon Bond IV (CB-IV) and
SAPRC93, were used in the analysis. CB-IV is the mechanism used
in the EPA UAM-IV model, and SAPRC93 is the latest version of
the SAPRC family of chemical mechanisms.
The surface layer initial conditions used were the observed concentrations
at 0500 EST at Bronx (see Tables 9-1
and 9-2). The hydrocarbon
data were first converted to CB-IV species. Carbonyl data were
suspect at Bronx on August 1, hence FORM (Formaldehyde) was assumed
to be 3 percent of NMOC and ALD2 (Aldehydes) was assumed to be
2 percent of NMOC. Concentrations of CB-IV species were converted
to SAPRC93 species using a scheme provided by Lurmann et al. (1991).
Aloft initial conditions used were obtained from aircraft data
for New Haven, CT and Brookhaven, NY (Tables 9-1 and 9-2). Again,
the hydrocarbon data were first converted to CB-IV species and
then to SAPRC93 species. The aircraft measured NO and NOy
rather than NOx. The average NOy concentration
observed was 8 ppb, which was split into NO2, PAN,
and HNO3 (Table 9-1). The aloft initial ozone was varied
in increments of 10 ppb from 40 ppb to 90 ppb for different simulations.
Mixing height and temperature data measured at Rutgers, New Jersey
were used (see Table 9-3).
The mixing height was assumed to stay at the 1400 EST value after
1400 EST even though it actually dropped below that value later
in the day.
A simple emissions profile was used to input the emissions to
the model (Table 9-4). Total emissions used were 1 ton/day of
NOx, 2 tons/day of ROG, 10 tons/day of CO and 0.5 tons/day
of Isoprene, assuming a 5x5 km2 horizontal grid box.
NOx emissions were assumed to be 90 percent NO and
10 percent NO2. Most of the NOx, CO, and
ROG emissions were input in the morning hours, whereas Isoprene
emissions were assumed to increase linearly from 0500-1400 EST
and then decrease linearly from 1400-2100 EST. Different sensitivity
simulations were performed where surface layer initial conditions
and emissions (both ROG and NOx) were changed by a
factor of 2 or 0.5 in different combinations. The motivation behind
using different sensitivity cases was to see how the impact of
aloft ozone would vary with different surface conditions.
Tables 9-5 and 9-6 show
the contribution of different terms to surface level maximum ozone.
The maximum ozone was predicted to occur at 1700 EST for all cases.
Contributions of different terms were calculated using the initial
and final mass of ozone in the surface and aloft layers and the
evolution of the mixing height. For each sensitivity simulation,
the results for initial aloft ozone concentrations of 40 ppb and
90 ppb are shown in the tables. The 40 ppb minimum value was selected
to represent clean air.
The results demonstrate the nonlinear nature of the chemistry.
Increasing the initial ozone concentration aloft by 50 ppb does
not necessarily result in 50 ppb higher ozone at the surface.
The effect of aloft ozone on surface ozone is different for different
conditions at the surface. The maximum impact of aloft ozone occurred
when the initial surface NOx and emissions of NOx
were doubled. For that case, a 50 ppb increase in the initial
aloft ozone above clean air resulted in an increase in surface
level maximum ozone of 47 ppb for the CB-IV mechanism, and 54
ppb for the SAPRC93 mechanism. When 90 ppb aloft ozone was used,
it contributed between 44 and 66 percent towards the maximum surface
level ozone for the CB-IV mechanism and between 32 and 49 percent
for the SAPRC93 mechanism. The difference in the percent contribution
using the two mechanisms is mainly because the SAPRC93 mechanism
is much more reactive than the CB-IV mechanism resulting in relatively
more production of ozone in the surface layer.
Figures 9-2 and 9-3
show the maximum surface layer ozone concentrations for different
initial aloft ozone concentrations for different sensitivity simulations
for the CB-IV and SAPRC mechanisms, respectively. For the base
case, the maximum ozone predicted by the SAPRC93 mechanism is
about 50 ppb higher than that predicted by the CB-IV mechanism.
For example, at 90 ppb aloft ozone, SAPRC93 predicts 200 ppb ozone
and CB-IV predicts 149 ppb ozone at the surface layer. The
highest surface level ozone is predicted when both ROG and NOx
were doubled (CB-IV predicting 184 ppb and SAPRC93 predicting
251 ppb at 90 ppb aloft ozone).
Figures 9-4 and 9-5
show the increment in maximum ozone at the surface for different
initial aloft ozone concentrations for the CB-IV and SAPRC mechanisms,
respectively. Increments were calculated from the ozone levels
predicted for 40 ppb aloft ozone. Although the CB-IV and SAPRC93
mechanisms predict significantly different ozone concentrations
(Figures 9-2 and 9-3),
the increments in surface ozone predicted by the two mechanisms
are more similar. Both mechanisms predict that the highest effect
of aloft ozone occurs under double NOx conditions.
Using the CB-IV mechanism, a 50 ppb increment in ozone aloft results
in a 47 ppb increase in maximum surface ozone concentration for
the double NOx conditions. Under all other conditions
the increase in maximum surface ozone is between 17 and 22 ppb
when the aloft ozone is increased by 50 ppb. Using the SAPRC93
mechanism, a 50 ppb increment in ozone aloft results in a 54 ppb
increase in maximum surface ozone concentration for the double
NOx conditions. Under all other conditions the increase
in maximum surface ozone is between 20 and 40 ppb when the aloft
ozone is increased by 50 ppb. The SAPRC93 mechanism predicts a
broader range of impact of aloft ozone than does the CB-IV mechanism
for conditions other than the double NOx conditions.
For both mechanisms the least impact of aloft ozone on surface
ozone is predicted to occur under double ROG conditions.
Figures 9-2 through 9-5 show that the increase in surface ozone concentration is roughly a linear function of the initial aloft ozone concentration for a given set of surface conditions. Figures 9-4 and 9-5show that the range of contribution to surface concentrations is roughly 34 to 108 percent of the aloft increment above clean air, depending on the initial conditions, emissions, and reaction mechanism used. Most of the results were bunched in the lower half of this range.
From Tables 9-5
and 9-6 an estimate
of the contribution to maximum surface ozone of clean air, excess
ozone transported aloft, and ozone formed from same-day surface
emissions can be derived for each model run. For the range of
conditions examined here, with a carryover concentration of 90
ppb aloft, the ranges of contribution to the maximum ozone were
roughly 14-30 percent for clean air, 23-39 percent for excess
ozone transported aloft, and 33-63 percent for same-day surface-based
ozone. For the same runs, surface-based ozone plus clean air contributed
roughly 61-77 percent of the maximum ozone and the excess ozone
transported aloft contributed 23-39 percent.
These results should not be used in a quantitative fashion. They
do not represent real conditions, and some assumptions may be
unrealistic. For example, the aloft ozone was varied without varying
the aloft NOx or VOC. To better understand the results,
the processes and assumptions of the model should be further examined
to see whether they are realistic for the ranges of inputs used,
and the potential errors of the approach should be assessed. We
believe, however, that the results are quite instructive in a
qualitative manner.
The impact of aloft ozone on surface ozone concentrations appears
to be positive but is nonlinear. It is likely that the impact
on the surface concentrations, when mixed into reactive urban
plumes, will be less that the aloft ozone increment above clean
air. That is, an increase in the aloft ozone concentrations does
not necessarily result in the same increase in the surface level
ozone. The impact of aloft ozone on surface ozone varies depending
on the conditions at the surface. The highest effect of aloft
ozone on surface ozone was predicted for high NOx conditions
at the surface, and the lowest effect was predicted for high surface
ROG conditions. NOx rich conditions appear to result
in a higher percentage impact of aloft ozone on the surface concentrations.
However, this does not necessarily result in higher surface ozone
concentrations because the surface ozone is also affected by the
production of ozone in the surface layer. For example, doubling
the NOx emissions and initial conditions reduced the
surface level ozone compared to the ozone predicted by the base
case, using SAPRC93 mechanism; but a higher percentage of the
aloft concentrations were reflected at the surface.
The SAPRC93 mechanism is more reactive than the CB-IV mechanism.
The SAPRC93 mechanism predicted maximum surface ozone concentrations
which were higher by 40-60 ppb for different conditions, for 90
ppb aloft ozone.
If these results are correct, a reduction of the regional background
ozone would result in a positive, but smaller reduction in the
maximum ozone seen in urban plumes.
| Submit your comments, feedback, questions, and ideas pertaining this page. Your input will be automatically added to the existing annotations. In order to add a new comment, you must be registered with the OTAG/AQA Peoples Page. |
