Environmental Engineering Reference
In-Depth Information
3. Results
The three different methods were evaluated by simulating the July 29-August 2,
2000 ozone episode in Central California. The uncertainties in wind speed and
direction to be used in Method 3 for switching from vertical winds adjustment to
concentration renormalization were determined by comparing the winds generated
by the meteorological model with profiler data at 24 locations in Central California.
The more accurate vertical advection scheme in Method 2 resulted in better
ozone performance than the upwind scheme in Method 1. The normalized error in
hourly-average ozone concentrations was reduced at 71 out of 122 sites, with a
reduction of the average error from 56.63% to 49.81% (
Table 1).
Method 3
effectively limited the trajectory deviations seen both in Method 1 and Method 2.
It introduced some mass conservation errors but at a lesser degree than the pure
renormalization approach. The normalized error was reduced at 72 out of 122 sites
when Method 1 was replaced with Method 3. The average normalized error
decreased from 56.63% to 55.16% (
Table 1)
.
In Method 2, the adjusted winds are solved iteratively for each layer using the
secant method, starting from the bottom and moving to the top. Another option is
the iterative solution of the wind adjustments for all the layers at once, as a
system. This may result in better convergence characteristics since in our solution
technique, errors may be accumulating as we go from the bottom to the top. This
hypothesis was tested by using the Newton-Raphson method for solving the
system of nonlinear equations. The simultaneous solution did not improve
convergence and resulted in larger wind adjustments.
Table 1.
Normalized errors (%) for ozone at CCOS sites
Method-1
Method-2
Method-3
Average
56.63
49.81
55.16
Max
228.75
184.95
221.82
Min
10.97
11.17
10.75
St. dev.
39.37
31.38
37.34
4. Conclusion
The adjustment of vertical winds helps achieve mass conservation in AQMs. It
yields better ozone performance when applied with more accurate vertical advection
schemes such as Bott's. However, since those schemes are non-linear, an iterative
solution is needed. This introduces small mass conservation errors. Recall that the
mass was strictly conserved with the first-order upwind scheme. Both methods
introduce trajectory deviations compared to wind fields generated by meteoro-
logical models. Therefore we developed a third method which mixes the use of
