Page 373 - Fundamentals of Air Pollution 3E

328 20. Air Pollution Modeling and Prediction

Several point source algorithms, HPDM (17, 18) and TUPOS (19, 20),
incorporate the use of fluctuation statistics (the standard deviations of
horizontal and vertical wind directions) and non-Gaussian algorithms for
strongly convective conditions. Because during strong convective condi-
tions, thermals with updrafts occupy about 30-35% of the area and slower
descending downdrafts occupy 65-70% of the area, the resulting distribu-
tion of vertical motions are not Gaussian but have a smaller number of
upward motions, but with higher velocity, and a larger number of down-
ward motions, but with lower velocity. These skewed vertical motion distri-
butions then cause non-Gaussian vertical distributions of pollutant concen-
trations,

C. Multiple Sources and Area Sources
The problem, already noted, of not having the appropriate plume trans-
port direction takes on added importance when one is trying to determine
the effects of two or more sources some distance apart, since an accurate
estimate of plume transport direction is necessary to identify critical periods
when plumes are superimposed, increasing concentrations.
In estimating concentrations from area sources, it is important to know
whether there is one source surrounded by areas of no emissions or whether
the source is just one element in an area of continuous but varying emis-
sions.
To get an accurate estimate of the concentrations at all receptor positions
from an isolated area source, an integration should be done over both the
alongwind and crosswind dimensions of the source. This double integration
is accomplished in the PAL model (21) by approximating the area source
using a number of finite crosswind line sources. The concentration due to
the area source is determined using the calculated concentration from each
line source and integrating numerically in the alongwind direction.
If the receptor is within an area source, or if emission rates do not vary
markedly from one area source to another over most of the simulation area,
the narrow-plume hypothesis can be used to consider only the variation
in emission rates from each area source in the alongwind direction. Calcula-
tions are made as if from a series of infinite crosswind line sources whose
emission rate is assigned from the area source emission rate directly upwind
of the receptor at the distance of the line source. The ATDL model (22)
accomplishes this for ground-level area sources. The RAM model (8) does
this for ground-level or elevated area sources.
Rather than examine the variation of emissions with distance upwind
from the receptor as already described, one can simplify further by using
the area emission rate of only the emission square in which the receptor
resides (23). The concentration c is then given by