Environmental Engineering Reference
In-Depth Information
It is recognised that, in principle, at least, a diagnostic approach to deriving pro-
files etc. in the canopy layer is needed for practical outputs such as temperatures at
street level and winds in the canopy. A certain amount of modification to raw pro-
files is used when UM output is the input for the off-line Nuclear Accident ModEl
(NAME) transport and dispersion model, but further work is needed to improve the
internal representation of the canopy.
9.2 Heterogeneity
The UM surface exchange scheme (MOSES II; Essery et al. 2001, 2003) uses a tiled
approach to surface heterogeneity. This assumes independent, 1D vertical fluxes
from different surfaces. The approach, based upon the concept of a blending height
for fluxes (Mason, 1988), is based upon the mathematical approach of matched
asymptotic expansions though more simple, heuristic arguments also exist.
We assume that, very close to a locally homogeneous surface 'patch', fluxes are
in equilibrium with the local surface. However, sufficiently far away from the sur-
face, fluxes are in equilibrium with a uniform, effective surface representing the
aggregated effect of the whole surface. This approach, formulated for a given char-
acteristic length-scale of heterogeneity, in this case a height (scale) exists below
which local equilibrium and above which aggregated equilibrium may be assumed.
This is the blending height. Since different surface patches may have different sta-
bilities, in principle the blending height is stability dependent, and should be derived
iteratively (the blending height depends on the overall stability and vice versa).
An iterative solution has been implemented. However, in practice provided the first
model layer is within the surface layer, use of the first model level works equally
well. Under extremely stable conditions, the approach probably has some difficul-
ties. Note the blending height differs from the diffusion height (the height at which
all horizontal heterogeneity has dissipated). This is roughly an order of magnitude
higher than the blending height.
In practice, the blending height approach cannot be applied to very small-scale
heterogeneities because it is to close to the surface. Mason (1988) gives a rule of
thumb of L/200 as a rough estimate of blending height, where L is the length scale
of heterogeneity. This implies patches of different surfaces are at least, 100-200 m
across. In an urban area this may stretch the concept. It may be reasonable for park-
land but not necessarily for urban gardens etc. The assumption is that the patches
are sufficiently large that the error incurred by ignoring the transition region from
one patch to the other is negligible. An essential assumption is that tiles act inde-
pendently (it does not matter how they are distributed in a grid mesh) and that local
homogeneity can be assumed so that MOST can be used to compute exchange coef-
ficients. However, see the Sect. 9.3 on the 2-tile approach.
The current scheme uses nine tiles of which one is urban. The characteristics of
the urban tile can vary from point to point in principle, though in practice, at coarse
resolution, fixed urban characteristics tend to be used. The urban tile fraction has
been determined from IGBP (AVHRR-based) land-use (Brown et al., 2003), which
Search WWH ::




Custom Search