In this paper we present the results of two wind-tunnel simulations of
dispersion from upwind point sources through a large group of obstacl
es, and compare these with an associated field study (Davidson et al.,
1995, Atmospheric Environment 29, 3245-3256), Detailed flow-field and
plume concentration data were obtained from simulations at scales of
1:20 and 1:200. With these data we are able to provide experimental co
nfirmation of many of the ideas developed during the field study and t
o confirm the experimental results obtained in the field. In doing so,
we show that the upstream flow-field parameters are the most effectiv
e means of scaling the three data sets to achieve broad quantitative a
greement. Measurements and flow visualisation of the flow-field confir
m that there are a number of mechanisms influencing the behaviour of a
plume as it passes through an obstacle array: in particular the diver
gence and convergence of streamlines and changes to the structure of t
he turbulence within the array, However, although the turbulence withi
n the array is shown to be of greater strength and smaller scale than
at corresponding locations outside the obstacle array, it is found tha
t there is little change in the transverse diffusivity (and therefore
in the lateral plume width). The concentration data confirm that the d
ivergence of streamlines near the upstream end of the obstacle array h
as a significant effect on the vertical width of a plume (sigma(2)). C
hanges to the structure of the turbulence appear to have little effect
, however, since the transverse diffusivities within the obstacle arra
y are unchanged. Thus, the mean lateral spread and decay of mean conce
ntration of the plume with downstream distance resemble that of a cont
rol plume; that is; a plume released under identical conditions where
the obstacle array is not present. We also confirm that the mean struc
ture of a plume has a Gaussian form as it passes through an array of o
bstacles. By contrast, concentration measurements with a high-frequenc
y-response detector confirm that the small-scale, high-strength turbul
ence rapidly mixes the plume internally, dramatically reducing the str
ength of concentration fluctuations within the plume. Since the wind t
unnel is shown to be an effective means of modelling this type of fiel
d situation, with the appropriate scaling, these studies were extended
to consider the effects on plume behaviour of changes in source posit
ion, array configuration and array height. Copyright (C) 1996 Elsevier
Science Ltd