124
assumption of scale separation is thus weak. On the other hand, relatively high model
resolution, of the order of 10 kill is needed to realistically simulate the structure of OMCs,
so that a scale of typically a few tens of km could be taken as typical of the mesoscale
component in Eq. (67).
The vertical flux of ¢, F,p is thus given by
F,p = w¢ = w¢ +w¢ +w¢' + w¢ + w¢ +w¢' +w'¢ +w'¢ +w'¢'
(68)
where the vertical velocity w has been also separated according to Eq. (67). When
averaging over the large scale (or resolvable GCM scale) terms 3, 6, 7, and 8 disappear
based on the definition of turbulent component (see section 2) which applies in our case
to both the large scale and mesoscale circulations. The further assumption that OMCs
are not explicitly resolved by GCMs leads to neglection of the terms 2, 4 and 7, so that
we are left with
(69)
The first term in Eq. (69) is the vertical flux associated with large (resolvable) scale
motions, the second is associated with OMCs and the third is due to turbulent motions.
The problem is thus to quantify and possibly parameterize the second term in Eq. (69).
The following results are taken from the recent work of Seth and Giorgi (1995), who
performed a 3D-day simulation of OMCs with the three dimensional mesoscale model of
Giorgi et aL (1993a,b). The fields necessary to drive the model were taken from ECMWF
analyses of observations over the central-eastern U.S. for June 1990. Two experiments
were performed, a control run in which the whole domain was covered by wet forest and a
perturbed run in which a strip of dry grass of 200 km width extending in the north-south
direction was surrounded by wet forested areas. The mesoscale fluxes and circulations
were calculated as the difference between the perturbed vs control runs.
Figures lla-c,show east-west cross-sections of the difference between perturbed and
control run for u component of the wind, temperature and humidity at 3 pm of day
11 of the simulation, when strong OMCs were simulated by the modeL The zonal wind
component indicates the formation of two residual sea-breeze-like circulations at each edge
of the dry strip, counterclockwise at the western edge and clockwise at the eastern edge,
with ascending motion over the dry grass strip and descending motion over the wet areas.
These circulations are in response to the low level heating over the dry grass (see Fig.
lIb). Note that for moisture we find negative mesoscale moisture signal in the lower PEL
regions over the low evaporating grass areas, with regions of positive mesoscale humidity
signal aloft. This is caused by the ascending motions over grass which carry relatively
moist low level air into the drier upper levels. Similarly, the ascending motions produce
a negative temperature signal above the PEL over grass. This is due to the transport
assumption of scale separation is thus weak. On the other hand, relatively high model
resolution, of the order of 10 kill is needed to realistically simulate the structure of OMCs,
so that a scale of typically a few tens of km could be taken as typical of the mesoscale
component in Eq. (67).
The vertical flux of ¢, F,p is thus given by
F,p = w¢ = w¢ +w¢ +w¢' + w¢ + w¢ +w¢' +w'¢ +w'¢ +w'¢'
(68)
where the vertical velocity w has been also separated according to Eq. (67). When
averaging over the large scale (or resolvable GCM scale) terms 3, 6, 7, and 8 disappear
based on the definition of turbulent component (see section 2) which applies in our case
to both the large scale and mesoscale circulations. The further assumption that OMCs
are not explicitly resolved by GCMs leads to neglection of the terms 2, 4 and 7, so that
we are left with
(69)
The first term in Eq. (69) is the vertical flux associated with large (resolvable) scale
motions, the second is associated with OMCs and the third is due to turbulent motions.
The problem is thus to quantify and possibly parameterize the second term in Eq. (69).
The following results are taken from the recent work of Seth and Giorgi (1995), who
performed a 3D-day simulation of OMCs with the three dimensional mesoscale model of
Giorgi et aL (1993a,b). The fields necessary to drive the model were taken from ECMWF
analyses of observations over the central-eastern U.S. for June 1990. Two experiments
were performed, a control run in which the whole domain was covered by wet forest and a
perturbed run in which a strip of dry grass of 200 km width extending in the north-south
direction was surrounded by wet forested areas. The mesoscale fluxes and circulations
were calculated as the difference between the perturbed vs control runs.
Figures lla-c,show east-west cross-sections of the difference between perturbed and
control run for u component of the wind, temperature and humidity at 3 pm of day
11 of the simulation, when strong OMCs were simulated by the modeL The zonal wind
component indicates the formation of two residual sea-breeze-like circulations at each edge
of the dry strip, counterclockwise at the western edge and clockwise at the eastern edge,
with ascending motion over the dry grass strip and descending motion over the wet areas.
These circulations are in response to the low level heating over the dry grass (see Fig.
lIb). Note that for moisture we find negative mesoscale moisture signal in the lower PEL
regions over the low evaporating grass areas, with regions of positive mesoscale humidity
signal aloft. This is caused by the ascending motions over grass which carry relatively
moist low level air into the drier upper levels. Similarly, the ascending motions produce
a negative temperature signal above the PEL over grass. This is due to the transport