Evolution of Urban Surface Exchange in the UK Met Office’s Unified Model - Meteorological and Air Quality Models for Urban Areas - page 71

Best, 2005) shows the improvement in nocturnal temperature forecast for a typical
radiation night. A significant improvement was produced even in the operational
forecast model (12 km) for urban areas, though errors in the diurnal cycle response
were still evident. The approach has some similarities with the force-restore method
of treating building thermal properties.
The thermal canopy was first implemented operationally within the “Site Spe-
cific Forecast Model”, a 1D form of the UM driven from the 12 km model (Clark,
1998). Figure 9.3 shows results from a 3 month autumn/winter trial comparing the
forecasts for two sites about 20 km apart, one largely rural, the other largely urban.
For reference, the then-operational 12 km model is also compared - this contained
essentially no urbanization. The variability of the temperature difference is well
reproduced, and the bias roughly halved. Some impact is also evident on relative
humidity, but clearly not enough.
Figure 9.4 shows the forecast vs observed urban heat island. The intensity of
the urban heat island is somewhat under-estimated using this approach (typically a
factor of two) - this was also observed after implementation of the scheme in the
12 km forecast model. Some improvement has been achieved through the addition
of an anthropogenic heat source to the recently-implemented 4 km forecast model.
Fig. 9.4 Average forecast vs.
observed temperature
difference between Heathrow
(largely urban) and Beaufort
Park (largely rural).
Mesoscale model ( solid ),
SSFM ( dashed ) over all 06
UTC forecasts (T+7-T+24,
i.e. 13 UTC-06UTC) from
Nov-Dec 1997 and Jan 1998.
Error bars denote two
standard deviations from the
mean