Performance of Different Sub-Grid-Scale Surface Flux Parameterizations for Urban and Rural Areas - Meteorological and Air Quality Models for Urban Areas - page 22

Fig. 3.3 Like Fig. 3.2 but for urban area measurement sites in Berlin

Increasing the resolution in this case results in a better agreement with the mea-
surements, since the urban area is resolved more detailed. Even though the I lt =

40
case shows a worsening of performance for wind direction in urban areas com-
pared to the 16 km parameter averaging case, the pattern of improvement of per-
formance by increasing the resolution is similarly found as for the I lt = 30 case.

Further significant differences occur for temperature and dew point temperature
in urban areas for the very locally driven meteorological situation with a locality
index of 40.
Overall, the thermodynamical values are more sensitive to the parameterization
and resolution applied than the dynamical values in rural as well as in urban areas,
since they are directly affected by the sensible and latent heat fluxes. The dynamical
values show a smaller dependence on resolution for rural and urban areas for param-
eter averaging than for the flux aggregation approach. When comparing the hit rates
between the rural areas and Berlin for all parameters it is found that METRAS per-
forms better for rural than for urban areas, although METRAS captures some of the
urban phenomenon's associated with an urban heat island like the enhanced dryness
for example.
A dependence on the meteorological situation described by the locality index I lt
could not be found so far. The current results suggest that the humidity in the lower
boundary layer might play a significant role in the model performance and should
be accounted for when classifying meteorological situation.