Performance of Different Sub-Grid-Scale Surface Flux Parameterizations for Urban and Rural Areas - Meteorological and Air Quality Models for Urban Areas - page 21

The desired accuracy, A, for the air temperature and dew point is set to ± 2ºC, for

1ms − 1 , for wind direction 30 ◦ and for pressure wind speed ± ± ± 1.7 hPa.

3.3 Results and Discussion

In general, the flux aggregation method with blending height concept performs bet-
ter than the parameter averaging method in nearly all cases, for all resolutions and
parameters as shown by Figs. 3.2 and 3.3. Increasing the resolution for each of both
parameterization methods mostly pays back for the less well performing parame-
ter averaging method, since sub-grid scale surface characteristics are resolved more
explicitly by higher resolutions. This leads to more realistic averaged surface char-
acteristics. In contrast, for the flux aggregation method with blending height concept
differences in hit rates due to changes in resolution are mostly within 5%, which can
be considered as negligible. Exceptions are found for some of the urban measure-
ment sites, e.g. the wind direction in the Berlin area for an index of 30 was not
simulated well with the 16 km resolution and is improved with higher resolutions.

Fig. 3.2 Differences in hit rates for rural areas for all simulations with flux aggregation with
blending height concept (fl16 and fl08) and parameter averaging (pa08 and pa04) compared to
the theoretically worst case (parameter averaging with 16 km resolution). The results are grouped
into changes in hit rates for wind direction (dd), wind speed (ff), temperature (te) and dew point
temperature (td) and for all I lt cases. The uncertainty of calculated hit rates is ± 5%; changes in hit

rates within this range are covered by a white transparent rectangle