Verification and Case Studies for Urban Effects in HIRLAM Numerical Weather Forecasting - Meteorological and Air Quality Models for Urban Areas - page 134

(a) (b)

Fig. 14.4 Diurnal variability for 00 UTC forecasts for the average (a) relative humidity at 2 m and
(b) wind velocity at 10 m for the (a) all and (b) six urban/suburban Danish stations during July
2004 as a function of the forecast length based on the DMI-HIRLAM-T15/S05 and -U01+urb
model runs vs. observations

Fig. 14.4. As seen the urbanized U01 model has good predictive skills compared
with two other models (S05 and T15) of lower resolution. In general, the air tem-
perature is well predicted during morning and evening hours. The wind velocity is
overpredicted, but it is of the higher accuracy compared with other models. Analy-
sis of the diurnal variability for the relative humidity showed comparable results for
all models during night and evening hours, but the urbanized version has a better
predictability during daytime.

14.4 Conclusions

The urbanization of DMI-HIRLAM model with modified roughness, anthropogenic
heat fluxes and building effects allowed the modelling of effects over urbanized
areas. On average, the differences between NWP control vs. urbanized runs over the
Copenhagen metropolitan area and surroundings the following was found:

• For typical wind conditions, the differences for: (1) wind at 10 m is less than

0.5 m s − 1 (with a maximum up to 1.5, at midday); (2) air temperature at 2 m is
less than 0.25 ºC (with a maximum up to 0.5 ºC, at night time); and (3) relative
humidity is a few percent (with a maximum up to 5%, at midday).

• For low wind conditions, the differences for: (1) wind at 10 m is more than 1

m/s(withamaximumupto3ms − 1 , at night time); (2) air temperature at 2 m
is more than 0.5 ºC (with a maximum up to 1.5 ºC, at night time); (3) relative
humidity is a few percent (with a maximum up to 7%, at midday).

The long-term runs with the DMI-HIRLAM-U01/I01 high resolution urbanized
models showed a slight improvement in overall NWP model performance, but this
improvement is more significant over the urban areas.