The results show great improvements for the variables (precipitation, wind and temperature) evaluated against independent observations.
Method
Observational adjusted model data (re-analysis) from the European forecasting centre in Reading, England (ECMWF) for 40 years has been downscaled. The new model mesh has a spacing of about 30km in Norwegian waters. The downscaling method is based on a global model applying a stretched grid. In this way, we improve the resolution in our waters, but this includes poorer resolution on the opposite side of the earth. This stretching saves computer power, and thus allows us to improve the resolution in the area of focus.
The large-scale solution from the old dataset has been imposed on the new one, and thus the large-scale circulation is similar in the two datasets. The small-scales have been free to evolve and among other things, the extremes are improved.
Results
Comparison with independent daily observations of precipitation, temperature and wind over a 30 years period, we find significant improvements. In the new dataset the deviation in daily precipitation is significantly reduced (from 50% to 11%), and the extremes significantly improved (e.g. from -59% to -29% for the 99.9 percentile). The temperature deviation has been reduced with about a degree and the wind also shows improvements.
Reference:
I.Barstad*, A.Sorteberg*, F.Flatøy* and M.Deque¤ (2008): Precipitation, temperature and wind in Norway: dynamical downscaling of ERA40. Publisert i Climate Dynamics (online: 19 okt 2008).
*=BCCR
¤=Meteo-France