- surface observations in Switzerland (hourly) and Europe (3-hourly) for 2 m-temperature, dewpoint, pressure (station pressure and reduced to sea level), wind (speed and direction), cloud cover, precipitation, and wind gusts.
- upper air data from radiosonde ascents for temperature, wind, geopotential and humidity
- radar data for precipitation
- windprofiler data and flux measurements in Payerne for different near-surface and boundary layer parameters (short and longwave surface radiation, albedo, sensible and latent heat flux).
Main results
Percentage of correct forecasts of 2m-temperature within 2, 5, 8 degrees C in Spring 2005 over Switzerland for gridpoints < 800 m. COSMO direct model output (black line), COSMO output adjusted with Kalman filter (red line).
large.gif, 27 KBSurface observations:
2m-temperature: With the Kalman-filter correction, 75% of all hourly forecasts show an error of < 2 degrees C for the first day of the forecast dropping to a high 65% for day 3 (left, top). During the evening and at nighttime, there is a cold bias (especially in winter). The amplitude of the diurnal cycle is too large and the maximum about 1.5 hours too early (left, bottom).
2m-dewpoint: The diurnal cycle ist not well captured, but the Kalman filter corrects the systematic biases.
10m-wind: The wind speed is overestimated (especially during nighttime). For mountain gridpoints there is a strong underestimation (due to the same parameterization of the boundary layer over flat and complex topography).
precipitation: The low amounts (0.1 mm in 6 hours) are overestimated. In summer the mean diurnal cycle of convective precipitation has its maximum about 4 hours too early.
Verification of the 2m-temperature diurnal cycle in Spring 2005 over Switzerland for gridpoints < 800 m. Observations (ANETZ, thin black line), COSMO direct model output (dashed black line), COSMO output adjusted with Kalman filter (dotted red line).
large.gif, 27 KBVertical profiles:
There is a cold bias from the surface up to 750 hPa (mainly in summer). The mean error in wind speed is small, with a positive bias in the boundary layer (at least during the cold season) and a small negative bias above 800 hPa.




