Improved road weather forecasting by using high resolution

Improved road weather forecasting by using high resolution satellite data Claus Petersen and Bent H. Sass Danish Meteorological Institute

Background • It has been realized that prediction of cloud cover and precipitation play a key role in prediction of the road surface temperature and the road conditions. • Prediction of cloud cover requires a NWP model which can model clouds and dataassimilation of cloud cover and precipitation observations.

Viking project • Title – Development of new generation of cloud and precipitation analyses for the automatic Road Weather Model • Duration – 2003 -2005 • Goal – Improvement of the forecasts for slippery roads by developing a new prediction model

Numerical Weather Prediction (NWP) model Horizontal resolution 0. 15 x 0. 15 (degree) Vertical levels 40 Rotation of south pole Lon. =80 Lat. =0 (degree) Number of grid points 610 x 568=346480 Dynamic time step 360 (s) Physical time step 360 (s) Boundary update Every 3 rd hour Boundary age 0 -6 hours First guess age 3 or 6 hours Forecast frequency Every 4 th hour Forecast length 60 hours Data-assimilation 4 times daily+2 reassimilation cycles

Already used observations

Model domain of NWP model and network of road stations • Horizontal resolution • Vertical levels • Number of grid points • Dynamic time step • Physical time step • Boundary update 0. 15 x 0. 15 40 82 x 98=8036 72 s. 360 s. 1 hour • Boundary age • First guess age • Forecast frequency • Forecast length • Data-assimilation period • Road stations 0 -5 hours 0 -1 hour Every hour 5 -24 hours 300

Data sources

Single channels or composite

Cloud mask Cloud top temperature Precipitation intensity Cloud type

Application of cloud observations

1 hour forecast with data-assimilation of satellite data FORECAST Observed cloud mask 1 hour forecast of wind and temperature 1 hour forecast of precipitation, mslp

1 hour forecast without data-assimilation of satellite data FORECAST Observed cloud mask 1 hour forecast of wind and temperature 1 hour forecast of precipitation, mslp

6 hour forecast with data-assimilation of satellite data Observed cloud mask 6 hour forecast of wind and temperature 6 hour forecast of precipitation, mslp

6 hour forecast without data-assimilation of satellite data Observed cloud mask 6 hour forecast of wind and temperature 6 hour forecast of precipitation, mslp

21 hour forecast with data-assimilation of satellite data Observed cloud mask 21 hour forecast of wind and temperature 21 hour forecast of precipitation, mslp

21 hour forecast without data-assimilation of satellite data Observed cloud mask 21 hour forecast of wind and temperature 21 hour forecast of precipitation, mslp

Road Condition Model G: Ground heat flux S: Direct insolation D: Diffuse insolation R: Infrared radiation H: Sensible heat flux L: Latent heat flux F: Flux correction

User interface

Verification of cloud forecast • First two weeks of March 2005 • Danish SYNOP stations • Limited MSG 1 data • Verifcation for model run every hour

Best practice • A general method has been developed to assimilate cloud observations into a NWP model. • Verification and case studies indicate that prediction of cloud cover is improved for short range forecasting but that results can be further improved with more experience. • Further verification and investigation of the road surface temperature dependency of cloud cover are needed. • Satellite data will be used in the road weather model from this season • The potential use of satellite data in other road application is very large.

QUESTIONS CONTACT Claus Petersen cp@dmi. dk Danish Meteorological Institute LINKS www. dmi. dk www. eumetsat. int http: //nwcsaf. inm. es