COSMO_2005 DWD 9 Sep 2005Page 1 (9) COSMO General Meeting Zürich, September 2005 Erdmann Heise German Weather Service Report on Workpackage Improve diagnosis of convective and turbulent gusts

COSMO_2005 DWD 9 Sep 2005Page 2 (9) WP Improve diagnosis of convective and turbulent gusts In the diagnostic determination of gusts three changes have been made: The effect of water loading is included in the buoyancy determination of downdrafts The tuning parameter responsible for distributing the downward kinetic energy of the gusts to all horizontal directions is increased from to. If the convective precipitation rate is below mm/h, convective gusts are suppressed. a) Convective gusts

COSMO_2005 DWD 9 Sep 2005Page 3 (9) WP Improve diagnosis of convective and turbulent gusts a) Convective gusts The ‘Berlin-case’ of 10 July 2002 showed only moderate gusts with the operational version of the gust diagnosis (left). In the new version (right) gusts are much higher, but still not as vigorous as in the observations.

COSMO_2005 DWD 9 Sep 2005Page 4 (9) WP Improve diagnosis of convective and turbulent gusts a) Convective gusts On 12 August 2004 a conver- gence line with heavy con- vection crossed Gernany. It caused gusts up to 35 m/s in south-western Germany. Other regions were free of gusts. The operational version significantly underestimated the gusts, but overestimated the area of mod- erate gusts. The new version improves both the maximum value and the area covered by moderate gusts, but simulates erroneous gusts in north-eastern Germany. (The figures show convective and turbulent gusts.)

COSMO_2005 DWD 9 Sep 2005Page 5 (9) WP Improve diagnosis of convective and turbulent gusts An error was noted in a first experiment using the new version of gust diagnosis. Therefore a new experiment had to be initiated. It will simulate the period 01 July 2004 to 31 August Up to now no verification results are available. a) Convective gusts

COSMO_2005 DWD 9 Sep 2005Page 6 (9) WP Improve diagnosis of convective and turbulent gusts b) Turbulent gusts Following Brasseur (MWR, 129, 2001, 5-25) the basic relation for a determination of turbulent gusts is given by Here E(z) is turbulent kinetic energy, Δθ v is the difference of the virtual potential temperature between the environment and a dry-adiabatically rising/descending parcel, and z p is the largest height for which the inequation holds. Therefore, z p is the largest height from which a parcel can move downward to the earth’s surface. The gust at the surface is then given by the mean wind in z p. In order to avoid too vigorous gusts in the case of deep mixed layers, a maximum value of 2000 m is prescribed for z p.

COSMO_2005 DWD 9 Sep 2005Page 7 (9) WP Improve diagnosis of convective and turbulent gusts b) Turbulent gusts On May 13, 2004, UTC, gusts were reported in only few areas. The most pronounced reports are in the Rhone valley and in the Golfe du Lion. In this region the inclusion of the surface turbulent kinetic energy proved to be necessary in order to simulate high enough gusts with the Brasseur method. In Brittany and at the coast of the Baltic Sea gusts up to 12 m/s were reported. The values of the Brasseur method fit better to the obser- vations than the operational method. The lower gusts values in the latter version in most of the LM-area also fit better to the observations. No observations were avail- able in Spain, Italy, Slovenia and Croatia.

COSMO_2005 DWD 9 Sep 2005Page 8 (9) WP Improve diagnosis of convective and turbulent gusts b) Turbulent gusts Average verification results of a parallel experiment 01 October - 31 December 2004: Statistical measures for the verification using all stations. Green numbers: Experiment more than 5 % better than Reference Red numbers: Experiment more than 5 % worse than Reference

COSMO_2005 DWD 9 Sep 2005Page 9 (9) WP Improve diagnosis of convective and turbulent gusts b) Turbulent gusts Results of a parallel experiment 01 October - 31 December 2004: a) positive The frequency of gusts is reduced The probability of detection for severe gusts is increased (except for October) The false alarm rate for all gusts is reduced b) negative The probability of detection for all gusts is reduced considerably The equitable threat score for all gusts is reduced The false alarm rate for severe gusts is increased The verification results differ considerably for different station heights and for different times of the day. This might lead to some reformulation. But then a new experiment has to be realised. An Interim Report on the results is available.