Federal Department of Home Affairs FDHA Federal Office of Meteorology and Climatology MeteoSwiss The Latent Heat Nudging Scheme of COSMO EWGLAM/SRNWP Meeting, , Zürich, Switzerland Daniel Leuenberger, MeteoSwiss S. Klink, K. Stephan and Ch. Schraff, DWD

2 The LHN scheme of COSMO Content Introduction The Latent Heat Nudging scheme Results Summary and Outlook

3 The LHN scheme of COSMO Introduction QPF poorest performance area of NWP, especially in summer (convection) Rainfall assimilation to mitigate spinup effect Radar observations for km-scale NWP models Some requirements placed upon operational km-scale radar data assimilation methods computational efficiency timely assimilation of high-frequency observations ability to cope with complex microphysics

4 The LHN scheme of COSMO The Latent Heat Nudging Scheme Originates from Jones and Macpherson (1997) Method: modification of diabatic (latent) heating rates in order to simulate the observed precipitation rates Assumption: vertical integrated latent heat release in one column  precipitation rate at the ground Realisation: add nudging increment to temperature tendency

5 The LHN scheme of COSMO Model fair scaling of local profile The Latent Heat Nudging Scheme Model much too wet limited scaling of local profile Model much too dry limited scaling of nearby profile (gridpoint search) Add humidity increment in order to maintain relative humidity Model Radar Rainrate Model Diabatic Heating z

6 The LHN scheme of COSMO Advection of Precipitation Violates LHN assumption Introduces spatial and temporal shift (LH → RR sfc ) Temporal shift use vertically integrated preci- pitation flux RR ref as reference Spatial shift Updraft region : large LH, small RR obs reduced limit of scaling factor, limited LHN increments revised grid point search Downdraft region: small LH, large RR obs scale only positive LH parts of profile + - from R. A. Houze, Cloud Dynamics Book

7 The LHN scheme of COSMO radar LMK ass without LHN LMK ass with LHN Strong low-level advection Courtesy of DWD

8 The LHN scheme of COSMO LHN with correction radar Strong low-level advection weight of LH increments decreased linearly from 1 to zero when low- level wind speed increases from 20 to 30 m/s (low-level wind speed v ll := ½ v ¼ v ¼ v 700hPa ) LHN without correction Courtesy of DWD

9 The LHN scheme of COSMO Results Introduction The Latent Heat Nudging scheme Results Summary and Outlook

10 The LHN scheme of COSMO ETS FBI threshold = 2.0 mm/h UTC runs threshold = 0.1 mm/h same ETS despite smaller FBI +0h +4h higher ETS undershooting (w. resp. to no-LHN) LHN no LHN Courtesy of DWD Statistics from 45 Summer Days (2006)

11 The LHN scheme of COSMO ETS FBI threshold = 0.5 mm/h 12 UTC runs +0h +4h 00 UTC runs LHN no LHN Courtesy of DWD Statistics from 45 Summer Days (2006)

12 The LHN scheme of COSMO ETS FBI threshold = 2.0 mm/h UTC runs threshold = 0.1 mm/h higher ETS despite smaller FBI +0h +4h LMK: precip areas too small all models: strong precip underestimated Statistics from 45 Summer Days (2006) LHN no LHN LME (7km) Courtesy of DWD

13 The LHN scheme of COSMO Sensitivity to Environment Assimilation scheme is not whole story! Convective environment crucial for the success of short-range QPF Quantify effect with an ensemble LHN simulation  x=2.2km 10 members use COSMO-LEPS (7km) as driving ensemble deterministic run started from 7km analysis same radar rainfall in all members 10h assimilation starting at 06 UTC, 8h free forecast

14 The LHN scheme of COSMO Mean Area Precipitation (300x200km) RAD det t = 0

15 The LHN scheme of COSMO Radar With LHN Without LHN (dashed: determininistic) Benefit of LHN t = 0

16 The LHN scheme of COSMO Outlook Introduction The Latent Heat Nudging scheme Results Summary and Outlook

17 The LHN scheme of COSMO Summary Large step forward in LHN for km-scale LM Still open problems and questions LHN has (too) much to correct for Convective environment important Use LHN in concert with other observations humidity (e.g. from GPS, radiometer, satellites) radial winds (from radar)

18 The LHN scheme of COSMO Outlook LHN running preop. in LMK (2.8km) at DWD since Operational in beginning of 2007 Monitor LHN performance and improve algorithm Improve quality of radar observations Implement LHN parallel test suite with 2.2km LM at MeteoSwiss Improve radar quality control mechanism in the LHN algorithm

19 The LHN scheme of COSMO Questions? Thank you for your attention !

20 The LHN scheme of COSMO Example: RR obs = 3.1 mm/h RR ref = 0.1 mm/h RR search = 2.1 mm/h But, how much should be the scaling factor  ? LH mod z *  z Tinc lhn Revised grid point search

21 The LHN scheme of COSMO Example: RR obs = 3.1 mm/h RR ref = 0.4 mm/h RR search = 2.1 mm/h LH serach z *  z Tinc lhn but, how large should be the scaling factor  ? (want to add ‘0.7 * RR ref = 0.28 mm/h) (revised version of RR max in the old version, RR max and  could become very large) conditions: RR search close to RR obs LH ref small enough LH search large enough Revised grid point search

22 The LHN scheme of COSMO radar weaker forcing stronger forcing radar weaker stronger UTC 4 UTC 6 UTC mm/h hourly precipitation (4, 6 UTC ): old stronger forcing produces strong gravity waves Revised grid point search Courtesy of DWD

23 The LHN scheme of COSMO det hPa Flow