480 likes | 626 Views
M. Baldauf , A. Seifert, J. Förstner, T. Reinhardt, C.-J. Lenz, P. Prohl, K. Stephan, S. Klink, C. Schraff. LMK (LM Kürzestfrist) - Development of a very short range forecast model COSMO General Meeting, Zürich 21.09.2005. Methods I - LMK-Configuration. grid length: dx = 2.8 km
E N D
M. Baldauf, A. Seifert, J. Förstner, T. Reinhardt, C.-J. Lenz, P. Prohl, K. Stephan, S. Klink, C. Schraff LMK (LM Kürzestfrist) - Development of a very short range forecast modelCOSMO General Meeting, Zürich21.09.2005
Methods I - LMK-Configuration • grid length: dx = 2.8 km • direct simulation of the coarser parts of deep convection (!?) • interactions with fine scale topography • about 50 vertical layers, lowest layer in 22 m (new: 10 m) above ground • center of the domain 10° E, 50° N • 421 x 461 grid points • boundary values from LM (LME)(x = 7 km)
LMK-forecast runs (18h-forecasts, started every 3h) -> LAF-Ensemble LMK-data assimilation cycle • continuous assimilation -> nudging • short cut-off (<1h) • use all available data, especially radar reflectivites possible alterations in this procedure by: • additional soil moisture analysis (SMA) • split in a ‘forecast run analysis’ and a pure data assimilation cycle • GME, LME and LMK cannot be run at the same time on the computer • bigger time window needed for maintenance of the computer Methods II : LAF-Ensemble (Rapid update cycle)
M7: Radar M8: Latent Heat Nudging M9: LMK 2.8 km & explicit convection M10: Verification • M7: Radar • 5-min. DX-Radar-composits (from 16 German Radars) • Detection (Correction) of disturbing echoes --> Quality control flags = Quality product • European DX composit (15 min.)
Detection of Spokes Berlin 21.06.2005 22:10 UTC Original ‘Pos. 3’ of quality product after spoke detection Latent Heat Nudging
Further development of the concept is needed Hamburg 30.07.2004 6:15 UTC original after detection of spokes falsely recognized strongly disconnected spokes
'Klops’ reasons: • total reflection byconvection bubble? • aerosols? detection by:- histograms
Currently in progress: • Detection of other artificial structures: • anomalous propagation • arched formed echoes (from other radars) • physical reason for ‚Klops‘ occurrence? • --> ‚quality product‘ • Open tasks: • European radar composit (based on DX data)‚COSMO radar composit‘ ?European/‘COSMO‘ quality product?
M7: Radar M8: Latent Heat Nudging M9: LMK 2.8 km & explicit convection M10: Verification M8: Latent Heat Nudging • Thermodynamic feedback and interactions of LMK • Improvements of the method • Alternative methods (?) WG + plenary lecture Chr. Schraff
M7: Radar M8: Latent Heat Nudging M9: LMK 2.8 km & explicit convection M10: Verification M9: LMK 2.8 km & explicit convection • Numerical schemes • Physical Parameterisations • Boundary conditions • case studies • LMK Testsuite
LMK- Numerics • Grid structure: horizontal: Arakawa C vertical: Lorenz • time integrations: time-splitting between fast and slow modes: 3-timelevels: Leapfrog (+centered diff.) (Klemp, Wilhelmson, 1978) 2-timelevels: Runge-Kutta: 2. order, 3. order, 3. order TVD • Advection: for u,v,w,p',T:hor. advection: upwind 3., 4., 5., 6. order for qv, qc, qi, qr, qs, qg, TKE: Courant-number-independent (CNI)-advection:Motivation: no constraint for w (deep convection!) Euler-schemes: WG lecture J. Förstner CNI with PPM advection Bott-scheme (2., 4. order) Semi-Lagrange (trilinear, triquadratic, tricubic) • Smoothing: 3D divergence dampinghorizontal diffusion 4. order
Questions currently considered: advection of moisture variables: advection or conservation form of equations? treatment of the density: diagnostic or extra transport equation? Eulerian or Semi-Lagrange approach? Properties of time-splitting (Runge-Kutta) schemes: stability WG lecture M. Baldauf --> plenary lecture J. Steppeler convergence, accuracy conservation Coupling with physics?underestimation of precipitation in convective events because of too weak coupling? Problems with steep orography over the Alps?Cold pools in narrow, deep valleys
Gaussian hill Half width = 40 km Height = 10 m U0 = 10 m/s isothermal stratification dx=2 km dz=100 m T=30 h analytic solution: black lines simulation: colours + grey lines w in mm/s Test of the dynamical core: linear, hydrostatic mountain wave
Test of the dynamical core: density current (Straka et al., 1993) ‘ after 900 s. (Reference) by Straka et al. (1993) RK3 + upwind 5. order RK2 + upwind 3. order
LMK- physics • turbulence: 1D, 1-equation model (prognostic TKE)3D, complete coordinate transformations WG lecture M. Baldauf1-equation model'moist turbulence'(buoyancy production of TKE altered by condensation processes) • cloud microphysics:6-class-scheme (qv, qc, qi, qr, qs, new: graupel qg) WG + plenary lecture T. Reinhardt6-class/2-moments-scheme (for research/benchmark purposes) • radiation: 2-flux-scheme (Ritter, Geleyn, 1992)15 min. update frequency, horizontally averaged WG lecture T. Reinhardt • soil-vegetation-model 2 levels --> 7 levels • convection:- no cumulus convection parametrization - ‚simple‘ shallow convection: WG lecture A. Seifertapply only shallow convection part from Tiedtke (1989) only for cloud 'heights' < 250 hPa
LM Radar LMK, Testsuite 1.7 Explicitly resolved convection in LMK (case study '26.08.2004')
Example: 18.07.2004, 00 UTC + 10 h LMK Radar LM
Example: 18.07.2004, 00 UTC + 12 h LMK Radar LM
Example: 18.07.2004, 00 UTC + 18 h LMK Radar LM
Currently under consideration in physical processes: • shallow convection: • initiation? • cloud microphysics: • tuning of parameters? • turbulence: • is 3D turbulence significant for the 2.8 km scale?horizontal diffusion coefficient? • little experience about moisture turbulence • More general tasks: • behaviour of LMK for the interaction with fine scale orography:Föhn storms, severe downslope winds, ...(compare ALADIN/HIRLAM: several ‚Bora‘ studies) • general underestimation of precipitation in convective situations
M7: Radar M8: Latent Heat Nudging M9: LMK 2.8 km & explicit convection M10: Verification • M10: Verification • Traditional verification (synoptic / upper air) • Verification by synthetic satellite and synthetic radar data • Development of new methods (pattern recognition, upscaling, ...) WG lecture C.-J. Lenz
> 0.1 mm/h > 2 mm/h LM, 7km, diagn. prec. LM, 2.8 km, diagn. prec. (TS 1.1) LMK, progn. prec. (1.3, 1.4) > 10 mm/h LM and LMK TSS for precipitation events > 0.1 mm/h, > 2 mm/h, > 10 mm/h January 2004, 00-h-runs
> 0.1 mm/h > 2 mm/h LM, 7km, diagn. prec. LMK, 6-Klassen-Wolkenphysik (1.7) LMK, 5-Klassen-Wolkenphysik (1.6) LMK, mit Datenassim. (2.1) > 10 mm/h LM and LMK TSS for precipitation events > 0.1 mm/h, > 2 mm/h, > 10 mm/h September 2004, 00-h-runs
TSS for precipitation, Exp. 5193 with LHN,‘07.07.2004 - 19.07.2004’ 00 h runs 12 h runs LMK, LM
Project LMK: Milestones • Summer 2003Start of the project LMK • End 2003: First test-suite with LM at high resolution is running. • End 2004:Prototype version of the LMK-System with data assimilation but without LHN running. • Autumn 2005:Prototype version of the LMK-System with LHN is running.LMK in quasi-operational mode. • Early 2006:Start of a pre-operational test-phase.Fine-Tuning and final evaluation of all components of the system. • End 2006:Start of the operational application.
LMK - (LM-Kürzestfrist)‘Aktionsprogramm 2003’ of DWD Goals • Development of a model-based NWP system for very short range (‘Kürzestfrist’) forecasts (2-18 h) of severe weather events on the meso- scale, especially those related to • deep moist convection (super- and multi-cell thunderstorms, squall-lines, MCCs, rainbands,...) • interactions with fine-scale topography(severe downslope winds, Föhn-storms, flash floodings, fog, ...)
Small clusters reasons: - echoes from moving obstacles (ships, wind power plants) detection by: - neighbour pixels - vertical extension of signals
Neu zugängliche Testsuiten-Ergebnisse TS 2.1: mit Datenassimilation • Hauptlaufanalysen (Sept-Nov. 2004) TS 2.2: mit Datenassimilation • ‘Aerosolbug’ behoben • Druckproblem behoben • Semi-Lagrange-Transport für Feuchtevariablen • Hauptlaufanalysen • Vorhersagen (z.Z. Juli + Aug. 2004)
Planung M9: Geplante LMK-Testsuiten TS 2.2b: wie TS 2.2, aber mit Bott-2-Advektion in Nichterhaltungsform; ab '1.7.2004' erste TS auf der cos1! TS 2.3: mit 7-Schichten-Bodenmodell (Verbesserung von Gefrier/Auftauprozessen -> Winterfall) ab '1.2.2005' Rand vom LME, neues Parameterfile, neue Schichteneinteilung (50 Schichten, dz=20 m unten) TS 3.1: wie TS 2.3 + Latent Heat Nudging ab '1.4.2005' Start: wenn TS 2.3 am '1.4.05' angelangt ist Anfangsdaten von TS 2.3
Final Remarks and Outlook • Preliminary test of LM at 2.8km grid-spacing (operational set-up,convection switched off) from Dec 2003 - Feb 2004- stable and robust- deficiencies from explicit convection -> shallow convection needed- sometimes convective structures verify subjectively, sometimes not- pressure discontinuity at lateral boundaries is solved now- spin-up of about 1 h (also at lateral boundaries) • Re-run of the testsuite for the same starting from Jan 2004 using- RK3-TVD time-split integration with 5th order horizontal advection (no horizontal diffusion required)- full budget equations for rain and snow (prognostic precipitation)- SLEVE (smooth level vertical) coordinate • Upgrade with new physics / dynamics and data assimilation • Deterministic and probabilistic products • Many open questions concerning skill and predictability
LMK - Subprojekte (‚Maßnahmen‘) M7: Radar • 5-min DX-Radar-komposits, qualitätskontrolliert • Europäisches DX Komposit • (Assimilation von Radarwinden) M8: Latent Heat Nudging • Thermodynamische Rückkopplung und Wechselwirkungen des LHN • Verbesserungen der Methode • Alternative Methoden (?) M9: LMK 2.8 km & explizite Konvektion • Numerische Schemata • Physikalische Parametrisierung • Seitliche und obere Randbedingungen • Fallstudien und Vergleiche • LMK Testsuite M10: Verifikation • Traditionelle Verifikation (Syn/Temp) • Verifikation anhand von synthetischen Satelliten- and synthetischen Radardaten • Entwicklung neuer Methoden (Mustererkennung, upscaling, ...)
see also 5. SRNWP - Workshop 2003, Bad Orb Further planned tests: • Idealized test cases especially for the dynamical core • Linear mountain flow • isotherm, hydrostatic • non-isotherm ! • non-hydrostatic ! • ‚Schär‘ • Non-linear mountain flows ! • mountain vortex street ! • cold bubble (Straka et al., 1993) • warm bubble in shear flow (Weisman, Klemp, 1982) • test cases especially for cloud physics scheme • IMPROVE-2
Data assimilation: special requirements for LMK • Assimilation of structures on the meso--scale necessary <-- deep convection • Observations:--> need for high resolution, rapidly updated data fields --> radar observations ('precipitation scan') horizontal resolution: r~1 km, ~1°, temporal resolution: t~5 / 15 min (Germany/Europe) ) max. range: ~ 120 km • assimilation method: - fast - relatively easy to implement --> Latent Heat Nudging
LMK - Dynamics • Model equations: non-hydrostatic, full compressible, advection form • Base state: hydrostatic • Prognostic variables: cartesian wind components u, v, w • pressure perturbation p‘, Temperature T (or T‘=T-T0)humidity var. qv, qc, qi, qr, qs, qg ('prognostic precip.') • TKE • Coordinate systems: rotated geographical coordinates • generalized terrain-following height coordinate • user-defined vertical stretching • Open questions: • shallow / deep atmospheric equationsi.e. are terms ~w in advection earth curvature and Coriolis force important?(diploma thesis)
Tiefe/flache Atmosphäre LM-Gleichungen für tiefe Atmosphäre in Kugelkoordinaten: flache Atmosphäre: • r ~ a • vernachlässige Terme in Advektion und Corioliskraft
Metric terms of 3D-turbulence scalar flux divergence: terrain following coordinates analogous: ‚vectorial‘ diffusion of u, v, w (Baldauf (2005), COSMO-Newsl.) earth curvature Idealized test: 3D isotropic diffusion over sinusoidal orography without metric terms with metric terms
LM Radar LMK, Testsuite 1.7 Example of explicitly resolved convection in LMK (case '26.08.2004')
Zusammenfassung der Verifikationsergebnisse: - Böen: durchweg höhere Qualität der LMK-Prognose gegenüber der LM-Routine infolge des feineren numerischen Gitters - Niederschlag: LMK-Ergebnisse liefern teilweise bessere Resultate bei der Verifikation als das LM - Verschiedene Tests (e.g. Parameterisierung der flachen Konvektion, Datenassimilation) führen zu besseren Verifikationsergebnissen bei einigen Parametern, andere Parameter verifizieren weniger gut
2.8 km from: R. A. Houze, Jr.: Cloud Dynamics International Geophysics Series Vol. 53 Deep moist convection Schematic model from a Colorado storm case study (Raymer Hailstorm)
Need for a shallow convection parameterization: Testsuite 1-5 without shallow convection pressure jump problem operational LM the ‚truth‘ (?) Testsuite 1-6 with simple shallow convection param.
Shallow convection based on Tiedtke-scheme N-S-crosssection rh(with shallow convection) Diff.: rh(with sh. conv) - rh(without sh. conv.) Dr. F. Theunert (AGeoBW)