Improved Weather Prediction Employing Ensemble Forecasts and Data Assimilation T. Gerz, K. Dengler, A. Tafferner, F. Holzäpfel Institut für Physik der Atmosphäre, DLR, Oberpfaffenhofen Limited area model for numerical weather prediction Wind, turbulence and temperature forecasts for wake vortex advisories Convective Precipitation forecasts for the TMA of an airport The DLR Project Wetter & Fliegen 1
COSMOairport a limited area model derived from COSMO-DE COSMOFRA COSMOMUC The DLR Project Wetter & Fliegen 2
Questions We seek improvements in Wind, turbulence and temperature forecasts for wake vortex advisories Convective Precipitation forecasts for the Terminal Manoeuvring Area of an airport 1 Does a hourly updated model run (TLE) improve the forecasts? 2 Does assimilated data from local measurements improve the forecasts? Compare forecasts of COSMOairport and the operational COSMO-DE with respective observations The DLR Project Wetter & Fliegen 3
Model Chain for Frankfurt Airport: COSMOFRA COSMO-EU COSMOFRA FRA ~280 km wake vortex predictor P2P transport and decay of wake vortices u,v,w,ρ, θ v, TKE, EDR, p altitude [m] Points along glide path The DLR Project Wetter & Fliegen 4
Model Chain for Munich Airport: COSMOMUC COSMO-EU COSMOMUC MUC ~360 km wake vortex predictor P2P transport and decay of wake vortices u,v,w,ρ, θ v, TKE, EDR, p altitude [m] Points along glide path The DLR Project Wetter & Fliegen 5
Model Chain for Munich Airport: COSMOMUC COSMO-EU COSMOMUC MUC FC dbz at 850 hpa ~360 km The DLR Project Wetter & Fliegen 6
Downscaling DWD data base COSMO-EU/DE analyses radar data other data Op. COSMO-DE Initial and boundary values assimilation COSMOairport GRIB 60 x 34 km 2 Synthetic satellite images 360 x 360 km 2 GRIB 360 x 360 km 2 projection on glide path Wake Vortex Deep Convection Winter Weather The DLR Project Wetter & Fliegen 7
Time-Lagged Ensemble Predictions using a rapid update cycle The DLR Project Wetter & Fliegen Folie 8
COSMOairport 1-hour assimilation/nowcast cycle (t a ) (t a +1h) (t a +2h) (t a ) local data real time local data COSMO-EU initial & boundary time DATA ASSIMILATION forecast O U T P U T DATA ASSIMILATION DATA ASSIMILATION forecast Boundary data from COSMO-EU Boundary plus assimilated data; Initial data: Restart or NewStart Nudging of local data DATA ASSIMILATION forecast The DLR Project Wetter & Fliegen 9
COSMOMUC for Wake Vortex Advisory and Thunderstorms The DLR Project Wetter & Fliegen 14
COSMOMUC for Wake Vortex Advisory Munich Airport (MUC) LIDAR SODAR/ RASS +SONIC Model set-up: model centred around MUC: 360 x 360 km 2 horizontal resolution Δx=2.8 km vertical resolution 15 130 m, 16 levels to 1100 m boundary conditions from COSMO-EU one run each hour for 6 h forecast horizon (time-lagged ensembles TLE with 6 members) assimilation of precip., SYNOP, TEMP, AMDAR Output: u,v,w, θ v, TKE, EDR each 10 min Forecasts compared with SODAR/RASS (sound detection and ranging / radio acoustic sounding system) 22.06.2010-06.10.2010 (107 days) vertical resolution: 20 m from 40 to 500 m data each 10 minutes, SONIC 20 Hz horizontal wind velocity and direction vertical wind velocity turbulence kinetic energy virtual potential temperature The DLR Project Wetter & Fliegen 15
COSMOMUC for Wake Vortex Advisory SODAR: 14.07.2010 frontal passage just after 22:30 UTC at MUC ( wind jump ) Wind direction Cross-wind Parallel-wind - + + - COSMOMUC Run 18 UTC -> 22:30 UTC Run 19 UTC -> 22:30 UTC The DLR Project Wetter & Fliegen 16
COSMOMUC for Wake Vortex Advisory 18 UTC run -> 22:30 UTC Vertical profiles of wind velocity along glide path 19 UTC run -> 22:30 UTC Glide path The DLR Project Wetter & Fliegen 17
COSMOMUC for Thunderstorms TMA of Munich MUC Model set-up: model centred around MUC: 360 x 360 km 2 horizontal resolution Δx=2.8 km vertical resolution 15 130 m, 16 levels to 1100 m b.c. from hourly analyses of COSMO-EU one run each hour for 6 h forecast horizon (time-lagged ensembles TLE with 6 members) assimilation of precip., SYNOP, TEMP, AMDAR Output each 5 min of TOT_PREC, u 10m, PMSL, CAPE_ML, CIN_ML, DBZ_CMAX, T 2m, TD 2m, T s, QV S, DBZ_850 Forecasts compared with precipitation data (German precipitation radar network) Radar, Δt = 5 min, Δx = 1 km The DLR Project Wetter & Fliegen 19
Observation 17.7.2010 01 UTC Forecast Matched field DAS: 0.77 DIS: 0.28 AMP: 0.49 FAR: 0.09 FBI: 0.30 COSMO-DE DAS: 0.62-19 % DIS: 0.16-43 % COSMOMUC No data assimilated AMP: 0.46-6 % FAR: 0.15 + 66 % COSMOMUC Local data assimilated FBI: 0.43 + 43 % DAS: 0.65-15 % Forecasts start at 21 UTC of 16. July and run 4 h DIS: 0.19-30 % AMP: 0.46-6 % FAR: 0.18 +100 % FBI: 0.39 + 30 % The DLR Project Wetter & Fliegen 25
Compared to the operational model COSMO-DE yields Summary a limited area model COSMOairport with rapid updates (time-lagged ensembles) and assimilation of local data small RMS errors of wind for short lead time forecasts (- 2 Std) significant reduction of RMS error by almost 2 m/s for u p improved forecasts of TKE from ground to 900 m altitude improved forecast of frontal passages improved forecast of position of convective precipitation cells Future Work Assimilation of data from LLWAS (ITWS) of DWD in FRA und MUC Use of COSMO-DE, COSMO-EPS for boundary values Adaptation of COSMOairport to allow for higher spatial resolution The DLR Project Wetter & Fliegen 26