eucos observing system experiments with the cosmo ......• eucos stands for “eumetnet composite...
TRANSCRIPT
-
EUCOS observing system experimentswith the COSMO Local Model
Jean-Marie BettemsSwiss Federal Office for Meteorology and Climatology3hd COSMO General Meeting, Athens, October 2001
• EUCOS stands for “EUMETNET Composite Observing System”
• The main goal of the EUCOS program is to define the optimumground based observing sytem for short range NWP
• Two scenarios have been proposed, both associated with a drasticreduction of the number of radiosonde stations over continentalEurope and an increased number of AMDAR platforms
• A special observation campaign took place from 20 September to14 November 1999 to support observing system experiments
• Observing system experiments with both global and limited areamodels are made to evaluate the impact of the proposed scenarios
• OSE studies have been done with the ECMWF global model (T319),with the DMI-HIRLAM-G (50km) model and with the high resolutionCOSMO Local Model (7km)
-
3hd COSMO general meeting, Athens octobre 2001 [email protected]
Experimental set-up (1)
• A 27 days period was chosen, from 19th October 1999 to 15th November 1999;
this period includes seven MAP heavy precipitation events and was also chosen by
DMI
• Three observing systems have been considered:
S1 is the standard observing systembefore EUCOS field experiment
S2 has 37 radiosonde stations less, but 13 upgraded stations (-50% bulletins)
S3 is S2 complemented by 121 more AMDAR platforms (+400% bulletins)
• For each observing system: a 27 days continuous assimilation and a daily 24h
forecast have been calculated with the non-hydrostatic COSMO Local Model
• All experiments were based on the operational configuration of the Local Model
installed at MeteoSwiss:
1. 385x325 mesh, about 7km horizontal resolution, 45 layers in the vertical
2. data assimilation based on the nudging scheme, with use of all conventional
observing systems (SYNOP, SHIP, DRIBU, AIREP, AMDAR, TEMP, PILOT)
3. aicraft data grouped in multi-level reports before assimilation
4. boundary conditions from the ECMWF operational deterministic forecast
(forecast) and from the operational 4d var stream (assimilation)
5. soil model updated once a day from the driving model
• Results have been evaluated with two verification packages:
1. verification of the vertical structure of the atmosphere against a set of 28
radiosondes regularly distributed over the whole LM domain
2. verification of near-surface weather parameters over Switzerland
-
3hd COSMO general meeting, Athens octobre 2001 [email protected]
Experimental set-up (2)
a. all observations influencing the model flow during the considered 24h period.
TEMP PILOT AIRCRAFT SYNOP
S1active stations 55 12 135 1078
active reportsa 202 24 4671 18542
S2active stations 25 12 135 1078
active reports 110 24 4671 18542
S3active stations 25 12 216 1078
active reports 110 24 21261 18542
Local Model domain
60˚N
50˚N
40˚N
0˚ 20˚E20˚W
Typical daily coverage of AMDAR platforms in S3
Typical number of observations assimilated in 24h
-
4th SRNWP Workshop, Bad Orb septembre 2001 [email protected]
Impact of radiosonde network modification
Number of active radiosondes over LM domain reduced from 55 to 25
From the remaining stations 12 have been upgraded to measure every 6hours
• Significant degradation of model quality at analysis time
both dynamical and thermodynamical fields are affected
mainly visible in the distribution of model fields (meso-β structures)
consistent with the results of an earlier OSSE study [Bettems, 1999]
• Negative impact in the very short range forecast
after 12hour most differences have disappeared
-
4th SRNWP Workshop, Bad Orb septembre 2001 [email protected]
Relative humidity +00h (12UTC) All stations
Bias Standard deviation
Vertical structure of the atmosphere
S1S2S3
-
4th SRNWP Workshop, Bad Orb septembre 2001 [email protected]
Wind speed +00h (12UTC) All stations
Bias Standard deviation
Vertical structure of the atmosphere
S1S2S3
-
4th SRNWP Workshop, Bad Orb octobre 2001 [email protected]
Wind speed +00h (12UTC) Innsbruck (not part of S1/S2/S3)
Bias Standard deviation
Vertical structure of the atmosphere
S1S2S3
-
4th SRNWP Workshop, Bad Orb octobre 2001 [email protected]
Total precipitation, mean over 5 grid points All active ANETZ stations
Near-surface weather parameters over Switzerland
S1S2S3
(Obs. frequence of Payerne RS is double in S2/S3)
-
4th SRNWP Workshop, Bad Orb septembre 2001 [email protected]
Impact of additional AMDAR platforms
Number of active AMDAR platforms in LM domain increased from ~135 to ~216
Number of AMDAR reports increased by a factor 4
• Over the whole model domain the impact is positive but very small
• Clear local improvement, in regions well covered by AMDAR platforms
humidity field is also improved
does not compensate quality loss brought by the radiosonde network degradation
-
4th SRNWP Workshop, Bad Orb octobre 2001 [email protected]
Wind speed +00h (12UTC) Stuttgart (many AMDARs around that location)
S1S2S3
Bias Standard deviation
Vertical structure of the atmosphere
-
4th SRNWP Workshop, Bad Orb septembre 2001 [email protected]
Wind direction All representative ANETZ stations below 800m
Near-surface weather parameters over Switzerland
S1S2S3
-
4th SRNWP Workshop, Bad Orb octobre 2001 [email protected]
46ON
48ON
4OE
4OE
6OE
6OE
8OE
8OE
10OE
10OE
12OE
12OE
6h Sum of total rain and snow in mmLM Forecast 06.11.1999 12 UTC +6h VT: Sat 6 Nov 1999 18 UTC
Copyright MeteoSwiss
0.1 0.5 1 5 10 20 50 100 121.8
.5 rosa - bettems Tue May 8 10:52:37 2001 S3a
Exper
imen
t S3
46ON
48ON
4OE
4OE
6OE
6OE
8OE
8OE
10OE
10OE
12OE
12OE
6h Sum of total rain and snow in mmLM Forecast 06.11.1999 12 UTC +6h VT: Sat 6 Nov 1999 18 UTC
Copyright MeteoSwiss
0.1 0.5 1 5 10 20 50 79.5
5.5 rosa - bettems Fri May 11 01:03:54 2001 S1a
Exper
imen
t S1
46ON
48ON
4OE
4OE
6OE
6OE
8OE
8OE
10OE
10OE
12OE
12OE
6h Sum of total rain and snow in mmLM Forecast 06.11.1999 12 UTC +6h VT: Sat 6 Nov 1999 18 UTC
Copyright MeteoSwiss
0.1 0.5 1 5 10 20 50 100 122.8
.5 rosa - bettems Sun May 6 18:21:45 2001 S2a
Exper
imen
t S2
Heavy precipitation case – 6th November 1999
6h precipitation sum from LM
MAP Alpine radar composite
(12UTC - 18UTC)
-
3hd COSMO general meeting, Athens octobre 2001 [email protected]
Conclusion
• The results obtained at a higher horizontal resolution with theLocal Model (~7km) differ substantially from those obtainedwith the global ECMWF model (~60km) or the DMI-HIRLAM-Glimited area model (~50km).
• ECMWF and DMI studies have shown that a reduction of thenumber of active radiosonde stations had a very small negativeimpact, and that the additional AMDAR data slightly degradethe forecast quality.
• This study suggests a significative degradation of the shortrange forecast quality associated with the introduction of theproposed EUCOS scenario, if no compensatory action are taken.
AMDAR observations have a clear potential, and an improvedhorizontal coverage would mitigate the expected quality loss.
However, additional information about local structures in thehumidity field are needed (e.g. GPS).
• EUCOS stands for “EUMETNET Composite Observing S...• The main goal of the EUCOS program is to define ...• Two scenarios have been proposed, both associate...• A special observation campaign took place from 2...• Observing system experiments with both global an...• OSE studies have been done with the ECMWF global...EUCOS observing system experiments with the COSMO ...Jean-Marie Bettems Swiss Federal Office for Meteor...• A 27 days period was chosen, from 19th October 1...• Three observing systems have been considered: S1...• For each observing system: a 27 days continuous ...• All experiments were based on the operational co...• Results have been evaluated with two verificatio...• The results obtained at a higher horizontal reso...• ECMWF and DMI studies have shown that a reductio...• This study suggests a significative degradation ...