high-resolution modelling in mountainous areas: map results
DESCRIPTION
High-resolution modelling in mountainous areas: MAP results. Evelyne Richard Laboratoire d’Aérologie CNRS / Univ. Paul Sabatier Toulouse, France. What are the skills of high-resolution models to forecast orographically influenced precipitation? - PowerPoint PPT PresentationTRANSCRIPT
High-resolution modelling in mountainous areas:
MAP results
Evelyne RichardLaboratoire d’AérologieCNRS / Univ. Paul SabatierToulouse, France
What are the skills of high-resolution models to forecast orographically influenced precipitation?
Does explicit (versus parameterized) convection lead to a gain in predictability?
IOP 2a – 17 September 1999
A short, intense, isolated, convective event
70 mm within 12 hours
Sensitivity experiments performed with Meso-NH
MAP IOP 2a: IR Meteosat
MAP target area
18:00 UT 19:00 UT 20:00 UT
MAP – IOP 2AComposite radar reflectivity @ z = 2km
25
0 k
m
21:00 UT 22:00 UT 23:00 UT
Toce Ticino watershed
Observation
Simulation (Δx =2km)
20:00 UT
23:00 UT
Reflectivity @ 2000m
ECMWF: OP. ANA 1999ECMWF: OP. ANA 1999
RADAR
12 hour accumulated precipitation
SIMULATION
17:00 UT 18:00 UT
19:00 UT 20:00 UT
Composite radar reflectivity @ z = 2km
Z > 60 dBz
25
0 k
m
Radar Retrieval (S-Pol)
Simulation (Meso-NH)
(x) hail + graupel
(o) hail
graupel
hail
18:00 UT
19:00 UT
20:00 UT
rain rain
12
km
100 km
Great !
My model is doing a good job
ECMWF: OP. ANA ECMWF: OP. ANA 19991999
ARPEGE: OP. ANA. 1999ARPEGE: OP. ANA. 1999 ECMWF: OP. ANA. ECMWF: OP. ANA. 20022002
ECMWF: REANALYSISECMWF: REANALYSIS
MAP - IOP2A:
Intense Convection
Strong sensitivity to initial state
Low predictability
RADAR OBSERVATIONS
ANA. OP. 1999ANA. OP. 1999 REANALYSISREANALYSIS
850hPa water vapor mixing ratio : 17 September 1999 12UTC
IOP 2a
REANA (NO MAP REANA (NO MAP DATA)DATA)
ECMWF ANALYSIS MAP ECMWF REANALYSIS
Low-level convergence between the Ligurian and Adriatic flows
Increase in the model resolution -> higher mountains -> the Ligurian flow is blocked
Streamlines at 1000 m, 17/09/99 12 UTC
MAP IOP 2aLascaux et al., 2004
Prectitability ?Still a long way to
go !
IOP2b 20/21 September 1999:
Orographic enhancement of a frontal system
200 mm within 30h
Model intercomparison : MC2, MM5, MOLOCH,Meso-NH
How do the models compare with each other ?
19 Sept. 1999 12:00 20 Sept. 1999 12:00
METEOSAT infrared
Sensitivity to the analysis
ECMWF Op. Analysis MAP Reanalysis
Max: 512 mm Max: 482 mm Mean: 78 mm Mean: 87 mm
The different models:
• MESO-NH 10 KM + 2.5 KM
• MOLOCH 10 KM + 2 KM
• MM5-RE 27 KM + 9 KM + 3 KM
• MM5-E1 18 KM + 6 KM + 2 KM
• MC2 40 KM + 10 KM +2 KM
Initial and boundary conditions from ECMWF operational analyses
From 19 Sep. 12 UTC to 20 Sep. 18 UTC
(30 hours)
MAP - IOP2B - 19-20 Sep. 1999
Intercomparison exercise
4 non-hydrostatic models with horizontal resolution of 2 to 3 km
Initialization based upon ECMWF operational analysis
Accumulated precipitation from the 19th 15 UTC to the 20th 18UTC
Toce-Ticino watershed
Rain gauges
Time evolution of the mean hourly precipitation rate
Radar
Time evolution of the correlation coeffecient (wrt rain gauges)
Heidke skill scores as a function of precip. class
1h precip. 27h precip.
Comparison with rain gauge measurements (121 points)
Correlation 1h
Correlation 6h
Correlation 27h
Mean Bias
MESONH 0.33 0.49 0.62 + 28 % MOLOCH 0.31 0.47 0.62 - 22 % MM5-RE 0.27 0.43 0.55 - 16 % MM5-E1 0.37 0.53 0.63 + 28 %
MC2 0.30 0.47 0.63 - 32 % RADAR 0.48 0.55 0.60 - 54 %
Toce watershed
1532 km2
Hydrological response
Grossi et al., 2004
Grossi et al., 2004
How does the flow over complex terrain modify the growth mechanisms of precipitation particles?
Three Doppler radars
Monte Lema – Ronsard – S Pol
• Dual Doppler analysis
• 3D wind fields rerievals
• Microphysical retrievals
Monte Lema
S Pol
Ronsard
Medina and Houze, 2003Medina and Houze, 2003
U/Nh < 1
U/Nh > 1
dry snow
wet snow
light rain
graupelriming
heavy rain
coalescence
Blocked and stable case
Unblocked and unstable case
To what extend the models able to reproduce this contrasted behaviour in the microphysics ?
SnowGraupel
Hail
Cloud Rain
Ice
IOP2A
IOP2a ( Strong convection)- Deep system-Large amount of hail and graupel
Mean vertical distribution of the hydrometeors
IOP8 ( Stratiform event)- Shallow system- Large amount of snow
IOP8
Snow
Dominant microphysical processes:
DEPOSITION on ice(and sublimation)
Growth of graupelby RIMING
AUTOCONVERSIONof pristine ice
MELTING-CONVERSIONof the snow (into graupel)
ACCRETION of cloud dropletsby raindrops
Depletion of graupel by
WET GROWTH of hail
IOP2a IOP8
Conclusion:
IOP 2A (Predictability)
The use of non-hydrostatic high-resolution models will improve the precipitation forecast but only to some extend.
Further improvement is tied to the improvement of the model initial state
Adding mesoscale data in a global assimilation system is insufficient
•Mesoscale data assimilation system
•Limited area ensemble forecast
(MAP D-PHASE)
IOP 2b (Model Intercomparison)
– Very good consistency of the accumulated precipitation pattern
– Model results over/under estimate the total precipitation by a factor ranging from +30% to -30%
– The accuracy of the model precipitation is rather weak for the hourly rainfall but fairly reasonable for the precipitation accumulated over the 30h time period of the event
– However model results are not yet accurate enough to be used for hydrological forecast on small watersheds
Explicit microphysical schemes do provide fairly realistic results …
The contrasted microphysical behaviour between different IOPs is reasonably reproduced
• Convective – flow over ----> Strong riming and coalescence
• Stratiform – blocked flow ----> Melting of snow
http://www.aero.obs-mip.fr/map/MAP_wgnum
(1) LA CNRS/UPS, Toulouse, France
(2) CNRM, Météo-France, Toulouse, France(3) RPN, Montréal, Canada
(4) ISAC, CNR, Bologna, Italy(5) University of L ’Aquila, Italy(6) University of Milano, Italy
(7) University of Munich, Germany(8) University of Brescia, Italy
(9) University of Waterloo, Canada
N. Asencio (2), R. Benoit (3), A. Buzzi (4), R. Ferretti (5), F. Lascaux (1), P. Malguzzi (4), S. Serafin (6), G. Zängl (7), J-F. Georgis (1), R. Ranzi (8), G. Grossi (8), N. Kouwen (9)