Stochastic ensemble - Simon Fresnay

Predictability of a catastrophic rainfall event (Var 2010). Evaluation of .... Surface low may at H-12 may have a precursor role on the HPE. Pressu re levels ...
Télécharger le PDF
4MB taille 0 téléchargements 266 vues
commentaire
Report
Predictability of a catastrophic rainfall event (Var 2010) Evaluation of precursors Simon Fresnay (1*) Philippe Arbogast (2) Dominique Lambert (1) Karine Maynard (3) Evelyne Richard (1) (1) Laboratoire d’Aérologie, University of Toulouse and CNRS (2) CNRM-GAME, Toulouse (3) Météo-France

[email protected] European Geosciences Union | General Assembly 2013 | Vienna, Austria

*Allocated a doctoral grant from AXA Research Fund

Predictability of a catastrophic rainfall event (Var 2010) – S. Fresnay – EGU 2013 Context The Var Case Predictability Sensitivity experiment

Context • Heavy precipitation events (HPEs) in the Mediterranean

Conclusion

Climatology of HPEs in Southern France (before 2006) FRANCE

Var

• Key factors • upper-air trough • southerly flow (France) • low-level humidity convergence • orographic forcing

SPAIN

Mediterranean Sea

• The Var region, in Southern France • under-investigated (meteorology) • 6 events with RR > 200mm since 1967

Predictability of a catastrophic rainfall event (Var 2010) – S. Fresnay – EGU 2013 Context The Var Case Predictability Sensitivity experiment

Conclusion

1. The Var Case A) Overview • 397mm in 24hrs (Les Arcs) • multi-stage event • unusual (climatology) • misforecast

1-hr rainfall time series

Stage1 6-hr precip RADAR+gauges

Stage1 : 149mm/6hrs (Hyeres) Stage2 : 261mm/9hrs (Les Arcs)

Predictability of a catastrophic rainfall event (Var 2010) – S. Fresnay – EGU 2013 Context The Var Case Predictability Sensitivity experiment

Conclusion

1. The Var Case B) Stage 1 and forecasting issues ARPEGE dx=10km

Operational forecast 6-hr precip - Stage 1

max=16mm

AROME dx=2.5km max=43mm

init=H-12

1-hr rainfall time series

Stage1 6-hr precip RADAR+obs

H+0 Stage1 : 149mm/6hrs (Hyeres) Stage2 : 261mm/9hrs (Les Arcs)

Predictability of a catastrophic rainfall event (Var 2010) – S. Fresnay – EGU 2013 Context The Var Case Predictability Sensitivity experiment

2. Predictability A) Precipitation Data: ensemble of ARPEGE operational forecasts initialisation at H-12 … H-54 (N=8) 6-hr precip stage 1 (H+0…H+6) init=H-54

H-48

H-42

H-36

OBS

H-30

H-24

H-18

H-12

> Scenarios extend between 10% and 70% of the rainfall observed > Net forecasting issues (forecasts lose skill with time)

Conclusion

Predictability of a catastrophic rainfall event (Var 2010) – S. Fresnay – EGU 2013 Context The Var Case Predictability Sensitivity experiment

2. Predictability B) Dynamics 250-hPa jet (40 m/s)

MSLP (1008 hPa)

H+0

Longitude (degE)

H-12

S/SW upper-level jet Surface low at jet right entrance: > Deep convection, cyclogenesis

Conclusion

Predictability of a catastrophic rainfall event (Var 2010) – S. Fresnay – EGU 2013 Context The Var Case Predictability Sensitivity experiment

Conclusion

2. Predictability B) Dynamics MSLP (1008 hPa)

H+0

Jet E-W displacement

Longitude (degE)

H-12

S/SW upper-level jet Surface low at jet right entrance: > Deep convection, cyclogenesis

Longitude (degE)

250-hPa jet (40 m/s)

bifurcation

H-12

H-6

H+0

Bifurcation = high sensitivity of the upper-level dynamics > Role of convective processes to disrupt predictability?

Predictability of a catastrophic rainfall event (Var 2010) – S. Fresnay – EGU 2013 Context The Var Case Predictability Sensitivity experiment

Conclusion

3. Sensitivity experiment A) Initial condition perturbation Different analyses at H-12 on the surface pressure south of Mediterranean

CTRL (ARPEGE)

ECMWF

Predictability of a catastrophic rainfall event (Var 2010) – S. Fresnay – EGU 2013 Context The Var Case Predictability Sensitivity experiment

Conclusion

3. Sensitivity experiment A) Initial condition perturbation Different analyses at H-12 on the surface pressure south of Mediterranean Dynamical balance initialisation

CTRL (ARPEGE)

ECMWF

MSLP1

Predictability of a catastrophic rainfall event (Var 2010) – S. Fresnay – EGU 2013 Context The Var Case Predictability Sensitivity experiment

Conclusion

3. Sensitivity experiment B) Results CTRL

MSLP1

OBS

Pressure levels

6-hr precipitation

Vertical section θe , omega

> Role of the low-level modification on the activation of deep convection 12 hours later

Predictability of a catastrophic rainfall event (Var 2010) – S. Fresnay – EGU 2013 Context The Var Case Predictability Sensitivity experiment

Conclusion

3. Sensitivity experiment B) Results CTRL

MSLP1

Jet E-W displacement

Pressure levels

Longitude (degE)

CTRL MSLP1

H-12

IC perturb > Role of the low-level modification on the activation of deep convection 12 hours later

H-6

H+0

> Role of the low-level modification on the western displacement of the jet 6 hours later

> Surface low may at H-12 may have a precursor role on the HPE

Predictability of a catastrophic rainfall event (Var 2010) – S. Fresnay – EGU 2013 Context The Var Case Predictability Sensitivity experiment

Conclusion

3. Sensitivity experiment C) Ageostrophic circulation 250 hPa wind (>40m/s) ageo.wind (arrows)

H-9

H-6

H-3

H+0

CTRL

MSLP1

> Net enhancement of upper-level wind and cross-jet ageostrophy (jet entrance) from H-6

Predictability of a catastrophic rainfall event (Var 2010) – S. Fresnay – EGU 2013 Context The Var Case Predictability Sensitivity experiment

Conclusion

3. Sensitivity experiment C) Ageostrophic circulation 250 hPa wind (>40m/s) ageo.wind (arrows)

H-9

H-6

H-3

H+0

CTRL

MSLP1

> Net enhancement of upper-level wind and cross-jet ageostrophy (jet entrance) from H-6

Predictability of a catastrophic rainfall event (Var 2010) – S. Fresnay – EGU 2013 Context The Var Case Predictability Sensitivity experiment

Conclusion

3. Sensitivity experiment C) Ageostrophic circulation Cross-jet sections (ageo.wind, divergence) (avg 40-42N)

CTRL

H-6

MSLP1

Predictability of a catastrophic rainfall event (Var 2010) – S. Fresnay – EGU 2013 Context The Var Case Predictability Sensitivity experiment

Conclusion

3. Sensitivity experiment C) Ageostrophic circulation Cross-jet sections (ageo.wind, divergence) (avg 40-42N)

CTRL

H-6 jet entrance

MSLP1

mid-level based convection low-level convection (Var) Jet entrance interacts with the enhanced midlevel based convection

Predictability of a catastrophic rainfall event (Var 2010) – S. Fresnay – EGU 2013 Context The Var Case Predictability Sensitivity experiment

Conclusion

3. Sensitivity experiment C) Ageostrophic circulation Cross-jet sections (ageo.wind, divergence) (avg 40-42N)

(avg 41-43N)

CTRL

H-6

H-3 jet entrance

MSLP1

mid-level based convection low-level convection (Var) Jet entrance interacts with the enhanced midlevel based convection

Predictability of a catastrophic rainfall event (Var 2010) – S. Fresnay – EGU 2013 Context The Var Case Predictability Sensitivity experiment

Conclusion

3. Sensitivity experiment C) Ageostrophic circulation Cross-jet sections (ageo.wind, divergence) (avg 40-42N)

(avg 41-43N)

(avg 42-44N)

CTRL

H-6

H-3 jet entrance

MSLP1

H+0 Deep convection

mid-level based convection low-level convection (Var) Jet entrance interacts with the enhanced midlevel based convection

Enhanced low-level convection finally leads up to deep convection

Predictability of a catastrophic rainfall event (Var 2010) – S. Fresnay – EGU 2013 Context The Var Case Predictability Sensitivity experiment

Conclusion

Summary and conclusion > Low predictability of the upper-level and precipitation > Low-level modification at H-12 impacts the jet and precipitation at H+0 > Ageostrophic circulation

H-3

No deep convection (CTRL) H+0

jet dry

cold

W

dry

cold

warm

E

W

warm

E

Predictability of a catastrophic rainfall event (Var 2010) – S. Fresnay – EGU 2013 Context The Var Case Predictability Sensitivity experiment

Conclusion

Summary and conclusion > Low predictability of the upper-level and precipitation > Low-level modification at H-12 impacts the jet and precipitation at H+0 > Ageostrophic circulation

H-3

Deep convection (MSLP1) H+0

jet dry

cold

dry

cold

warm

W

E

> Jet streak highly sensitive to convective processes > Precursor role of H-12 surface low > Complex interplay of various ageostrophic components

W

warm

E

Predictability of a catastrophic rainfall event (Var 2010) Evaluation of precursors Simon Fresnay (1*) Philippe Arbogast (2) Dominique Lambert (1) Karine Maynard (3) Evelyne Richard (1) (1) Laboratoire d’Aérologie, University of Toulouse and CNRS (2) CNRM-GAME, Toulouse (3) Météo-France

[email protected] European Geosciences Union | General Assembly 2013 | Vienna, Austria

*Allocated a doctoral grant from AXA Research Fund

OBS

init=H-18

ECMWF

ARPEGE

init=H-12

init=H-6

Arbogast et al (2008, QJRMS) PV inversion method Inversion of the Ertel PV with an implicit balance obtained by digital filter initialisation (DFI) → provides 3D-balanced fields suitable as initial fields Iterate trajectories for a low-order model 1=DFI 2=PV constraint (variational pb) M0

2 1 Balanced subspace

PV

height N = expected solution (theoretical) M11 = consistent with the model (DFI), but wrong PV M2 = good PV but imbalanced (gravity waves would restore the balance)

Step 0: compute first iterate M0 from PV (usually geostrophic=linear) Step1: apply DFI to get a balanced solution. DFI tends to change the PV (filtering removes energy) Step2: restoring the PV while minimizing energy difference from M11 (variation problem under the PV constraint)→we get a closer state from M1 with the right PV Step3: reapply DFI to get a balanced solution, etc.

Surface modification: inversion step

Free troposphere

PV inversion

Φb BL Surface

Climatology

MSLP

U θ Ub , θb Interp.

Us , θs

H-3 T2 T1

B2

B1

T2

latitude T1 B1

B2

T2

T1 B1

B2