Limited-area ensembles using targeted SVs from ECMWF

Limited-area ensembles using targeted SVs from ECMWF

Inger-Lise FrognerNorwegian Meteorological Institute

Meteorologisk institutt met.no

Topics

• Experiments using targeted SVs for perturbing HIRLAM (QJRMS (2002) 128, pp 1321-1341)

• Targeted vs. not targeted perturbations

• Two case studies• Summary/Plans

Meteorologisk institutt met.no

Verification area

• Target area SVs

Integration areaHIRLAM

Meteorologisk institutt met.no

SET-UP LAMTEPS

• LAMTEPS - Limited Area Model Targeted Ensemble Prediction System

• HIRLAM perturbed with targeted EPS from ECMWF (TEPS)

• 48h optimization interval for TSVs.• HIRLAM analyses used• Two different approaches:

– Perturbed ini. and boundary conditions (PB)– Perturbed only initial conditions (NP)

• Test periods in June and Feb. 1997• 28 km, 31 levels, 20+1 ensemble

members

Meteorologisk institutt met.no

I

i

N

n

D

didind pe

DNIspread

1 1 1

2)(11

Comparison of TEPS and

operational EPS

Meteorologisk institutt met.no

Thin lines: winterHeavy lines: summer

Meteorologisk institutt met.no

Verified against “superobservations”

THR = 20mm/24hTOTAL PRECIP.

WINTER PERIODVerification time: 40h

TEPSLAMTEPS (PB)

Meteorologisk institutt met.no

THR = 30mm/24hTOTAL PRECIP.

WINTER PERIODVerification time: 40h

TEPS

LAMTEPS (PB)

Meteorologisk institutt met.no

•

TEPS

LAMTEPS (PB)LAMTEPS (NP)

THR = 2m temp. colder than -6K

compared to climatology.

WINTER

Meteorologisk institutt met.no

THR = 2m temp.

colder than -6K

compared to

climatology.

WINTER

TEPSLAMTEPS (PB)LAMTEPS (NP)

Meteorologisk institutt met.no

TEPS

LAMTEPS (PB)

LAMTEPS (NP)

THR = 2m temp. warmer than 4K

compared to climatology.

SUMMER

Meteorologisk institutt met.no

TEPS

LAMTEPS (PB)

LAMTEPS (NP)

THR = 2m temp. warmer than 4K

compared to climatology.

SUMMER

Meteorologisk institutt met.no

Targeted vs. not targeted perturbations

• Motivation: – TEPS perturbations initially small inside the

HIRLAM integration area.– EPS perturbations more uniformly spread

around the hemisphere

• Two different approaches:– Perturbing using TEPS -> LAMTEPS– Perturbing using “operational” EPS -> LAMEPS

• Test period in September 1999 (23 – 30)• Lateral boundary conditions perturbed

Meteorologisk institutt met.no

Meteorologisk institutt met.no

EPS

TEPS

LAMEPS

LAMTEPS

THR = 15mm/24hTOTAL PRECIPITATION

Verification time: 66h

Autumn

Meteorologisk institutt met.no

EPS

TEPS

LAMEPS

LAMTEPS

THR = 15mm/24hTOTAL PRECIPITATION

Verification time: 90h

Autumn

Meteorologisk institutt met.no

Two case studies from Sept. 1999 with heavy precipitation

• Case 1: 42 hours forecast from 30 September 1999

• Case 2: 42 hours forecast from 25 September 1999

• “super-observations” representative of the HIRLAM grid constructed (Ghelli and Lalaurette, 2000)

• Kriging method used to obtain estimated area precipitation from point observations

Meteorologisk institutt met.no

Case 1: 24h accumulated precipitation, valid at 2 October 1999

AnalysisHIRLAM 50km

(MSLP)

Meteorologisk institutt met.no

Observed 24haccumulatedprecipitation