AEMET-γSREPS: convection-permitting LAM-EPS [Presentación]

(1)

AEMET-

γ

SREPS

Convection-permitting LAM-EPS

VI Simposio Nacional de Predicción 17/19

th

_{-September 2018}

Servicios centrales AEMET

(2)

Outline

• Who are we ?



AEMET predictability group

• What is

AEMET-

γ

SREPS

?



20-members

non-hydrostatic convection-permitting

EPS

• Designing

AEMET-

γ

SREPS



Multi-boundary conditions + multi-mesoscale NWP models

• γ

SREPS

in operations



EcFlow suite gSREPSoper

• γ

SREPS

verification



Subjective and objective

• Current and future

γ

SREPS

developments

(3)

(4)

• Since 2002 an small core group working on Limited Area

(LAM) Ensemble Prediction

Systems

(EPS) depending on Research Deparment

• Members of

HIRLAM

-GLAMEPS-HarmonEPS and involved in several projects:

EUMETNET SRNWP EPS II, PreFlexMS, COASTEPS, EPS eCOST action, etc., and

collaborations on EPSs with

IPMA

-AEMET, AROME-EPS

MétéoFrance

, etc.

AEMET-

_γ

SREPS Predictability group

José Antonio García-Moya

(retired on May 2018)

BCsand experiments on all aspects of EPS

Alfons Callado

γ

SREPS operational development and model error: SPPT

Pau Escribà

Assimilation LETKF

Mauri Martínez

(CLARITY project)

γ

SREPS support and verification

Marc Compte

(scholarship)

γ

SREPS support

Jesús Rodríguez

(EUMETNET SRNWPEPSII project) calibration

David Gil

(collaboration) web page

David Quintero

machine learning

and

γ

SREPS support

(5)

• Full participation in

VI Simposio Nacional de Predicción

: 5 presentations

• AEMET-

γ

SREPS in other presentations:

• El episodio de Oliva-Gandía del 3-4 noviembre de 1987: récord nacional de precipitación en 24 horas (C. Santos et al.) • Episodio de cizalladura del 10-9-2017 en el aeropuerto de Barcelona (A. Méndez et al.)

• Coastal trapped disturbancesen el litoral catalán: un peligro aeronáutico (S. González et al.)

• Episodio de tiempo severo con elevado impacto en Cataluña del 18 de octubre de 2017 (R. Pascual and G. Cuevas)

AEMET-

_γ

SREPS Predictability group

José Antonio García-Moya

Super-CHAIRMAN

Alfons Callado

“AEMET-

γ

SREPS: convection permitting LAM-EPS”

Pau Escribà

“El

AEMET-γSREPS en un caso de convección por convergencias en el Mediterráneo”

Mauri Martínez

“nowcRadiation – Software de nowcasting para la predicción de la radiación solar”

David Gil

“Presentación de la aplicación web AEMET-

γ

SREPS”

David Quintero

“Postproceso estadístico de modelos meteorológicos

con Machine Learning: aplicaciones al γSREPS y al HARMONIE”

(6)

What is

(7)

Turia 1957-10-14 Flash flood>350mm/d

La Safor 2007--Cut-off low>450mm/d CVal 2009-09-27/29 Cut-off low>300mm/d Cádiz 2009-12-24 Tornado Algeciras 2012-03-06 Storm > 35 mm/h

Montserrat 2000-06-09 Flash-floods > 200mm

Vista Alegre 1999-09-07 Tornado

Tous 1982-10-20 Flash-floods>1000mm Cantábrico 2011-05-30

CAP gale (Galerna)

Spain

Severe

Weather

Spain

Renewable

Energies

AEROgenerators

Serra De Burgo (Orense)

Valladolid 2011-05-30 Hailstorm

(8)

AEMET-

γ

SREPS system

NMMB WRF ARW

HARMONIE-ALARO

HARMONIE-AROME Multi-BCs

Multi-NWP

CMC / GEM JMA / GSM

MF / ARPÈGE NCEP / GFS

ECMWF / IFS

We are now updating

WRF-ARW from 3.6.0 on

sigma levels to 3.9.1.1 on

hybrid !!!

• 20-members

non-hydrostatic convection-permitting

EPS

• Since April 2016 daily running at

00 and

12 UTC

up to

36 hours

(9)

• Multi-model

• HARMONIE-AROME

• HARMONIE-ALARO

• WRF-ARW (NCAR)

• NMMB (NCEP)

AEMET-

γ

SREPS

1 INITIAL

CONDITIONS

HH+00

HH+24

HH+48

X X

X

K

P

D

t

X

_

_

_



Initial conditions

and

Multi-BCs

• ECMWF – IFS

• NCEP – GFS

• MétéoFrance – ARPÈGE

• CMC – GEM (Canadian)

• JMA - GSM (Japanese)

2 BOUNDARY

CONDITIONS

3 MODEL

ERROR

• Developing a

convection-permitting

LAM-EPS

• 3 sources of uncertainties

2019

LETKF

assimilation

25 members

¿ GEM-LAM ?

(10)

Designing

(11)

Horizontal resolution in

GRIB

files

HARMONIEs

lambert

‘NLON' => '576’,

'NLAT' => '480',

‘EZONE’ => ’11’

'GSIZE' => '2500.',

WRF-ARW

map_proj = ‘lambert',

e_we = 566,

e_sn = 470,

dx = 2500,

dy = 2500,

NMMB

map_proj = 'rotated_llb',

e_we = 568,

e_sn = 472,

dx = 0.022545,

dy = 0.022500,

 NMMB bigger area than HARMONIE to interpolate to common

Lambert area without missing values

• Intended to

run each NWP

with the same /

very

similar

internal

GRID

• Exactly

the

(12)

Vertical levels consistence

between models

• AROME

and

ALARO

•

65 hybrid

sigma-pressure

layers

• No top

• WRF-ARW

:

•

72 sigma levels



hybrid on

3.9.1. version

• Top 40 hPa

• NMMB

•

72 hybrid levels

• Top 40 hPa



Consistency AT LOW LEVELS into the

PBL



The very close vertical levels

distribution !!!

22

(13)

NWP models’ settings

HARMONIEs

WRF-ARW

NMMB

AROME

physics

ALARO

physics

65 Hybrid

sigma-pressure

vertical

levels

60 s

time step

ARW

dynamical core

66 72 sigma (ETA)

hydrostatic-pressure levels

up

to 40 hPa [

where it is 64 HARMONIE vertical level

]

12 s

time step

NMM

dynamical core

66 72 Hybrid

sigma-pressure

up to 40 hPa

5.625 s

time step

Lambert Conformal Conic projection:

lon -2.5º / lat 40.0º centre

**565 * 469**

grid-points

Rotated lon-lat B-grid:

lon -2.5º / lat 40.0º centre

**568 * 472**

grid-points Calling radiation every 15 minutes

8 LBC relaxation points around grid area

Summary of models in

γ

SREPS:

4 NWP MODELS

It has been intended to integrate both NWP models with the closer possible settings in order to be the comparison the more fairly possible.

X

(14)

(15)

• EcFlow

suite management

• 00 and 12 UTC cycle over

IBERIA_2.5

domain

Downloading

NCEP / GFS

BCs

20 members

EPS

γ

SREPS at

ECMWF

Cray XC40

Running

NMMB

+

NCEP/GFS BCs

UPP

NCEP Unified Post-processor

HARMONIE

AROME

+

ECMWF/IFS BCs

HARMONIE

ALARO

+

ECMWF/IFS BCs

Finishing !!!

Cray XC40 ECMWF

(16)

γ

SREPS

Objective

VERIFICATION

AEMET-

γ

SREPS

is always the better LAM-EPS !!!

(17)

• AEMET-

γ

SREPS



without assimilation

• HARMONIE-LETKF



As HarmonEPS, but with LETKF

• HARMONIE-3DVAR-EPS

 HarmonEPS



3DVAR-NMC + EC-SLAF

• HARMONIE-3DVAR-LAG



8 mbrs det. HARMONIE

AEMET-

γ

SREPS

is always the better LAM-EPS !!!

(18)

γ

SREPS

Subjective

VERIFICATION

Not significant conclusions … based on

only few cases !!! See

Pau Escribà

presentation

(19)

Qualitative verification results

Stratiform precipitation

(20)

Maresme

case study:

low predictability

HARMONIE 40

HARMONIE 38

γ

SREPS PROB > 5mm

γ

SREPS max AccPcp

AEMET-

γ

SREPS

(21)

Current and future

γ

SREPS

(22)

Foreseeable future work

plan: 2018-2020

γ

SREPS

“scientific” developments:

• Assimilation:

LETKF

without discarding

3DVAR…

[



Pau

]

• GNSS+

RADAR

assimilation [



Jana Sánchez

]

• 25 members



Including the 5th

mesoscale

convection-permitting

NWP model:

¿Canadian

GEM-LAM

? [



Alfons

]

(23)

Foreseeable future work

plan: 2018-2020

γ

SREPS

“technical” developments:

• γ

SREPS

IBERIA

at

ECMWF

systems:

• Time critical application

level 2 [



Pau

]

• IBERIAxxm_2.5

bigger domain (~better organized

convection) and up to

72 hours



2020 Bologna

HPCF supercomputer

• γ

SREPS

CANARIAS_2.5

00 and 12 UTC at

AEMET

systems (

BULL-ATOS

):

• Currently under implementation [



Alfons

+

David Q.

]

• ¿ “explotación” future mamagement ? [



Sergio Cotera

]

(24)

Foreseeable future work

plan: 2018-2020

γ

SREPS

user “products” developments:

• Probabilistic forecast products:

• Forecasters

WEB page

[



David G.,

see presentation

]

−

With one click:

– EPSgrams

– Probabilistic vertical profiles – Probabilistic cross-sections

• PANEL-VIM

integration [



ATAP Marcos G.

]

• Specific socio-economic

impact

products:

• AEROgrams

ENAIRE

[



Mariona Pons

&

Alejandro Méndez

]

−

¿

Wind shear

products ?

• Wind

/

solar

renewable energy sector:

−

E.g. probabilistic solar

GHI

+

DNI

for

REE

[¿



Mauri

?]

Barcelona airport

(25)

Happy AEMET-γSREPSing !!!

[email protected]

AEMET

-

γ

SREPS will be shortly available for AEMET forecasters !!!

For your case studies, ask for AEMET-

γ

SREPS !!!

VI Simposio Nacional de Predicción 17/19

th

_{-September 2018}

Servicios centrales AEMET

AEMET Predictability Group