assimilation of radar information in the alpine model of meteoswiss

17
Radar in aLMo Assimilation of Radar Information in the Alpine Model of MeteoSwiss Daniel Leuenberger and Andrea Rossa MeteoSwiss

Upload: jaser

Post on 03-Feb-2016

19 views

Category:

Documents


0 download

DESCRIPTION

Assimilation of Radar Information in the Alpine Model of MeteoSwiss. Daniel Leuenberger and Andrea Rossa MeteoS wiss. Introduction. Radar information is gaining importance in mesoscale data assimilation Latent Heat Nudging (LHN): Assimilation method for precipitation information - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: Assimilation of Radar Information in the  Alpine Model of MeteoSwiss

Radar in aLMo

Assimilation of Radar Information in the Alpine Model of MeteoSwiss

Daniel Leuenberger and Andrea RossaMeteoSwiss

Page 2: Assimilation of Radar Information in the  Alpine Model of MeteoSwiss

Dani

el.L

euen

berg

er@

Met

eoSw

iss.c

hSR

NWP

– COS

T-71

7 Lis

bon,

8.O

ctob

er 2

003 2

Radar in aLMoIntroductionRadar information is gaining importance in mesoscale data assimilationLatent Heat Nudging (LHN): Assimilation method for precipitation informationTrigger model precipitation where radar detects precipitation (heating), supress it elsewhere (cooling)4DDA, yet computationally very efficientConceptionally simpleAssimilation of non-prognostic variables not straight forwardHeuristic approach of weighting observations and model

Page 3: Assimilation of Radar Information in the  Alpine Model of MeteoSwiss

Dani

el.L

euen

berg

er@

Met

eoSw

iss.c

hSR

NWP

– COS

T-71

7 Lis

bon,

8.O

ctob

er 2

003 3

Radar in aLMoRadar ObservationsSwiss radar network: 3 C-Band Doppler RadarsBest estimate of surface rain: preprocessed (e.g. clutter reduction, vertical profile corrections)Resolution: 2 x 2 km2, 5 min

Radar Quality Map

Page 4: Assimilation of Radar Information in the  Alpine Model of MeteoSwiss

Dani

el.L

euen

berg

er@

Met

eoSw

iss.c

hSR

NWP

– COS

T-71

7 Lis

bon,

8.O

ctob

er 2

003 4

Radar in aLMoReal Case StudyCase

System of severe convection over SwitzerlandTriggered around 22h30 UTC over the Massif CentralDevelopment ahead of weak cold frontModerately unstable environment as observed by the Swiss radiosonde Payerne at 00UTC (CAPE ~250 J/kg)Strong wind shear (~ 30 m/s at 6000m)

SimulationsOperational Alpine Model (aLMo) of MeteoSwiss (x=7km)Convection ParametrizationStarted 21.8.00 00 UTC from GME of DWDCTRL (no forcing)LHN during 6hLHN+ during 3h, free forecast afterwards

Page 5: Assimilation of Radar Information in the  Alpine Model of MeteoSwiss

Dani

el.L

euen

berg

er@

Met

eoSw

iss.c

hSR

NWP

– COS

T-71

7 Lis

bon,

8.O

ctob

er 2

003 5

Radar in aLMo

010203040506

CTRL LHN Radar

Case Study of the 21.8.2000 StormHourly Sums of Precipitation: Forcing during 6h

Page 6: Assimilation of Radar Information in the  Alpine Model of MeteoSwiss

Dani

el.L

euen

berg

er@

Met

eoSw

iss.c

hSR

NWP

– COS

T-71

7 Lis

bon,

8.O

ctob

er 2

003 6

Radar in aLMo

01020304

1h Free Forecast

05

2h Free Forecast

06

3h Free Forecast

Case Study of the 21.8.2000 StormHourly Sums of Precipitation: Forcing during 3h

CTRL LHN+ Radar

Page 7: Assimilation of Radar Information in the  Alpine Model of MeteoSwiss

Dani

el.L

euen

berg

er@

Met

eoSw

iss.c

hSR

NWP

– COS

T-71

7 Lis

bon,

8.O

ctob

er 2

003 7

Radar in aLMoFindings IModel is able to assimilate radar observationsGood impact in analysis, sfc winds in line with observationsSome impact in free forecast up to 03hModel loses information quickly, i.e. storm dies too earlyWhy is the model not able to maintain storm?

Environment not representative?Model resolution ?

Try to find reasons by means of idealized simulations

Page 8: Assimilation of Radar Information in the  Alpine Model of MeteoSwiss

Dani

el.L

euen

berg

er@

Met

eoSw

iss.c

hSR

NWP

– COS

T-71

7 Lis

bon,

8.O

ctob

er 2

003 8

Radar in aLMoIdealized SimulationsSetup

Environment from Payerne sounding of 21.8.00 00UTCFine mesh (x = 1km), no CPS, no soil model, no radiationTrigger convection with warm bubble: No storm development

wind shearPayerne profile

KW profile

Payerne sounding

KW sounding

Klemp Wilhelmson EnvironmentLarge amount of CAPE (~ 1200 J/kg)Moderate wind shearFavorable for splitting supercell storms

Page 9: Assimilation of Radar Information in the  Alpine Model of MeteoSwiss

Dani

el.L

euen

berg

er@

Met

eoSw

iss.c

hSR

NWP

– COS

T-71

7 Lis

bon,

8.O

ctob

er 2

003 9

Radar in aLMoOSSE SetupReference run

Convection initiated with warm bubbleModel sfc rain serves as „artificial radar observations“

LHN AnalysisSame environment as reference runNo warm bubble initiationLHN during 3h (artificial rain rates from reference run)

LHN ForecastLHN during first 30, 60 min Free run afterwards

Page 10: Assimilation of Radar Information in the  Alpine Model of MeteoSwiss

Dani

el.L

euen

berg

er@

Met

eoSw

iss.c

hSR

NWP

– COS

T-71

7 Lis

bon,

8.O

ctob

er 2

003 10

Radar in aLMoInsertion Frequency of Precipitation Input

LHN linearly interpolates between subsequent observationsExamine relevance of insertion frequency t to LHN Analysis

t = 10minlinear interpolation

t = 1min

Page 11: Assimilation of Radar Information in the  Alpine Model of MeteoSwiss

Dani

el.L

euen

berg

er@

Met

eoSw

iss.c

hSR

NWP

– COS

T-71

7 Lis

bon,

8.O

ctob

er 2

003 11

Radar in aLMoLHN Analysis (LHN during 3h)

“OBS“ t = 10min

t = 4mint = 1min

Page 12: Assimilation of Radar Information in the  Alpine Model of MeteoSwiss

Dani

el.L

euen

berg

er@

Met

eoSw

iss.c

hSR

NWP

– COS

T-71

7 Lis

bon,

8.O

ctob

er 2

003 12

Radar in aLMoDomain Sum of LH Nudging Increment

Page 13: Assimilation of Radar Information in the  Alpine Model of MeteoSwiss

Dani

el.L

euen

berg

er@

Met

eoSw

iss.c

hSR

NWP

– COS

T-71

7 Lis

bon,

8.O

ctob

er 2

003 13

Radar in aLMoLHN Forecast (t = 1min)

“OBS“

Free forecast after 30 min

Analysis (LHN during 3h)

Free forecast after 1h

Page 14: Assimilation of Radar Information in the  Alpine Model of MeteoSwiss

Dani

el.L

euen

berg

er@

Met

eoSw

iss.c

hSR

NWP

– COS

T-71

7 Lis

bon,

8.O

ctob

er 2

003 14

Radar in aLMoSensitivity to Horizontal Grid Spacing

Analysis (LHN during 3h)Free forecast after 1h

x = 2km

x = 5km

Page 15: Assimilation of Radar Information in the  Alpine Model of MeteoSwiss

Dani

el.L

euen

berg

er@

Met

eoSw

iss.c

hSR

NWP

– COS

T-71

7 Lis

bon,

8.O

ctob

er 2

003 15

Radar in aLMoFindings IILHN capable of analysing and initiating supercell stormGood temporal sampling of the observed phenomena is importantRepresentative large-/mesoscale environment importantEven a poorly resolved forcing is able to initiate and maintain storm evolution in appropriate environmentSupercell storm very stable dynamics: are findings ‚portable‘ to other situations?

Page 16: Assimilation of Radar Information in the  Alpine Model of MeteoSwiss

Dani

el.L

euen

berg

er@

Met

eoSw

iss.c

hSR

NWP

– COS

T-71

7 Lis

bon,

8.O

ctob

er 2

003 16

Radar in aLMoOutlookReal-case study

Reduction of grid-size to 2kmStudy impact of errors in radar dataMore cases

Idealized OSSESensitivity of vertical forcing distributionAssimilation of model 3D latent heating fieldsAssimilation of horizontal windsConsider case which is less driven by dynamics

Page 17: Assimilation of Radar Information in the  Alpine Model of MeteoSwiss

Dani

el.L

euen

berg

er@

Met

eoSw

iss.c

hSR

NWP

– COS

T-71

7 Lis

bon,

8.O

ctob

er 2

003 17

Radar in aLMo

Thank you for your attention !