The Ensemble Systems of the

1

• The Ensemble Systems of the
• COSMO Consortium
• Chiara Marsigli, Andrea Montani, Tiziana
  Paccagnella
• ARPA-SIM
• The Hydro-Meteorological Service
• of the
• Emilia-Romagna Region Environmental Agency

2
COSMO Ensemble Prediction Systems

• COSMO-LEPS
• 2-5 days 10 km hor. res.
• Operational as ECMWF Time Critical Application
• COSMO-SREPS
• Short Range 10 km hor. res.
• Under development
• EELMK
• DWD project
• 2.8 km hor. res.

3
COSMO-LEPS

• It is a Limited-area Ensemble Prediction System
  (LEPS), based on the Lokal Modell of the COSMO
  Consortium
• It has been designed to improve the capability of
  forecasting severe local events in the time range
  48 C132 hours by performing a dynamical
  downscaling of some selected members of ECMWF EPS

4
The new COSMO-LEPS suite _at_ ECMWFsince February
2006
16 Representative Members driving the 16 Lokal
Modell integrations (weighted according to the
cluster populations) Employing either Tiedtke
or Kain-Fristch moist convection scheme (randomly
choosen)
4 variables Z U V Q
3 levels 500 700 850 hPa
d5
Cluster Analysis and RMs identification
Cluster Analysis and RM identification
2 time steps
European area
clustering period
Complete Linkage

• suite running as a time-critical application
  managed by ARPA-SIM
• ?x 10 km 40 ML
• LM_3.17 since 1/2/2006
• fc length 132h
• Computer time (4 million BU for 2006) provided
  by the COSMO partners which are ECMWF member
  states.

COSMO-LEPS clustering area
COSMO-LEPS Integration Domain
5

• Start 20 UTC
• 2 trances 96 Tasks HPCD (ltHPCE) IBM PW5 for
  each run.
• 45 Mins for each run 306x258x40

6
The new COSMO-LEPS suite _at_ ECMWFsince February
2006

• Core System
• 16 perturbed LM runs (ICs and 3-hourly BCs from
  16 EPS members) to generate probabilistic output
  (start at 12UTC ?t 132h)

• Additional Components
• 1 deterministic run (ICs and 3-hourly BCs from
  the high-resolution deterministic ECMWF forecast)
  to assess the relative merits between
  deterministic and probabilistic approach (start
  at 12UTC ?t 132h)
• 1 proxy run (ICs and 3-hourly BCs from ECMWF
  analyses) to downscale ECMWF information (start
  at 00UTC ?t 36h).

7
COSMO-LEPS products
New products Ensemble Mean and Ensemble
Standard Deviation for Z700 - T850 MSLP -
T2M - UV10M - TP fc0h to fc132h every 3
hours EM and ES are calculated without
considering the weights relative to each
integration.

• Products delivered at about 1 UTC to most of
  COSMO weather services, to Hungary (case studies)
  and to the COPS community (field campaign).

• about 4 GB a day

8
COSMO LEPS CLONES

• The Alpine Suite ?MAP D Phase Project
• (Demonstration of Probabilistic Hydrological
  and Atmospheric Simulation of flood Events in the
  Alpine region) a WWRP Forecast Demonstration
  Project
• The Northern Europe Suite ? FP6 Eurorisk-Preview
  (PREVention, Information and Early Warning)
  Windstorm Project

9
ALPINE suite (on SPCOLEPS billing units)

• Experimental suite from 15 July 2005 to 31
  January 2006.
• Same configuration as the operational COSMO-LEPS
  (red), but both clustering and integration domain
  are centred over the Alps (blue).
• Products were not disseminated, but saved at
  ECMWF.

• 10 integrations
• hor. Res. 10 km
• 32 ML
• 132h forecast range
• comparison between OPE and ALP suites on a
  case-study basis and over the full period (about
  200 days).

10
Clone of COSMO-LEPS suite over NW Europe

• Generation of a new limited-area ensemble
  forecast system for the prediction of wind and
  wind gusts over North-Western Europe.
• Aim provision of site-specific wind forecasts
  (wind speed, wind direction other variables to
  diagnose wind gust) to be post-processed and
  combined with other products by UK Met. Office.
• Point forecasts are calculated over 197 stations.
• Similar configuration as the operational
  COSMO-LEPS (red), but both clustering and
  integration domain are shifted northwards (blue).

Experimental suite started on September
2006. Computer time provided by an ECMWF special
project between ARPA-SIM and UK Met Office.
11
Future plans

• On ectmp, store ECMWF EPS bc and ics (all EPS
  members at both 00Z and 12Z!) for the last 3
  months to rerun recent case studies.
• Work on coding of synthetic radar and satellite
  images so as to archive them at ECMWF on MARS.
• (Start to) Implement verification at ECMWF.
• Back archiving on MARS of past runs (from
  5/11/2002 onwards).
• Switch on the Alpine suite for the MAP D-Phase
  campaign (June-November 2007)?
• Comparison with COSMO-SREPS

12
COSMO-SREPS

• COSMO-LEPS mainly designed for the 2-5 days range
• The spread is coming only from EPS initial and
  boundary conditions (uncertainties related to
  larger scale analysis errors )

• COSMO-SREPS has been thought as a limited-area
  ensemble system for the short-range
• add more spread in the short-range
• initial and boundary condition from a Multi-Model
  Multi-Boundary ensemble (INM SREPS)
• take into account also small scale uncertainty
  related to LM errors

13
System DesignICs and BCs multi-model approach

• a MUlti-Model MUlti-Boundaries ensemble is
  operational at INM (Spain) 5 different
  limited-area models (HIRLAM, HRM, MM5, UM, LM)
  are nested on 4 different global models (IFS,
  GME, AVN, UM) gt 20 members ensemble (25 km
  horizontal resolution)

Courtesy of INM
IFS LM 25km
GME LM 25km
AVN LM 25 km
UM LM 25 km
14
Model Perturbations

• Schemes (physics and numerics) Some schemes
  have been selected to be perturbed Tiedtke,
  Kain-Fritsch, 2-time-level, Runge-Kutta
• Parameters
• a first selection of the parameters has been
  done
• new tuneable parameters will be tested
• Surface forcings test a technique for the
  perturbation of surface parameters

Implementation of an ETKF (collaboration with
UKMO) to perturb initial conditions at high
resolution
15
System DesignICs and BCs multi-model approach
IFS ECMWF global
LM at 25 km on IFS
GME DWD global
P1 control P2 conv. scheme (KF) P3 turb.
parameter 1 P4 turb. parameter 2
LM at 25 km on GME
by INM Spain
UM UKMO global
16 LM runs at 10 km
LM at 25 km on UM
AVN NCEP global
LM at 25 km on NCEP
16
13 May 12-18 UTC
COSMO-LEPS PROB. MAPS fc. range 0-6 6-12 12-18 h
gt 10mm/6h
COSMO-SREPS PROB. MAPS fc. range 12-18 18-24
24-30 h
gt 10mm/6h
17
COSMO-SREPS fc. range 24-30
ecmwf
gme
avn
ukmo
14 May 0-6 UTC
18
ensemble spread
T 2m
12/05 12UTC
13/05 00UTC
13/05 12UTC
19
member behaviour (father)
20
intra-group distance whole domain
21
intra-group distance northern Italy
22
1-month test

• run of COSMO-SREPS with model perturbations for
  1 month of Autumn 2006
• chosen dates 13-18 September, 30 September, 1-7
  October, 13 October, 18-20 October, 27-31
  October, November
• provide b.c. on model levels on a sub-domain for
  EELMK (DWD) tests
• statistical analysis of the system
• objective verification comparison with
  COSMO-LEPS and PEPS
• test the use of the system for the determination
  of a flow-dependent B-matrix to be used in the
  1D-Var of satellite data with LM

23
1-month test model pert. with the same father
IFS ECMWF global
P1 control (ope) P2 conv. scheme (KF) P3 turb.
parameter 1 P4 turb. parameter 2 P5
LM at 25 km on IFS
by INM Spain
10 LM runs at 10 km
24
Future Plans

• implementation of an ETKF (collaboration with
  UKMO) to perturb initial conditions at high
  resolution
• MAP D-PHASE testing period (summer-autumn 2007)
  
• only a sub-period?
• not real time!
• perturbations selected on the basis of the
  1-month test
• verification -gt availability of observations
• comparison with the other available mesoscale
  ensemble systems

25
Thank you !
26

• TIGGE
• The THORPEX Interactive Grand Global Ensemble
• Philippe Bougeault
• Co-chair, GIFS-TIGGE WG
• TIGGE LAM
• Tiziana Paccagnella
• Chair, TIGGE LAM Panel

27
WMO/WWRP/RDP/ THORPEX THe Observing system
Research and Predictability Experiment

• TIGGE, the THORPEX Interactive Grand Global
  Ensemble, is a key component of THORPEX a World
  Weather Research Programme to accelerate the
  improvements in the accuracy of 1-day to 2-week
  high-impact weather forecasts for the benefit of
  humanity.
• The first TIGGE workshop was held at ECMWF from 1
  to 3 March 2005 and the full workshop report is
  available on the THORPEX Web site and has been
  published in the WMO series WMO/TD-No. 1273
  WWRP/THORPEX No. 5.

28
Key objectives of TIGGE discussed at the Reading
workshop

• An enhanced collaboration on development of
  ensemble prediction, internationally and between
  operational centres and universities
• New methods of combining ensembles from different
  sources and of correcting for systematic errors
  (biases, spread over-/under-estimation)
• A deeper understanding of the contribution of
  observation, initial and model uncertainties to
  forecast error

29
Key objectives of TIGGE (continued)

• A deeper understanding of the feasibility of
  interactive ensemble system responding
  dynamically to changing uncertainty (including
  use for adaptive observing, variable ensemble
  size, on-demand regional ensembles) and
  exploiting new technology for grid computing and
  high-speed data transfer.
• Test concepts of a TIGGE Prediction Centre to
  produce ensemble-based predictions of high-impact
  weather, wherever it occurs, on all predictable
  time ranges
• The development of a prototype future Global
  Interactive Forecasting System

30
Development in two phases

• Phase-1 data will be collected in near-real
  time (via internet ftp) at a small number of
  central TIGGE data archives. This can be
  implemented now at little cost and could handle
  the estimated 200 GB per day data volumes with
  current network and storage capabilities
• Phase-2 data archives will be distributed over a
  number of repositories, instead of all being held
  centrally, but efficient and transparent access
  to users will be maintained. This is a more
  flexible solution with the potential to eliminate
  routine transfers of large data volumes. But this
  will require substantial software development
  over a number of years, in coordination with the
  WMO Information System, and will require
  additional funding

31
Development in two phases

• Phase-1 data will be collected in near-real
  time (via internet ftp) at a small number of
  central TIGGE data archives. This can be
  implemented now at little cost and could handle
  the estimated 200 GB per day daa volumes with
  current network and storage capabilities
• Phase-2 data archives will be distributed over a
  number of repositories, instead of all being held
  centrally, but efficient and transparent access
  to users will be maintained. This is a more
  flexible solution with the potential to eliminate
  routine transfers of large data volumes. But this
  will require substantial software development
  over a number of years, in coordination with the
  WMO Information System, and will require
  additional funding

32
TIGGE infrastructure Phase 1

• Data collected in near-real time (via internet
  ftp) at central TIGGE data archives
• Can be implemented now at little cost
• Can handle current data volumes (estimated 200
  Gb/day) within available network and storage
  capabilities

Predictability science
Real-world applications
NHMS
academic
End user
TIGGE Centre A
TIGGE Centre B
EPS 1
EPS 2
EPS n
33
TIGGE infrastructure Phase 2
Predictability science
Real-world applications

• Data distributed over several repositories
• But keep efficient and transparent user access
• Flexible minimise data transfers
• Needs substantial software development
• Coordination with WMO Information System
• Requires additional funding

NHMS
academic
End user
Portal to distributed (virtual) archive
EPS 1
EPS 2
EPS n
34
TIGGE Partners

• Archive and Distribution Centres CMA, ECMWF,
  NCAR (more in phase 2!)
• Data providers
• BMRC (Bureau of Meteorology Research Centre)
• CMA (China Meteorological Administration)
• CPTEC (Centro de Previsao de Tempo e Estudos
  Climaticos),
• ECMWF
• FNMOC (Fleet Numerical Meteorology and
  Oceanography Center)
• JMA (Japan Meteorological Agency),
• KMA (Korea Meteorological Administration)
• Meteo-France,
• MS Canada (Meteorological Service of Canada)
• NCEP
• UKMO
• TIGGE Web site ECMWF
• Meta-data centre NCAR
• Verification Web site JMA

35
The GIFS-TIGGE WG

• Philippe Bougeault ECMWF
• David Richardson UKMO
• Barbara Brown NCAR
• Chen De Hui CMA
• Beth Ebert BMRC
• Martin Ehrendorfer Innsbruck
• Tom Hamill NOAA CDC
• Mark Roulston Penn State
• Pedro Silva Dias CPTEC

• Warren Tennant Africa
• Zoltan Toth NCEP
• Laurie Wilson MSC
• Yoshiaki Takeuchi JMA
• Hee Dong Yoo KMA

36
Data to include in priority

• Ensemble forecasts generated routinely (often
  operationally) at different centres around the
  world. This is the core data of the TIGGE
  archive. The total daily data volume is expected
  to be around 200GB, based on a preliminary list
  of required parameters developed at the workshop.
  
• Observational data and existing datasets
  including re-analyses and re-forecasts
• Additional special datasets generated during the
  TIGGE project for specific research and
  applications.

37
Content - formats

• GRIB2 agreed by all partners
• Units, names of fields, accumulation periods,
  etc.. Will be identical for all data providers
• Grids
• Data Providers will be asked to provide data on
  grids of their own choosing, which are as close
  as possible to the native grid employed to carry
  out the predictions
• Data Providers should ensure that appropriate
  software is available to the Data Centres to
  enable users to interpolate data to
  latitude/longitude grids and locations of their
  choosing
• Data Providers should ensure that when revisions
  to their systems are made the interpolation
  software will still work.

38
User access Data retrieval

• It is unavoidable that in Phase 1 each Archive
  Centre will provide data through a different user
  interface
• The user interfaces will allow to download any
  set of data (sub-setting and sub-sampling
  facilities) at single points or on a regular
  lat-lon grid defined by the user
• Automatic regular requests will be possible

39
TIGGE-LAM

• LAM-EPS are an important component of the future
  GIFS
• TIGGE-WG discussed roadmap towards TIGGE-LAM at
  its March 2006 meeting
• The perceived priorities are
• To facilitate exchange of initial and boundary
  conditions between ALL global ensembles and ALL
  LAM ensembles by agreed standard data formats and
  contents
• To agree content and standard format of LAM
  ensemble data for wide exchange and evaluation
• To exchange meta-data through a common Internet
  site
• To explore which Centres are willing to take a
  leading role in offering
• routine boundary conditions,
• elocatable LAM ensembles
• archiving services
• Dissemination
• etc.. for a TIGGE-LAM
• To define Scientific Issues
• To agree on verification and calibration
  methodologies and related implementation
  strategies

40
Actions towards TIGGE-LAM

• A panel of experts is being appointed work by
  email initially, then possibly a physical meeting
  if funding can be secured
• Tiziana Paccagnella (ARPA) has accepted to chair
  the panel and the panel work will start in Autumn
  2006
• The panel will report to TIGGE-WG

41
TIGGE LAM Panel - Terms of Reference

• 1. The Panel should propose to the GIFS-TIGGE WG
  a possible structure of the TIGGE-LAM component.
• 2. The Panel should maintain the necessary
  liaison with already existing LAM EPS initiatives
  and with related new projects. A good monitoring
  of existing operational LAMs should also be
  provided by the Panel to evaluate possible
  contributions to TIGGE-LAM on specific
  geographical regions, in coordination with the
  THORPEX regional committees.
• 3. The Panel should encourage a coordinated
  approach to LAM EPS and recommend solutions to
  facilitate the development of collaboration for
  the mutual benefit of WMO members.
• 4. The Panel should formulate recommendations
  regarding the formats of global ensemble data
  provided as initial and boundary conditions for
  limited-area ensembles, keeping in mind the GIFS
  objective to facilitate the exchange of
  information. If possible, universal formats
  should be sought. The panel should formulate
  proposals to facilitate the interoperability
  (different LAMs driven by different GCMs) of the
  different modelling systems contributing to
  TIGGE.
• 5. The Panel should formulate proposals for the
  creation of a coordinated distributed archive of
  limited-area ensemble forecasts. If possible
  formats for such an archive should be fully
  compatible with the formats already agreed for
  the archive of global ensemble forecasts under
  TIGGE

42
TIGGE LAM Panel - Terms of Reference (2)

• 6. The Panel should survey the interest from
  various centres to act as archiving centres for
  limited-area ensemble forecasts, and more
  generally as resource centres for TIGGE-LAM.
• 7. The Panel should contribute to the definition
  of scientific issues related to LAM EPS and
  particularly to TIGGE-LAM and should propose
  specific initiatives to explore these issues in a
  focused and organized way to advance LAM EPS.
• 8. The Panel, in a close coordination with the
  GIFS-TIGGE WG, should propose guidelines as
  regards LAM EPS validation, calibration and
  combination.
• 9. The panel will conduct its business initially
  by e-mail. Meetings will be arranged, pending
  available resources from the THORPEX Project
  Office or other sponsors, such as GEO.
• 10 The panel will report to the GIFS-TIGGE-WG (in
  charge of developing TIGGE under the THORPEX Core
  Steering Committee). The Chair of the TIGGE-LAM
  panel will be invited to GIFS-TIGGE WG meetings.

43
Actions towards TIGGE-LAMList of issues to be
tackled and discussed

• Interface to couple Global Ensembles (models) to
  LAMS
• LAM output and formats
• LAM archive
• Calibration and Verification
• Scientific topics to be addressed and
  investigated
• WEB SITE
• Links with other projects and bodies
• Funding
• Others.

44
Actions towards TIGGE-LAM

• TIGGE-WG will urge the THORPEX Community to work
  on scientific issues of the coupling between
  global and LAM ensembles
• Assess impacts of mismatch between the physical
  parameterizations of global and regional models
• Assess the need to define bias corrections
• Assess impacts of large mismatch in horizontal
  resolution
• What is the best way of generating perturbations
  for LAM EPS?

45
TIGGE-LAM output and formats

• The starting point is the list of parameters
  defined by the TIGGE WG. TIGGE-LAM can of course
  add or withdraw data to this list .
• Recommended Fields must be produced with the
  same unit and keeping the same philosophy as
  regards cumulating and averaging periods (e.g.
  precipitation must be cumulated from the forecast
  start). The homogeneity between data from Global
  and LAM ensemble will make easier (feasible)
  verification and comparison.

46
TIGGE-LAMLinks with other projects and bodies

• TIGGE GIFS (in Primis)
• SRNWP gt Cooperation on Ensemble forecasting
• NAEFS
• ET-EPS of the CBS
• Joint Working Group on Verification WGNEWWRP

47
Thank you again!
48
(No Transcript)