WMO Lead-centres for verification and archiving Richard Graham (Met Office Hadley Centre) and Won-Tae Yun (KMA) acknowledgements to all GPCs and members of.
Download ReportTranscript WMO Lead-centres for verification and archiving Richard Graham (Met Office Hadley Centre) and Won-Tae Yun (KMA) acknowledgements to all GPCs and members of.
WMO Lead-centres for verification and archiving Richard Graham (Met Office Hadley Centre) and Won-Tae Yun (KMA) acknowledgements to all GPCs and members of the WMO Expert Team on Extended and Long-range Forecasting (ET-ELRF) WMO workshop on Sub-seasonal to Seasonal Prediction, Met Office, Exeter, 1-3 December 2010 © Crown copyright Met Office Context • Forecasters like to compare output from different models, this is helped by: • easy access to products; • same map projections, contour conventions etc; • digital data for ‘user’ processing. • Multi-model combinations also important • Verification diagnostics also easier to use if presented uniformly • Need for international coordination 2m temperature forecasts for DJF2010/11 ECMWF JMA WMO CBS coordination of centres producing long-range forecasts: Global Producing Centres (GPCs) First step: agree standard forecast and verification output: • Designate GPCs: centres adhering to agreed procedures/standards in LRF – there are now 12 GPCs, forming an integral part of the WMO GDPFS Next step: increase coordination • 2 Lead Centres, facilitating user access to GPC products • Lead Centre for Long-range Forecast Multi-model Ensembles (LCLRFMME) • Lead Centre for the Standard Verification System for Long-range Forecasts (LC-SVSLRF) • Aim: improve access and usability of global LRF products to aid production of regional/national climate services © Crown copyright Met Office GPC designation criteria • have fixed production cycles and time of issuance; • provide an agreed minimum set of long-range forecast products, • 2m temp, precipitation, SST (averages of at least 1-month); • any lead-time between 0 and 4 months. • provide an agreed set of forecast quality measures (WMO Standard Verification System for LRF); • provide up-to-date information on forecast methodology used; • make products accessible to users (through website and Lead Centres) © Crown copyright Met Office The 12 WMOdesignated GPCs GPC name Centre System Configurati on (ensemble size of forecast) Resolution (atmosphere) Hindcast period used Beijing Beijing Climate Centre Coupled (48) T63/L16 1983-2004 CPTEC Centre for Weather Forecasts and Climate Studies 2-tier (15) T62/L28 1979-2001 ECMWF European Centre for Medium Range Weather Forecasts Coupled (41) T159/L62 1981-2005 Exeter Met Office Hadley Centre Coupled (42) 1.25°x1.85°/L 38 1989-2002 Melbourne Australian Bureau of Meteorology Coupled (30) T47/L17 1980-2006 Montreal Meteorological Service of Canada 2-tier (40) T32/T63/T95/2.0 °x2.0° (4model combination) 1969-2004 Seoul Korean Meteorological Agency 2-tier (20) T106/L21 1979-2007 Tokyo Japan Meteorological Agency Coupled (51) T95/L40 1979-2008 Toulouse Météo-France Coupled (41) T63/L91 1979-2007 Washington National Centres for Environmental Prediction Coupled (40) T62/L64 1981-2004 Moscow Hydromet Centre of Russia 2-tier (10) 1.1°x1.4°/L28 1979-2003 Pretoria South African Weather Service 2-tier (6) T42/L19 1983-2001 WMO Lead Centre for LRF Multimodel Ensembles (LC-LRFMME) Jointly operated by Korean Met Agency & NOAA NCEP Won-Tae Yun, Arun Kumar Primary functions: http://www.wmolc.org/ • collect LRF data from GPCs • core data: monthly pre-calculated anomalies (EM + members) • additional data: hindcast and forecast active started but not active • display GPC forecasts in standard formats (next 3-months only at present) • generate and display an agreed set of MME products • distribute GPC data (for those GPCs that allow it) • promote research into MME techniques © Crown copyright Met Office GPC forecasts for DJF 2010/11: ensemble mean pmsl anomaly (from LC-LRFMME web) Exeter ECMWF Toulouse Beijing Melbourne Seoul Montreal Washington Tokyo ‘Dynamic’ userselectable domain Coupled systems Un-coupled systems Pretoria Multi-model GPC pmsl forecasts, DJF 2010/11 from LC-LRFMME website Ensemble mean of 10 GPCs multi-model subsets produced interactively from LC-LRFMME website Ensemble mean of 7 GPCs (coupled systems) Ensemble mean of 3 GPCs (uncoupled systems) Presented at South East Europe Regional Climate Outlook Forum (SEECOF), 22-26 November 2010 What about last year? pmsl anomalies: observed & GPC predictions from Nov09 for DJF2009/10 Observed pmsl anomalies anomaly sign ‘consistency’ 10-model multi-model EM (wrt 61-90) Other products/activities Other variables: • Nino3.4 region plumes • 2m temperature; precipitation; 850 hPa temperature; 500hPa height; SST Methods of multi-model combining: • Operational: simple equal weights • Research: skill weighted combination; linear and non-linear methods Sub-seasonal: • 1-month means available in addition to 3-month means Note: ET-ELRF: last meeting recommended pilot projects on potential for monthly-range prediction for (e.g. rains onset) Priority prediction needs Africa: rainfall DFID-Hadley Climate Science Research Partnership Question: what next beyond standard 3-month means and terciles? • Poll of 9 African Regional Centres and National Meteorological Services • Highest priority: predictions of the temporal distribution of seasonal rainfall (e.g. onset, duration, dry spells) • Less clear at what lead-time onset prediction is needed to be useful © Crown copyright Met Office Lead Centre for Standard Verification of Long-range forecasts (LC-SVSLRF) http://www.bom.gov.au/wmo/lrfvs/ Jointly operated by Bureau of Met. BoM) & Met. Service of Canada (MSC) David Jones, Normand Gagnon • repository of global SVSLRF scores for GPC forecasts • Core of SVS: ROC maps and diagrams, reliability diagrams, MSSS • documentation on verification scores (Attachment II-8 of the WMO manual on the GDPFS) • observation datasets for verification, and code for calculating the SVSLRF scores Examples of submitted data http://www.bom.gov.au/cgi-bin/climate/wmo.cgi Reliability/sharpness diagrams ROC curves Courtesy: David Jones (BoM) ROC score maps Use of Lead Centre products at African Regional Climate Outlook Forums (RCOFs) GPC forecasts for E. Africa ‘short-rains’ SON 2010: ensemble mean rainfall anomaly Exeter ECMWF Toulouse Beijing Melbourne Tokyo Moscow Seoul Montreal Coupled systems Un-coupled systems Washington Sample verification: ROC scores for SON precipitation, 1-month lead Melbourne Toulouse Washington above below 0.6 Tokyo WMO Lead Centre information and final consensus, SON 2010 Model consistency Verification Ensemble mean of 10 GPCs Ensemble mean of 7 GPCs (coupled systems) Ensemble mean of 3 GPCs (uncoupled systems) Forecaster judgement Final consensus WMO Lead Centres in context of GFCS GPCs: role in the Climate Services Information System (CSIS) GFCS users CSIS obs/research Other Centres GPCs and their LCs Monitoring Centres WMO Global Seasonal Climate Updates (GSCUs) • International consensus on global-scale seasonal monitoring and outlook (will extend the WMO El Niño/La Niña updates); • Other climate indices (Indian Ocean Dipole, NAO) • Temperature, precipitation • Intended users: RCCs, RCOFs, NMHSs, but also ‘globally acting users’ (e.g. UN, Red Cross); • Scoping meeting and first meeting of GSCU Task Team took place 12-15 October 2010; • identified work needed before Pilot Phase can be begin • requirements of GSCUs on the Lead Centres are feeding into work plans, e.g.: • probability forecast products, extension of forecast range • verification of the multi-model products Summary • WMO CBS (working with CCl) has designated GPCs and associated Lead Centres for multi-modelling and verification strengthening coordination of LRF • Forecast and verification products are available to RCCs /NMHSs / RCOFs - active use is being made • Lead Centres will play a key role in WMO Global Seasonal Climate Updates (monitoring and outlook) – part of GFCS vision. • Key plans for the WMO Lead Centres (guided by the Expert Team for Extended and Long-range Forecasts) include: • development of probability forecast products • extension prediction range (to ~6 months) • verification of multi-model products • possible centralisation of the verification process • Issue of length of hindcast – but also of consistency of choice of the hindcast period across the GPCs Thank you! Any questions?