1 / 48

Seasonal forecasting: status and plans

Seasonal forecasting: status and plans. David Anderson Tim Stockdale, Magdalena Balmasda, Arthur Vidard, Alberto Troccoli, Paco Doblas-Reyes, Kristian Morgensen, Malcolm MacVean, Laura Ferranti, Frederic Vitart. ECMWF System-2 is the same as last year.

fayre
Download Presentation

Seasonal forecasting: status and plans

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Seasonal forecasting: status and plans David Anderson Tim Stockdale, Magdalena Balmasda, Arthur Vidard, Alberto Troccoli, Paco Doblas-Reyes, Kristian Morgensen, Malcolm MacVean, Laura Ferranti, Frederic Vitart

  2. ECMWF System-2 is the same as last year. • System-3 is under development. This involves substantial changes to the coupled system, though the changes should be largely transparent to the user. • Another major initiative is the development of a real-time, multi-model, multi-analysis system. The UKMO and Meteo France systems are implemented. Other Met Services may join in future: DWD, Spain.

  3. System 2 Recap • Forecast model • TL95L40 atmosphere • L29 ocean (Variable resolution 0.3X1in tropics) • Ocean Analysis • OI analysis of T • Corrections to salinity when T is assimilated. • Corrections to velocity when density is updated. • Ensemble of ocean analyses: 5 analyses perturbed by wind anomalies. • Ensemble generation • From each ocean analysis, 8 perturbations to SST are made, creating an ensemble of 40 members.

  4. Cycle 23r4(24r1) of atmosphere, TL95, L40 • Calibration based on 5-member ensemble from 1987 to 2001. • 40 member ensembles run for November and May for validation. • Plumes contain forecasts for Nino3, Nino3.4, Nino4 • Various plots on the web e.g. terciles, 15%iles, ensemble means.. Precip, T2m, upper level data

  5. The information on which a seasonal forecast is based lies mainly in the ocean. (Soil moisture, snow cover, sea ice, atmospheric state.. may also add some predictability). Considerable effort is put into analysing the ocean state. • An ensemble of ocean analyses is created. • Five ocean analyses are created by perturbing the wind stress with perceived uncertainty. (These analyses are used to create an ensemble of forecasts.) The purpose of creating an ensemble of ocean analyses is to represent some of the uncertainty in knowing the ocean state. These analyses are used in creating the ensemble of forecasts in System-2 (the current ECMWF seasonal forecast system and in the monthly forecast system).

  6. Operating method of ARGO floats. • Operating

  7. Data coverage for June 1982

  8. Data coverage for March 2002

  9. Build up of ARGO: Data coverage for February 2005 • XBT, MOORINGS, ARGO floats

  10. System 3

  11. System-3 • A new cycle of the atmospheric model - 29r1 with 40 levels • or following physics cycle with green house gasses and aerosols • and 62 levels. • Extended range and size of back integrations. • Strawman 11member, back to 1982. • Include bias correction in ocean assimilation. • Include salinity assimilation. • Include altimeter assimilation • Revised wind and SST perturbations. • New sea-ice specification algorithm. • Include ocean currents in wave model. • Use EPS Singular Vector perturbations in atmospheric initial conditions. • Forecasts out to 12 months (4X per year) • Will use ENACT/ENSEMBLES ocean data.

  12. There will be some extra data archived: • Pressure level data for 100HPa, and 50HPa at 12 hourly intervals. • A full set for 400 and 300 HPa (currently only T) • MSLP every 6 hours (currently 12hours) • Potential temperature on PV=2 surface.

  13. OSCAR currents Prototype of System 3 currents Velocity fields

  14. System 2 Zonal velocities: Correlation with OSCAR (NOAA) currents (15m depth) Period 1993-2004, seasonal cycle removed Prototype of System 3

  15. Spring barrier: predictability potential predictability *) seasonal recharge oscillator   skill ECMWF operational forecast 1987-2001  Estimate of predictability with parameters and noise properties from seasonal fit. From Gerrit Burgers KNMI

  16. Wind stress perturbations Standard deviation of zonal wind stress differences (in N/m**2, multiplied by 100) for January ERA40-CORE (1986-1993) ERA15-SOC (1986-1993) ERA40-CORE (1980-2000) ERA40-CORE (1958-1979)

  17. Constant GHG Correlation = 0.52 Variable GHG Correlation = 0.77 Anthropogenic effect: T2m predictions 1-month lead, summer (JJA) predictions of global T2m

  18. The multi-model

  19. Forecast System is not reliable: RMS > Spread A) Can we reduce the error? How much? (Predictability limit) B)Or can we only increase the spread? • Improve the ensemble generation: Need to sample model error • Improve calibration: A posteriori use of all available information

  20. The Met Office model is very similar to that used in Demeter. The ensemble strategy follows the ECMWF S2 strategy except that it uses ERA-40/Ops rather than ERA-15/Ops and the wind perturbations used in the ensemble ocean analysis are half amplitude. The ocean analyses use the ENACT data set. • The atmospheric resolution is 2.5 X 3.75 degrees, with 19 vertical levels. The ocean has resolution of 1.25X1.25, increasing to .28220 in the north south direction near the equator. There are 40 vertical levels. The calibration period is 1987-2004, 15 members ensemble.

  21. The Meteo France atmospheric model has 31 vertical levels, TL63 resolution. The ocean model is ORCA: 2X2 at mid latitudes, increasing to 0.50 near the equator. There are 31 vertical levels. The ocean analyses are produced without in situ data assimilation. Altimeter data are used and a moderate relaxation to observed SST is applied. Forecasts are available from 1993-present. (A 5-member ensemble from 1993 to 2004 inc. The real-time forecast ensemble is 41.)

  22. Differences can be considerably larger e.g. in Nino4

  23. Results from the real-time multi-model forecast system. Three different models, using three different analysis strategies. Green is ECMWF, blue ECMWF + MO, red ECMWF+MO+MF.

  24. Predictions from the 3 multi-model components: Sahel precipitation Ecmwf Météo France Met Office

  25. Predictions from the 3 multi-model components: Guinea Coast precipitation Ecmwf Météo France Met Office

  26. Results are from DEMETER

  27. ENACT, DEMETER, ENSEMBLES, MERSEA

  28. ENACT • Enact was an EU framework V project, seeking to advance ocean data assimilation strategies, to generate an ensemble of ocean analyses for climate assessment and to assess the impact of different assimilation strategies on forecast skill.

  29. DEMETER and ENSEMBLES Activities

  30. The DEMETER heritage • DEMETER ended in September 2003. However, work has been carried out on forecast quality assessment (additional verification of time series), analysis of the benefits of the multi-model, etc. • Research on model calibration and combination has led to strong collaborations with CPTEC (Brazil) and IRI. • A special issue of Tellus A has appeared in May 2005. • Additional work on applications and end-user verification: the case of malaria.

  31. Seasonal predictability, Southern Europe T2m Precipitation 2-4 (JJA) 4-6 (ASO)

  32. Calibrated downscaled predictions PAGE agricultural extent PAGE agroclimatic zones From Coelho et al. (2005)

  33. Northern box Calibrated downscaled predictions Southern box From Coelho et al. (2005)

  34. ENSEMBLES project • Integrated Project funded by the EC within the VIth FP, 69 partners. • Start date: 1 September 2004, Duration: 5 years • Integrated probabilistic prediction system for time scales from seasons to decades, and beyond. • Seasonal-to-decadal hindcasts will be used to assess the reliability of forecast systems used for scenario runs. • Comparison of the benefits of the multi-model, perturbed parameters and stochastic physics approaches to assess forecast uncertainty. • Great diversity of applications: health, crop yield, energy production, river streamflow, etc.

  35. Initial s2d activities (RT1) • Main goal: assess best method to estimate model uncertainty among multi-model, perturbed parameter and stochastic physics approaches. • Estimates of model uncertainty using a new multi-model ensemble, a recently developed stochastic physics scheme (ECMWF and Met Office) and the perturbed parameters approach (Met Office with 2 different versions of HadCM3). • Ocean initial conditions from ENACT and new sets. • Common output archived at ECMWF in MARS (atmosphere) and ECFS (ocean). • Pre-production for 1991-2001 with reduced start dates and expected completion for end 2005. • Additional experiments to test the consistency of the predictions and the impact of the ensemble size.

  36. Three different forecast systems to estimate model uncertainty • Multi-model, built from ECMWF, Met Office, Météo-France operational activities and DEMETER experience. • Perturbed parameter approach, from the decadal prediction system (DePreSys) at the Met Office. • Stochastic physics, from the stochastic physics system developed for medium-range forecasting at ECMWF. • Design of a set of common experiments to determine the benefits of each approach.

  37. A service that offers immediate and free access to data from: • DEMETER • •ERA-40 • •ERA-15 • •ENACT • with monthly and daily data, select area and plotting facilities, GRIB or NetCDF formats Data dissemination Different depending on access granted to ECMWF systems: • access: MARS http://www.ecmwf.int/services/archive/ • no access: public data server and OPenDAP (DODS) server

  38. Downscaling for s2d predictions http://www.ecmwf.int/research/EU-projects/ENSEMBLES/news/index.html

  39. MERSEA • Will use the 0.25 degree ocean analyses from MERCATOR. • Will couple the 0.25 degree ocean to an atmosphere of comparable resolution to test the impact on an active ocean on medium range forecasts. • Will assess the impact of the high resolution ocean analyses on seasonal forecasts. • INGV, Meteo France, ECMWF, MERCATOR.

More Related