1 / 21

Roberto Buizza European Centre for Medium-range Weather Forecasts

ECMWF medium-range/monthly ensemble. Roberto Buizza European Centre for Medium-range Weather Forecasts. Atmospheric model. Wave model. The ECMWF Integrated Forecasting System (IFS). EDA 25 T L 399L137. HRES T L 1279L137 (d 0-10). ENS 51 T L 639L91 (d0-10) T L 319L91 (d10-15/32).

lara-moses
Download Presentation

Roberto Buizza European Centre for Medium-range Weather Forecasts

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. ECMWF medium-range/monthly ensemble Roberto Buizza European Centre for Medium-range Weather Forecasts

  2. Atmospheric model Wave model The ECMWF Integrated Forecasting System (IFS) EDA25 TL399L137 HRES TL1279L137 (d0-10) ENS51 TL639L91 (d0-10) TL319L91 (d10-15/32) Seasonal S451 TL255L91 (m0-7/12) Atmospheric model Wave model Ocean model ORTAS4 Real Time Ocean Analysis ~8 hours ORAS4 Delayed Ocean Analysis ~12 days

  3. The ENS re-forecast suite to estimate the M-climate …28 6 13 20 27 March … Following Thomas Hamill work, a re-fc suite is part of all ECMWF ensemble systems. Each day, the M-climate is estimated using 500 EPS re-forecasts: • 20 years (1994 – 2013) • 5 ICs (-14d,-7d,0,+7d, +14d) • 5 members Some of the ENS products (e.g. the Extreme Forecast Indices) are bias corrected and/or calibrated using the model climate. 2014 2013 5 5 5 5 5 5 5 5 5 5 51 T639 L91 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 51 T319 L91 5 5 5 5 5 5 5 5 5 2012 20y 2011 ….. 1994

  4. ENS to provide probabilities of weather scenarii

  5. ENS to identify rare events 80% Mcli t+84-108h t+0-72h t+36-60h TP24 CDFs for Boulder (From Ivan Tsonevski) Boulder, 12-13 Sep 2013. This plot shows EFI fcs for TP24 valid for the 24-hours between 00 UTC of 12 and 13 September 2013. EFI was increasing approaching the date.

  6. How do they perform? Ensembles must be reliable Reliability is a key property that probabilistic forecasts must have. T850 24TP 10WS

  7. How do they perform? Ensembles must be reliable NH T500 – fc v analysis NH T500 – fc v radiosondes NH T500 – fc v AMSUA ENS reliability is sensitive to verification field (analysis or obs). t+24h t+24h t+24h w/o stoch. phys. with stoch. phys w/o stoch. phys. with stoch. phys w/o stoch. phys. with stoch. phys. t+120h t+120h t+120h w/o stoch. phys. with stoch. phys w/o stoch. phys. with stoch. phys. w/o stoch. phys. with stoch. phys. (From M Yamaguchi)

  8. ENS to predict anomalies weeks ahead Day 12-18 Day 19-25 9D 23D 6J 20J 3F 17F 3M This plot shows ENS weekly anomaly forecasts over North America in D13-JFM13 (the right panels show average 2mT over (120-60W;40-70N).

  9. ENS to predict TC genesis (anomalies) weeks ahead

  10. How do the ensembles perform? HRES, ENS-C, ENS-M

  11. How do the ensembles perform? HRES, ENS-C, ENS-M

  12. How do the ensembles perform? ENS-C, ENS-M, ENS-std

  13. How do ensembles perform? Scores’ trends CRPSS Z500 NH Scores’ trends can be used to monitor progress. This plot shows that for upper-level fields over the NH extra-tropics , performance has been improving continuously. Results indicate predictability gains of 1.5-2.0 days per decade.

  14. How do ensembles perform? Scores’ trends CRPSS TP24 EU Similar, although smaller in size, improvements can be seen by looking at 24h total precipitation, with skill gains of about 1.75 days between 2001 and 2012. The plot shows the forecast lead time when CRPSS crosses a 10% value.

  15. How do they perform? Ensembles must be reliable • 4 TIGGE ENS: ECMWF JMA NCEP UKMO • Verification Areas TIGGE-based results for Jul-Oct 2012. TC tracking data have been created with the ECMWF vortex tracker (Vitart et al. 1997, JC). Verified TCs are storms with a ‘Tropical Storm’ intensity or stronger (>35 knots). • BSS • 10 deg. X 10 deg. • BSS • 20 deg. X 10 deg. (From M Yamaguchi)

  16. How do the ensembles perform? TIGGE ENS

  17. How do the ensembles perform? ROCA[PR(2mT>U3)] NH ROC score for prob. Fc of 2-meter temperature in the upper tercile. Day 19-32 Day 12-18 Pers. d12-19 Pers. d5-11

  18. Nov 2013: ENS coupling from d0 to a better marine model … Work is progressing to introduce a better, unified wave-currents-sea-ice model (LIM). The new model based on NEMO is under testing at higher resolution, ORCA_025_Z75. In 2014 we will complete a ¼ degree NEMOVAR re-analysis, and we will then implement the ¼ model first in ENS, and then in the seasonal system (S5, planned for 2016/17). (From F Vitart) Nov 2013: coupling from initial-time to a new version of the ocean model (NEMO), with 1-way wave-currents coupling, improved skill, especially in the monthly time-range.

  19. Nov 2013: ENS coupling d0, L62>L91, new physics+SKEB Results based on 61 cases (JFM12, JJA12); 38r2 analyses and EDA perturbations. (From M Leutbecher)

  20. Nov 2013: EDA-based land-surface pert. in ENS ICs SWVL1 – std(t=48h) Nov 2013: EDA-based surface initial perturbations introduced in ENS. Preliminary results indicate increased reliability in the short forecast range, due to small spread improvements. EDA[sf] NOSPPT EDA[ua] NOSPPT EDA[sf,ua] NOSPPT EDA[sf] Initial perturbations improved by coupling more the ensemble of analyses and forecasts.

  21. The future: moving towards a more integrated approach Today The future EDA25 ERA 4DV ENS51 S451 HRES EnsembleN coupled analysis and re-analysis EnsembleN coupled forecasts ORAS45 PDF(0) << coupled analysis and re-analysis (past) stream PDF(T) << medium/long range stream PDF(0) << 4DV+EDA+ORAS4 PDF(0) << ERA+ORAS4 (past) PDF(T) << HRES+ENS/S4

More Related