1 / 25

Extra-tropical flow regimes and connections with tropical rainfall i n the MINERVA experiments

Extra-tropical flow regimes and connections with tropical rainfall i n the MINERVA experiments. Franco Molteni , Frederic Vitart , Tim Stockdale, Laura Ferranti (European Centre for Medium-Range Weather Forecasts, Reading, U.K.) Susanna Corti (ISAC-CNR, Italy)

ilana
Télécharger la présentation

Extra-tropical flow regimes and connections with tropical rainfall i n the MINERVA experiments

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. Extra-tropical flow regimes and connections with tropical rainfall in the MINERVA experiments Franco Molteni, Frederic Vitart, Tim Stockdale, Laura Ferranti (European Centre for Medium-Range Weather Forecasts, Reading, U.K.) Susanna Corti(ISAC-CNR, Italy) Ben Cash, David Straus (COLA/George Mason Univ., USA)

  2. The MINERVA experiments MINERVA: a COLA-ECMWF project sponsored by the NCAR Accelerated Scientific Discovery programme: • seasonal re-forecasts at T319, T639 (30yr, Nov+May IC) and T1279 (12yr, May IC) with IFS_cy38r1 + NEMO_v-3.1, run on NCAR Yellowstone HPC, 28M core-hours) Outline of results: • Predictive skill for NAO and PNA for seasonal (DJF) and month-2 (Dec) means • Probabilistic prediction of flow regime occurrence in the sub-seasonal range for the Atlantic and Pacific sectors • Teleconnections of Indo-Pacific rainfall and NH 500 hPa height: impact on NAO long-range predictions

  3. NAO, Dec (m2) T319 ac = 0.37 ERA Ens mean Ens members Z 500 Anomaly index = 1 σ T639 ac = 0.50

  4. NAO, DJF (m2-4) T319 ac = 0.26 ERA Ens mean Ens members Z 500 Anomaly index = 1 σ T639 ac = 0.51

  5. PNA, DJF (m2-4) T319 ac = 0.68 ERA Ens mean Ens members Z 500 Anomaly index = 1 σ T639 ac = 0.66

  6. A re-visitation of Pacific + Atlantic regimes: methodology • Data: • 5-day means of 500-hPa height from ERA-Interim • Dec.1979-Mar.1980 to Dec.2012-Mar.2013 (24 pentads*34 years = 816) • Definition of anomalies wrt 34-yr climate (low-pass filtered) • EOF analysis on 3 domains: • Euro-Atlantic (EAT: 80W-40E, 25-85N) • Pacific – North America (PNA: 160E-80W, 25-85N) • Pacific + Atlantic (PAT = PNA + EAT, 160E-40E, 25-85N) • Non-hierarchical cluster analysis using k-means algorithm • up to 6 clusters for EAT and PNA, up to 8 clusters for PAT • Significance test on signal-to-noise ratio (centroid variance / inter-cluster variance) against 500 red-noise data samples with same variance, skewness and lag-1 autocorrelation as individual PCs) • Refs.: Michelangeli et al. 1995, Straus et al. 2007

  7. Statistics for N-cluster partitions (%)

  8. Euro-Atlantic 4-cluster centroids NAO+ 31.5% Atl. Ridge 22.2% Blocking 25.0% NAO- 21.3%

  9. Pacific-North American 4-cluster centroids Arctic Low ( PNA- ) 27.7% Alaskan Ridge 20.6% Pacific Trough 27.7% PNA+ 24.0%

  10. Probabilistic prediction of regime occurrence • Bj [X(t)] : binary index of j-th cluster occurrence = (0, 1) • From cluster analysis: Bj [Xi] • Probabilistic index of cluster occurrence based on kernel estimator: Pj [X(t)] = ΣiK [X(t) – Xi] Bj [Xi] • Multi-normal Kernel function K = exp { -|X(t) – Xi|2 / (h s)2 } s2 = internal variance of clusters h = kernel width (0.25, 0.35, 0.50) • From time series of Bj [X(t)] for analysis and ensemble members, we compute 5-day and 15-day CRPS and mean abs. error of the ens. mean, as well as the associated skill scores : SS = 1. – S/Sclim

  11. T319: Skill score based on CRPS, all EAT clusters Score for 5-day means Score for 5-day means, 3-point filter Score for 15-day means 0.8 0.4 0.0 1 Nov (d0) 1 Jan (d61)

  12. T319: Skill scores based on CRPS and MAE, all clusters CRPS EAT MAE CRPS PNA MAE

  13. T319: Skill scores based on CRPS, 4 Eur-Atl. clusters NAO+ 31.5% Atl. Ridge 22.2% Blocking 25.0% NAO- 21.3%

  14. T639: Skill scores based on CRPS, 4 Eur-Atl. clusters NAO+ 31.5% Atl. Ridge 22.2% Blocking 25.0% NAO- 21.3%

  15. T319: Skill scores based on CRPS, 4 Pac.-N.Am. clusters Arctic Low 27.7% Alaskan Ridge 20.6%

  16. T639: Skill scores based on CRPS, 4 Pac.-N.Am. clusters Arctic Low 27.7% Alaskan Ridge 20.6%

  17. Skill for 15d-mean fc of NAO +/- regime indices NAO- NAO+

  18. Local correlation SST – precip, DJF 1981-2008

  19. Precip. teleconnections in DJF: GPCP 2.2

  20. Precip. teleconnections in DJF: System 4 (from Nov.)

  21. Z 500_hPavs.precip: ERA-Int. and System-4 ERA Sys4

  22. Correlations of Indo-Pac. rainfall and NAO (DJF)

  23. Impact of tropical rainfall correlation on teleconnections Cov. (Z500, WCIO) DJF Cov. (Z500, NINO4W) cor = 0.0 cor = 0.19 (obs) cor = 0.55

  24. Teleconnections with WCIO and NINO4 rainfall, DJF Nino4 T319 T639 WCIO T319 T639

  25. Summary • On seasonal timescale, T639 has the same predictive skill as T319 for PNA, but a (notably) higher skill for NAO; the NAO skill improvement is also seen in month-2 means. • On the sub-seasonal scale, considerable difference in predictive skill are found for different flow regimes. In the Euro-Atlantic sector, the NAO+ and Atlantic Ridge regimes are more predictable than NAO- and Blocking. • T639 shows a better skill than T319 in predicting the NAO+ regime occurrence, while skill for NAO- shows a stronger drop at day 20~30 • For Indo-Pacific rainfall, the MINERVA runs (as Sys-4) show stronger links between rainfall over the Western Indian Ocean and over the Maritime Continents / Central Pacific than those found in GPCP data. As a result, extratropical teleconnection patterns from these three tropical regions look more similar than in observations, and the NAO – Indian Ocean rainfall connection is underestimated. This problem is alleviated in T639 wrt T319, but only by 10~15%.

More Related