1 / 33

Bin Wang

Multi-Model Ensemble Seasonal Prediction System Development. 2007 APCC International Research Project. Bin Wang. IPRC, University of Hawaii, USA. Executive Summary.

hali
Download Presentation

Bin Wang

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. Multi-Model Ensemble Seasonal Prediction System Development 2007 APCC International Research Project Bin Wang IPRC, University of Hawaii, USA

  2. Executive Summary The Climate Prediction and its Application to Society (CliPAS) team is an international research project of the Asia-Pacific Economic Cooperation (APEC) Climate Center (APCC). Its goals is to provide APCC with frontier research in climate predictability and prediction and to facilitate APCC’s effort in developing first-rate model tools and technologies and continuously improving APCC operational forecast system. The strategy of APCC/CliPAS is to coordinate leading climate scientists in 12 institutions through well designed research projects and to share their expertise in climate prediction and its application. • In 2007 APCC international project, the APCC/CliPASteam has devoted to improving APCC operational multi-model ensemble (MME) seasonal prediction system through • implementing innovated MME schemes to APCCand • providing one-tier predictions for 2007 winterusing coupled models developed by three non-operational institutions in APCC/CliPAS team. • The APCC/liPAS team has strive to address forefront climate issues through coordinated multi-institutional retrospective forecast experiments and analysis of 21 models’ two-decade long hindcast.

  3. Executive Summary

  4. Executive Summary

  5. Implemented MME-S in APCC • SPPM and MME-S was tested on prediction of 850 hPa temperature precipitation using APCC hindcast data for the period 1983-2003 and operational forecast data for 2006 and 2007. • SPPM code was transferred to APCC and is now part of the Automated Forecast System. Advantage to using SPPM2 • Computational Estimates • (per 1 model, 1 variable, 22 years) CPPM- OLD version : 72 hours CPPM – New version : 12-15 hours SPPM v2 : 5 hours (suggestion: If you use 8 cpu simultaneously, it takes 10 hours for all models’ hindcast and forecast and two variables) MME-S Procedure STEP 1: Applying statistical correction using SPPM to individual models (1) Prior prediction selection (2) Second Step: Pattern Projection (2) Improved skill, especially for precipitation (3) Optimal choice of prediction STEP 2: Simple multi-model composite using available predictions

  6. Implemented SPPM and MME-S in APCC Temporal Correlation Skill of APCC MME Prediction for the period 1983-2003 Hindcast (83-03) and Forecast (2006) skills for four seasons JJA Precipitation JJA Temperature at 850 hPa

  7. Prediction of JJA T850 in 2007 Observed and Predicted Anomaly of JJA 850 hPa Temperature in 2007

  8. 6-month lead coupled predictions initiated from Nov 1, 2007 DJF Temperature DJF Precipitation • 3 non-operational coupled models made real-time 6-month lead prediction initiated from November 1, 2007 from FRCGC in Japan, SNU in Korea, and UH in USA. • This implementation is expecting to improve APCC operational MME prediction because the scientific results of 2006-2007 APCC international project show that the one-tier models have better skill than two-tier models in general

  9. Description of APCC Operational Models Total number of model being used 9 10 7 7

  10. Case study of the causes of the seasonal forecast for which most models failed Anomaly Pattern Correlation Skill • Forecast skill was very poor for most models during DJF 1989/90, MAM 1994, SON 2003, and DJF 2003/04 in which SST anomaly is very weak over equatorial Central and Eastern Pacific. • It is found that most of coupled models failed to predict SST anomalies over Tropical Oceans as well as extratropical Oceans, resulting in the failure in predicting atmospheric circulation and precipitation. • It is interestingly noted that prediction skill in winter season is strongly related to that in previous fall season during recent decade. Lee, June-Yi, J.-S. Kug, B. Wang, C.-K. Park, K.-H. AN, Saji H., and H. Kang, 2007: Assessment of APCC MME retrospective and realtime forecast for seasonal climate. Will be submitted to Clim. Dyn.

  11. Case study of the causes of the seasonal forecast for which most models failed • The observed warm anomalies over Equatorial Central Pacific and North Pacific were very difficult to be captured by current climate models. • For all cases, the spatial pattern of predicted anomalies was quite different among models, resulting in very weak anomalies of MME prediction all over the globe.

  12. Design a hierarchy of metrics to evaluate climate models’ performance on intraseasonal-to-seasonal prediction • We design a hierarchy of metrics to evaluate climate models’ performance on intraseasonal-to-seasonal prediction • We published and submitted several papers on evaluating climate models using the metrics. • Wang et al, 2007: Assessment of APCC/CliPAS 14-model ensemble retrospective seasonal prediction. To be submitted to J. Climate • Wang et al, 2007: How accurately do coupled climate models predict the leading modes of A-AM interannual variability? Clim. Dyn. DOI:10.007/s00382-007-0310-5 • Wang et al., 2007: Coupled predictability of seasonal tropical precipitation. CLIVAR Exchanges 12, 4, 17-18 • Kim et al., 2007: Simulation of intraseasonal variability and its predictability in climate predictio models. Clim. Dyn. DOI 10. 1007/S00382-007-0292-3 • We will make technical report to evaluating APCC operational prediction using some part of the metrics including mean states and interannual variability of monsoon and ENSO-monsoon relationship. The upper limit of practical predictability of global precipitation using APCC operational prediction will be also investigated.

  13. Updated and completed the four-season multi-institutional retrospective forecast experiments Current Status of Multi-Institutional Retrospective Forecast Experiments • The APCC/CliPAS team completed the four-season multi-institutional retrospective forecast experiments for the period 1979-2004 for advancing our understanding of climate predictability and determining the capability and limitations of the MME prediction. We collected 7 two-tier and 7 one-tier predictions from 12 institutions in Korea, USA, Japan, China, and Australia.

  14. Model Descriptions of CliPAS System APCC/CliPAS Tier-1 Models APCC/CliPAS Tier-2 Models

  15. CliPAS/APCC HFP Daily DATA Variables Institution

  16. Comparative Assessment of the One-Tier and Two Tier MME predictions 1 Seasonal prediction skill of JJA precipitation over A-AM and ENSO region • One-tier and two-tier MME predictions have been compared using 7 one-tier and 7 two-tier predictions in APCC/CliPAS project. In JJA, the one-tier MME system has better skill than two-tier MME for seasonal climate prediction as well as simulation of mean and annual cycle. On the contrary, the skill difference between two MME system is very small in DJF. • NCEP two-tier prediction was forced by predicted SST using NCEP one-tier system. The comparative results between NCEP one-tier and two-tier prediction support the necessity to use one-tier system for predicting summer rainfall. Performance of Annual Mode vs Seasonal Prediction Skill June-Yi Lee, Bin Wang, and co authors, 2007: Forecast skill comparison between one-tier and two-tier multi-model ensemble prediction. To be submitted to J. Climate

  17. Comparative Assessment of the One-Tier and APCC Two Tier MME predictions Temporal Correlation Skill of JJA precipitation 0.21 Skill Difference between APCC T2 MME and T1 MME in (a) CliPAS and (b) DEMETER 0.27 0.27

  18. Comparative Assessment of the One-Tier and APCC Two Tier MME predictions Temporal Correlation Skill of DJF precipitation 0.27 Skill Difference between APCC T2 MME and T1 MME in (a) CliPAS and (b) DEMETER 0.31 0.32

  19. Comparative Assessment of the One-Tier and Two Tier MME predictions 2 SAPI WNPPI Correlation between local SST and precipitation Lead-lag relationship between Nino 3.4 SST and JJA PRCP Index • One-tier prediction shows increased feedback from local SST to some extent, although it bears similar systematic error as two-tier, especially over East China Sea and Western North Pacific. • One-tier prediction shows improved ENSO-monsoon teleconnection over Indian Ocean, while it exhibits unrealistic impact of JJA precipitation over Western North Pacific on Nino 3.4 SST following SON and DJF.

  20. Comparative Assessment of the One-Tier and Two Tier MME predictions 3 Velocity Potential at 850 hPa (shaded) and 200 hPa (contoured) • Two-tier MME shows distinctive difference from one-tier prediction during El Nino onset and decaying summers. • Precipitation error is large over South Asia in one-tier prediction during El Nino onset summers. • Two-tier MME has large error over the same region during El Nino decaying summers. Convergence (dashed line) Divergence (dashed line)

  21. Prediction of Equatorial SST • The equatorial sea surface temperature (SS) anomalies are the primary sources of climate predictablity worldwide. The 7-coupled GCMs’ MME SST forecast skills beat the SNU dynamical-statistical model’s performance and far better than persistence forecast. • In particular, the current MME capture the temporal variation of the two leading modes realistically. However, the spatial westward shift of MME prediction between the dateline and 120E could potentially cause errors of global teleconnection that is associated with equatorial SSTA, degrading seasonal climate prediction skills over both tropics and extratropics. EOF/ Equatorial SST [10S-5N] Wang, Bin, June-Yi Lee, J. Shukla, I.-S. Kang, C.-K. Park and coauthors, 2007: Assessment of APCC/CliPAS 14-model ensemble retrospective seasonal prediction (1980-2004). To be submitted to J. Climate

  22. Prediction of Indian Ocean SST Correlation Skill of Indian Ocean SSTA • The temporal correlation skill (TCC) for SST predictions over the equatorial eastern Indian Ocean (EIO) reaches about 0.68 at a 6-month lead forecast. The prediction for the equatorial western Indian Ocean (WIO) SST is about 0.8 for November initiation but drops below 0.5 at the 4-month lead for May initiation. However, the TCC skill for IOD index (SST at EIO minus SST at WIO) drops below 0.4 at the 3-month lead forecast for both the May and November initiations. There exist a July prediction barrier and a severe, unrecoverable January prediction barriers for IOD index prediction. Wang, Bin, June-Yi Lee, J. Shukla, I.-S. Kang, C.-K. Park and coauthors, 2007: Assessment of APCC/CliPAS 14-model ensemble retrospective seasonal prediction (1980-2004). To be submitted to J. Climate

  23. Predictability of Global Tropical Precipitation SEOF Modes for Precipitation over Global Tropics [0-360E, 30S-40N] Percentage Variance Prediction skill of each mode 54.3% (CMAP) 83% (MME) • The first two SEOF modes are very well predicted. The third are also reasonably well predicted. But all other higher modes are not predictable as shown by the insignificant correlation skill in the spatial structures and temporal variation. • We defined the first three modes are predictable part of the interannual variation using the current coupled MME prediction system.

  24. Predictability of Global Tropical Precipitation Upper limit of predictability if there is no other prediction source in MME system 0.4 correlation is correspondent to 16% of fractional variance. (d) will be same as (a) If there is no systematic anomaly errors for the “predictable modes” in MME prediction. We can quantify the “predictability” by the fractional variance that is accounted for by the “predictable” leading modes in the observations. Such “predictable” modes can be determined by examining models’ hindcast results

  25. Experimental hindcast of MJO with the UH hybrid coupled model Observed (left) and forecast (right) rainfall (mm/day) averaged over 10oS–10oN. For convenience observed rainfall (contours) are overlaid on the forecast in the right panel. • Experimental hindcasts of MJO have been produced using UH hybrid coupled GCM for 4 months of the TOGA-COARE program in 1992-1993. The model was initialized with observations from January 1, 1993, and allowed to run freely for 2 months. A comparison of daily rainfall from the observations (left) and from a 100-ensemble-mean model output (right) reveals that the model was able to “forecast” the eastward movement and associated rainfall of the MJO beyond one month fairly accurately. Fu, Xiouhua, Bin Wang, Q. Bao, P. Liu, and B. Yang, 2007: Experimental dynamical forecast of an MJJO event observed during TOGA-COARE period. Submitted to GRL

  26. Experimental hindcast of ISO with the UH hybrid coupled model Observed (left) and forecast (right) rainfall (mm/day) averaged over 60oE–120oE. For convenience observed rainfall (contours) are overlaid on the forecast in the right panel. • Experimental hindcastsof Boreal summer monsoon ISO have been produced using UH hybrid coupled GCM for summer of 2006. The model was initialized with NCEP reanalysis data on June 11, 2006. A comparison of daily rainfall from the observations (left) and from a 100-ensemble-mean model output (right) reveals that the model was able to “forecast” the northward movement and associated rainfall of the ISO beyond one month fairly accurately.

  27. The impact of the systematic errors on ENSO-monsoon relationship Precipitation (shading) and SST (contour) Anomaly Systematic and Anomaly Errors of JJA SST Forecast • The errors in El Nino amplitude, phase, and maximum location of variability in coupled models are related with mean state errors such as colder equatorial Pacific SST and stronger easterly wind over western equatorial Pacific. • The breaking relationship between ENSO and Indian monsoon is evident in observation, whilst the MME produce clear negative relationship mainly related to SST anomaly bias . • The anomalous precipitation and circulation are predicted better in the ENSO decaying JJA than ENSO developing JJA. June-Yi Lee and Bin Wang, 2007: How is ENSO-monsoon relationship in coupled prediction affected by model’s systematic mean error? To be submitted to GRL

  28. Paper Preparation Published (or in press) Wang, Bin, June-Yi Lee, I.-S. Kang, J. Shukla, J.-S. Kug, A. Kumar, J. Schemm, J.-J. Luo, T. Yamagata, and C.-K. Park, 2007: How accurately do coupled climate models predict the leading modes of Asian-Australian monsoon interannual variability? Clim. Dyn. DOI: 10.1007/s00382-007-0310-5 Kim, H.-M., I.-S. Kang, B. Wang, and J.-Y. Lee, 2007: Simulation of intraseasonal variability and its predictability in climate prediction models. Clim. Dyn., DOI 10. 1007/S00382-007-0292-3. Wang, Bin and Qinghua Ding, 2007: The global monsoon: Major modes of annual variation in tropical precipitation and circulation. Dynamics of Atmospheres and Oceans. In press. Wang, Bin, June-Yi Lee, I.-S. Kang, J. Shukla, S. N. Hameed, and C.-K. Park, 2007: Coupled predictability of seasonal tropical precipitation. CLIVAR Exchanges, Vol. 12 No. 4. 17-18. In revision Jin, E. K, J. L. Kinter III, B. Wang and Co Authors, 2007: Current status of ENSO prediction skill in coupled ocean-atmosphere model. Climate Dynamics

  29. Paper Preparation Submitted Kug, J.-S., J.-Y. Lee, I.-S. Kang, B. Wang, and C.-K. Park, 2007: Optimal multi-model ensemble method in seasonal climate prediction. Submitted to Geophys Res Lett Fu, Xiouhua, Bin Wang, Qing Bao, Ping Liu, and Bo Yang, 2007: Experimental dynamical forecast of an MJO event observed during TOGA-COARE period. Submitted to GRL Emilia K. Jin and James L. Kinter III, 2007: Characteristics of Tropical Pacific SST Predictability in Coupled GCM Forecasts Using the NCEP CFS.Submitted to Clim Dyn Emilia K. Jin, James L. Kinter III, and Ben P. Kirtman, 2007: Impact of Tropical SST on the Asian-Australian Monsoon in GCM experiments. Submitted to Geo. Res. Let. To be submitted Wang, Bin, J.-Y. Lee, J. Shukla, I.-S. Kang, C.-K. Park and co authors, 2007: Assessment of APCC/CliPAS 14-model ensemble retrospective seasonal prediction (1980-2004). To be submitted to J. Climate Lee, June-Yi, Bin Wang, I.-S. Kang, J. Shukla, C.-K. Park and co authors, 2007: Performance of climate prediction models on annual modes of precipitation and its relation with seasonal prediction. To be submitted to Clim. Dyn. Lee, June-Yi and Bin Wang, 2007: How is ENSO-monsoon relationship in coupled prediction affected by model’s systematic mean errors? To be submitted to GRL. Lee, June-Yi, J.-S. Kug, B. Wang, C.-K. Park, K.-H. An, Saji H., H. Kang and co authors, 2007: Assessment of APCC MME retrospective and realtime forecast for seasonal climate. To be submitted to Clim Dyn.

  30. Conclusions (2) • The MME captures the first two leading modes of precipitation variability with high fidelity. • Potential to capture the precursors of ENSO in the A-AM domain. • The MME underestimates the total variances of the two modes and the biennial tendency of the first mode. • The correlation skill for the first principal component remains about 0.9 up to six months before it drops rapidly, but the spatial pattern forecast exhibits a drop across the boreal spring. • The coupled models’ MME predictions capture the first two leading modes of variability better than those captured by the ERA-40 and NCEP-2 reanalysis datasets. • Future reanalysis should be carried out with coupled atmosphere and ocean models.

  31. Challenges • Physical basis/Strategy • Correction of coupled model systematic errors in annual cycle • Improvement of the slow coupled modes • Improvement of coupled model initialization • Determine the roles of land-atmosphere interaction • Sub-seasonal prediction • Predictability of extreme events

  32. Directions • Improvement of models’ physics representation and correcting systematic errors. • Development of Multi-model one-tier system, including coupled data assimilation and reanalysis. • Improving slow coupled physics is a key for long-lead seasonal forecast. • Urgent need is to determine the role of land-atmosphere interaction in monsoon predictability. • Development of High resolution global models for prediction ofTC and other extreme events. • Determine predictability of ISO and improve monthly prediction.

  33. Thank You ! Any Questions and Comments?

More Related