The World Weather Open Science Conference (WWOSC 2014) 16 21 August 2014, Montreal, Canada ENSO prediction using Multi ocean Analysis Ensembles (MAE) with NCEP CFSv2: Deterministic skill and reliability Jieshun Zhu 1, Bohua Huang 1,2, Larry Marx 1, James L. Kinter III 1,2 Magdalena A Balmaseda 3, Zeng-Zhen Hu 4, Rong-Hua Zhang 5, and Peitao Peng 4 1 Center for Ocean-Land-Atmosphere Studies (COLA) 2 Department of Atmospheric, Oceanic, and Earth Sciences, George Mason University (GMU) 3 European Centre for Medium-Range Weather Forecasts (ECMWF) 4 Climate Prediction Center/National Centers for Environmental Prediction/NOAA (CPC/NCEP/NOAA) 5 Earth System Science Interdisciplinary Center/University of Maryland, College Park (ESSIC/UMD) 1
Outline Motivation--Why MAE? Substantial Uncertainties in Different ODAs MAE Prediction Experiments: focusing on ENSO SST Deterministic Metrics Probabilistic Metrics: Reliability Summary 2
Introduction to ODA systems Model + Data Analysis Systems: ECMWF, NCEP/NOAA, GFDL/NOAA, GMAO/NASA, Mercator-Ocean, JMA, BOM/Australia, COLA, Models: MOM v?, NEMO/OPA, POP, HOPE Coupled vs. Uncoupled Assimilation schemes: OI, 3D-VAR, 4D-VAR, Kalman Filter (different variants) Obs. Data: Mooring, XBT, CTD, Argo, Satellite 3
Multiple Ocean Analyses ORA S3, ECMWF (Balmaseda et al. 2008) COMBINE NV, ECMWF (Balmaseda et al. 2010) GODAS, NCEP (Behringer 2005) CFSR, NCEP (Saha et al. 2010) SODA2.1.6, UM/TAMU (Carton and Giese 2008) ECDA, GFDL (Zhang et al. 2007) Sources of Uncertainty Different forecast systems (i.e., models) Different assimilation schemes Different surface and ocean data inputs Sparse and uneven distribution of obs. 4
CFSR Why Multiple Ocean Initialization? HCA ECMWF HCA Heat content anomaly (HCA) from ODA analyses shows high uncertainty Example: Tropical Atlantic 1st EOF modes from different analyses give different patterns Projection spread is large among analyses Zhu et al. (Clim. Dyn., 2012a) 5
Heat Content Anomaly moderate high low ODA Heat Content Uncertainty (1979-2007) DATA SOURCE Zhu et al. (Clim. Dyn., 2012a) ECMWF: ORA-S3, COMBINE-NV NCEP: GODAS, CFSR UM/TAMU: SODA GFDL : ECDA 6
Heat Content Anomaly Ensemble average reduces noise effectively Leading EOF patterns become physically meaningful S/N ratio improves significantly Signal exists in all analyses (masked by high internal noise) Zhu et al. (Clim. Dyn., 2012a) 7
Uncertainty is relatively low in Tropical Pacific Heat Content Anomaly Slight phase shift Is ocean uncertainty negligible for ENSO prediction? 8
Scientific Questions What are the effects of uncertainty in upper ocean heat content on seasonal-tointerannual (SI) prediction? Will ensemble predictions initialized with multiple ocean analyses (i.e., MAE) improve SI predictive skill in tropics? 9
Experiment Design 12-month hindcasts initialized in April Forecast Model: NCEP CFS version 2 1) Atmosphere T126, L64 2) Ocean (MOM4) 0.5 x0.5 (0.25 lat, 10 S-10 N), L40 Multi-Ocean Initialization Experiments (1979-2007) 1) Ocean initial state (OIC): Monthly means from COMBINE-NV, ORA-S3, CFSR, GODAS 2) Anomaly initialization in OIC 3) Perturbed Atmosphere-land IC (4-member with each OIC, Apr 1-4, CFSR) 10
MAE Jun Sep Dec Zhu et al. (GRL., 2012b) 11
Prediction skill of the Nino3.4 is sensitive to OICs (April ICs: 1979-2007) MAE(0.73) MAE(0.61) Predictive skills of individual OICs have substantial differences ES_Mean is comparable to the best of individual predictions Zhu et al. (GRL., 2012b) 12
NINO3.4 Index 13
NINO3.4 Index MAE 14
NINO3.4 Index MAE Obs El Ninos El Nino strength is usually underestimated 15
NINO3.4 Index False MAE False Missed Model or IC biases are still the major problem False Missed Missed 16
MAE improves prediction Reliability A necessary condition for reliability: Ensemble spread ~ RMSE Reliability Diagram Additional Hindcast Experiments CFS Reanalysis and Reforecast (CFSRR, Provided by NCEP, 9-month): Lagged Ensemble (LE) approach. 17
MAE gives better ensemble spreads (April ICs: 1979-2007) Zhu et al. (Clim. Dyn., 2013a) 18
MAE gives better ensemble spreads (April ICs: 1979-2007) Zhu et al. (Clim. Dyn., 2013a) 19
Zhu et al. (Clim. Dyn., 2013a) 20
Summary There is considerable uncertainty in upper ocean heat content anomalies from different analyses; OIC uncertainty causes a noticeable spread in ENSO prediction; MAE provides more reliable ENSO prediction. 21
References Zhu, J., B. Huang and M. Balmaseda, 2012a: An ensemble estimation of the variability of upper ocean heat content over the tropical Atlantic Ocean with multi ocean reanalysis products. Clim. Dyn., 39, 1001 1020. Zhu, J., B. Huang, L. Marx, J. L. Kinter III, M. A. Balmaseda, R. H. Zhang, and Z. Z. Hu, 2012b: Ensemble ENSO hindcasts initialized from multiple ocean analyses. Geophy. Res. Lett., 39, L09602, doi:10.1029/2012gl051503. Zhu, J., B. Huang, M. A. Balmaseda, J. L. Kinter III, P. Peng, Z. Z. Hu, and L. Marx, 2013a: Improved reliability of ENSO hindcasts with multi ocean analyses ensemble (MAE) initialization. Clim. Dyn., 41, 2785 2795. Zhu, J., B. Huang, Z. Z. Hu, J.L. Kinter, and L. Marx, 2013b: Predicting US summer precipitation using NCEP Climate Forecast System version 2 initialized by multiple ocean analyses. Clim. Dyn. 41, 1941 1954. 22