The Predictive Skill and the Most Predictable Pattern in the Tropical Atlantic: The Effect of ENSO Zeng-Zhen Hu 1 Bohua Huang 1,2 1 Center for Ocean-Land-Atmosphere.

The Predictive Skill and the Most Predictable Pattern in the Tropical Atlantic:

The Effect of ENSO

Zeng-Zhen Hu1

Bohua Huang1,2

1Center for Ocean-Land-Atmosphere StudiesInstitute of Global Environment and Society

Calverton, Maryland, USA

2Department of Climate DynamicsCollege of Science

George Mason UniversityFairfax, Virginia, USA

COLA

Main questions to be examined:

(1)What is the predictive skill of a real prediction/hindcast system of a CGCM in the tropical Atlantic?

(2) What are the most predictable patterns of SST, H200, and precipitation in the tropical Atlantic?

(3) What does result in the skill and patterns?

(4) What is the impact of the model bias on the skill and patterns?

DataCFS predictions:

• All calender months: Jan. 1981-Dec. 2003

• 15 predictions of 9 month length from lead month 0 to 8

• Oceanic ICs: 11th, 21st of month 0, 1st

of month 1 (Saha et al. 2006; Huang and Hu 2006).

• The atmospheric IC: the NCEP reanalysis II (Kanamitsu et al., 2002). Same day & those within two days before and after

• Data available from http://nomad6.ncep.noaa.gov/cfs/monthly

• AGCM: NCEP T62&L64 • OGCM: MOM3, 1/3o(10oS-10oN),

1o(higher latitudes); 74oS-64oN, L40

Other data (Jan/Nov. 1981-Dec. 2004)

• NCEP Global Ocean Data Assimilation (GODAS) (Behringer et al. 1998; Behringer and Xue 2004)

• NCEP/NCAR reanalysis I and II (Kalnay et al. 1996, Kanamitsu et al., 2002).

• The precipitation fields on a 2.5ox2.5o (Xie and Arkin 1996, 1997).

• The SST dataset is the second version of the optimally interpolated (OI-2) SST (Reynolds and Marsico 1993, Reynolds and Smith 1994, Reynolds et al. 2002).

CFS and Persistence forecast skills for all IC (Jan 1980-Dec 2003)* The predictive skill is higher in the west than in the east* CFS is better than persistence when lead longer than 4 months

Low predictive skill & large warm bias in the eastern Atlantic

Observed and CCSM Simulated JJA SST

Observed and CCSM3 Simulated MAM SST (from Deser et al. 2006: J. Climate, 19, 2451-2481)

Observed and CCSM3 Simulated SST Variance in JJA and MAM (from Deser et al. 2006: J. Climate, 19, 2451-2481)

*TB is the best and Dipole is the worst, NB and SB are in between*prediction with IC from boreal summer&autumn is more accurate

than that from boreal spring*CFS is better than the persistency when lead longer than 3 months

The predictive skill for TB index is higher than that of the Dipole index

Average for ALL ICs:

� The CFS predictive skill depends on regions and seasonal of ICs: Predictions with IC from boreal summer&autumn are more accurate than those from boreal spring.

� Predictive skill is higher in the west than in the east, due to large systematic errors the southeastern Atlantic. It is a common bias in current CGCM.

� CFS well predicts the basin wide SST variation, but poorly simulates the meridional gradient variation crossing the equator.

� The CFS prediction is better than the persistence forecast when lead time longer than 3 or 4 months.

What are the most predictable patterns of SST, H200, and

precipitation in March?

Why choose March as target month?� Large variability (Nobre and Shukla 1996)

� Strong connection with TAV and ENSO (Tourre & White 2005)

� High predictive skill in CFS (this work)

� The general patterns are similar for different targeted month (this work)

Maximized signal-to-noise EOF is used to identify the most Maximized signal-to-noise EOF is used to identify the most predictable patternpredictable pattern

MSN EOF1 of SST: A basin wide warming or cooling

MSN EOF1 SST & ENSO (the observation & predictions; Nov. IC & 4 month lead)

MSN EOF1 (Nov. IC & 4 Mon lead)

Simultaneous regression of March SST onto Nino3.4 Index

Simultaneous regression of March SST onto MSN EOF1Predicted SST Observed SST

The difference in tropical South Atlantic is large between CFS and obs.

Low predictive skill & large warm bias

12-8OS temperature regression onto MSN EOF1 of precipitationTop: NCEP/NCAR reanalysis; Bottom: CFS prediction

Theory of Chiang & Sobel (2002):The existence of atmosphere convection is crucial for temperature

variations to propagate from troposphere to surface

MSN EOF1 of H200: A basin wide coherent

variation symmetric about the equator

Nino3.4& H200

MSN EOF1 of H200 is associated with ENSO

MSN EOF1 of precipitation: Contrary variation of e. tropical Pacific and the tropical land

MSN EOF1 of precipitation is also associated with ENSO

CFS overestimates the contrary variation

Nino3.4 & observed P

Discussion I:1.The most predictable patterns of SST, H200, and precipitation are

largely connected with ENSO2: The predictive skill of CFS for ENSO is high

Discussion II-a:Warm bias in the southeastern Atlantic is mainly the result of excessive

SW radiation reaching ocean surface

Discussion II-b:The excessive SW radiation is caused by short of low cloud cover in CFS

Discussion II-c: CFS model can not correctly simulate the vertical inversion layer of the

temperature Observed: T850 warm than T925; CFS: T925 cold than T1000

observed low cloud exists in could ocean with inversion layerobserved low cloud exists in could ocean with inversion layer

• The predictive skill of CFS depends on region and IC month: west better than east; higher with IC in summer or autumn than in spring. The CFS is generally better than the persistency when lead time longer than 3 months.

• The CFS well simulates the basin wide SST variation, but poorly predicts the meridional gradient in the tropical Atlantic Ocean.

• The most predictable patterns of SST, H200, and precipitation in the tropical Atlantic in March are associated with ENSO. The CFS shows the ability to predict ENSO on interseasonal time scales.

• In the southeastern ocean, the systematic warm bias is a crucial factor leading to the low skill and unrealistic predictable pattern in this region.

• The warm bias is due to that CFS does not simulate the location of the major low cloud cover in the tropical Atlantic Ocean realistically.

Summary of the main results

(1) Huang, B. and Z.-Z. Hu, 2006: Cloud-SST feedback in southeastern tropical Atlantic anomalous events. J. Geophys. Res. (Ocean) (in press).

(2) Hu, Z.-Z. and B. Huang, 2006: The predictive skill and the most predictable pattern in the tropical Atlantic: The effect of ENSO. Mon. Wea. Rev. (revised).

(3) Huang, B. and Z.-Z. Hu, 2006: Evolution of model systematic errors in the tropical Atlantic basin from the NCEP coupled hindcasts. Clim. Dyn., (revised).

(4) Huang, B., Z.-Z. Hu, and K. Pegion, 2006: On observed and simulated cloud-radiation-SST feedback in the tropical Atlantic. J. Climate (in preparation).

Related Publications

EOF with Maximized Signal-to-Noise (MSN EOF) (I)• The MSN EOF is a method to derive patterns that optimize the signal-to-

noise ratio from all ensemble members. • This approach was developed by Allen and Smith (1997) and has been used

in Venzke et al. (1999), Sutton et al. (2000), Chang et al. (2000), and Huang (2004).

• An ensemble mean is supposely including a forced and a random part, which may be attributed respectively to the prescribed external boundary conditions and the unpredictable internal noise. In our case, the forced part is associated with the predictable signals which show certain consistency among different members of the ensemble predictions.

• The leading MSN EOF mode is the one with the maximum ratio of the variance of the ensemble mean to the deviations among the ensemble members.

• In this work, we only analyze the leading MSN EOF mode, which is defined as the most predictable pattern and is significant at the 95% level using a F-test.

• Details of this method were documented in Allen and Smith (1997), Venzke et al. (1999), and Huang (2004).

MSN EOF (II)• Variable: Xensemble =Xforced+

Xrandom

• Covariance: Censemble =Cforced+ Crandom (If Xforced and Xrandom are temporally uncorrelated)

• Crandom = Cnoise/15 (Cnoise is the average noise covariance of the 15 ensemble members

• To find the eigenvector of Cforced, the key procedures is to eliminate the spatial covariance of the noise. Which is equivalent to a transformation F such that

FT CrandomF= I• The transformation guarantees

that FT CforcedF and FT CensembleF have identical eigenvalues.

• F is estimated from the first K weighted EOF patterns of the deviations X’i =Xi-Xensemble (i denotes the ith member within the ensemble).

• The matrix of eigenvectors (E) of FT CensembleF contains a set of optimal noise filters, which can be restored into physical space by É=FE.

• The optimal filter (the 1st column vector é= É) maximizes the ratio of the variances of the ensemble mean and within-ensemble deviations (Venzke et al., 1999).

• The optimally filtered time series of Xensemble (i.e., its projection onto é) gives the 1st MSN principal component (PC).

• In practice, one can simply first project Xensemble onto F to form the pre-whitened data in the noise EOF space and then conduct an SVD to get both E and all MSN PCs simultaneously (Huang 2004).

• The 1st MSN EOF pattern, which represents the dominant spatial response, is derived by projecting Xensemble onto the 1st MSN PC.

GCM Systematic Errors and the Potential Influence of Cloud-SST feedback in the Southeastern Tropical Atlantic Ocean

Zeng-Zhen Hu1

Bohua Huang1,2

1Center for Ocean-Land-Atmosphere StudiesInstitute of Global Environment and Society

Calverton, Maryland

2Department of Climate DynamicsCollege of Science

George Mason UniversityFairfax, Virginia

COLA

SST Systematic Errors in CFS Hindcasts: Jan1981-Dec2003

Observed and CCSM Simulated JJA SST(from Deser et al. 2006: J. Climate, 19, 2451-2481)

Observed and CCSM3 Simulated MAM SST (from Deser et al. 2006: J. Climate, 19, 2451-2481)

Observed and CCSM3 Simulated SST Variance in JJA and MAM (from Deser et al. 2006: J. Climate, 19, 2451-2481)

•There are large systematic errors in the tropical Atlantic, particularly in the southeastern Atlantic in both CFS and NCAR models. This is a common model bias in current CGCM.

* The systematic errors affect the predictive skill in the region.

What Causes the Systematic Errors in the Southeastern Tropical Atlantic ?

It May Result from the Poor Simulation of the Climatology of Cloud-Radiation-SST and their

Feedback in the Region in Current GCMs

Both the CFS and COLA CGCMs are unable to simulate the seasonal variation of the cloud cover and net SW radiation flux in the southeastern Atlantic. That is associated with the SST bias in the region.

How does the feedback work in the real world and in CGCMs?

CFS model can not correctly simulate the vertical inversion layer of the temperature

Observed: T850 warm than T925; CFS: T925 cold than T1000

observed low cloud exists in could ocean with inversion layerobserved low cloud exists in could ocean with inversion layer

Summary• Marine low clouds modulate the seasonal and interannual variability

in the southeastern tropical Atlantic Ocean.

• Observationally, SST anomalies near the southeastern boundary in February affect the seasonal increase of the marine low clouds in JJA.

• The JJA cloud anomalies in turn changes the solar radiation reaching the sea surface over a broad region and expand the SST anomalies into the open ocean.

• Current CGCMs do not simulate the location of the major low cloud cover in the tropical Atlantic Ocean realistically.

• The cloud-radiation-SST feedback is not adequately represented in CGCM simulated seasonal and interannual variations.