Tropical Atlantic studies with SPEEDO

Wilco HazelegerRein Haarsma

Camiel Severijns

Last year: Mean state of tropical Atlantic in SPEEDO (coupled Speedy-MICOM)

This year: Variability in the tropical Atlantic:- What are the mechanisms of tropical Atlantic variability?- What are the teleconnections with Europe?

Other projects at KNMI

Interannual variability: Tropical Atlantic VariabilityAtlantic Nino (T=2-10yrs)

Ruiz Barradaset al 2000

Interannual variability: Tropical Atlantic VariabilityInterhemispheric mode (T=2-10yrs)

Ruiz-Barradas et al

SPEEDO (SPEEDy-Ocean)

LBMLAND

FOCEANSEA

SLABG-MODELMICOM+

Sea ice

LAND PRESCRIBEDOCEAN

THERMODYNAMICOCEAN

DYNAMICOCEAN

ATMOSPHERE(Molteni, ICTP Trieste)

Hazeleger, et al TR-KNMI 2003New developments: upgrade physics package,dynamic vegatation model, atmospheric chemistry

atmsph. mixed layer

Speedy

SpeedO-Atlantic

Atmosphere (Speedy, global):

T30 (3.5 degree) primitive equations

7 layers

Simplified parameterizations

Ocean (MICOM-Atlantic: 40S – 70N):

Isopycnic coordinate primitive equation model (MICOM)

22 layers

1 degree horizontal resolution

SST prescribed outside the Atlantic and nudging at lateral boundaries

Land:Prescribed

MEAN STATE: SPEEDO-controlSST error Temperature along equator

first REOF SST and wind stress second REOF SST and wind stress

The coupled model results

SST error in exp with enhanced wind mixing Temp along equator

Rotated EOFsSST and associated wind stress

Cold tongue mode Interhemispheric mode

Cold tongue mode dominant in late summer, interhemispheric mode in spring

Heat budget of cold tongue mode

Entrainment latent heat flux

Short wave radiation horizontal advection

Heat budget of interhemispheric mode

Entrainment latent heat flux

Short wave radiation horizontal advection

Zonally averaged SST anomalies(8 yr filtered)

Seasons

Atlantic Subtropical Cell pathways

Schott, Stramma

EUC at 35W

Time series of seasonal data: a) 2N-2S, 50-160m positive u averagedb) Principal component of REOF 1Positive lag, REOF leads

Equatorial Undercurrent correlationswith cold tongue mode (unfiltered seasonaldata and 4-year filtered)

Seasonal data

4-year filtered data

0.2

0

-0.2

0.6

0

-0.2

North Brazil Current correlationswith cold tongue mode (seasonal data and 4-yearfiltered data)

NBC at 10S

Time series of seasonal data: a) 33W-36W, 40-250m, positive v averagedb) Principal component REOF 1Positive lag, principal component leads

Seasonal data

4-year filtered

0.2

0

0.4

0

-0.4

Lagged SVD of SST (Jul-Aug-Sept) and GH-500 (Oct-Nov-Dec)

Expl variance: 33%Correlation: 0.44

Other studies with MICOM at KNMI:

Process studies: Agulhas Ring decayrole southern oceans in THC (Drijfhout)

Impact diapycnal mixing on THC and sea levelrise (Katsman)

Mechanisms of tropical Pacific variability (Zelle,Oldenborgh)

What causes skewness of ENSO?

Response of SPEEDY and MICOM to a periodic wind forcing withan ENSO pattern

Skewness in thermocline, not in SST! (El Nino state minus La Nina state

Currently: using Lagrangian trajectory methodsTo trace sources of changes in Equatorial thermocline

(Roberto de Almeida, USP)

Impact of South Atlantic variability on tropical Atlanticcirculation in MICOM (ocean-only)

Summary and conclusions

SPEEDO produces realistic patterns and amplitude of Tropical Atlanticclimate variability:

a) Cold tongue mode generated by thermocline-upwelling feedbacks

b) Interhemispheric mode generated by upwelling and short-wave feedbacks (in the east) and by latent heat flux (in the west, a wind-evaporation-SST feedback).

Small impact of TAV on Northern Hemisphere variability

NBC transport may be a predictor for decadal variations of cold-tonguevariations, but correlations are weak

NB!! Thermocline variations do not couple well to SST in the eastern parts of the basin (this is only apparent in coupled models, in ocean-onlymodels Tair will generate variability, but surface flux anomaly should damp variability)!!

Heat budget of decadal variations(8yr filtered zonally averaged data)

SST anomalies

Entrainment latent heat flux

Heat budget of decadal variations

SST anomalies

Horizonal advection short wave radiation

Zonal velocity differences averaged between 50E and 20W in SPEEDO

Enhanced ocean wind mixing efficiency minus Control

First rotated EOF of SST in coupled SPEEDO and related variables

Second rotated EOF of SST in coupled SPEEDO and related variables

Control Wind mixing efficiency enhanced Buoyancy mixing efficiency enhancedObservations

SST error along equator depth of 20 degree isotherm

SST budget in MICOM ocean-only (1S,10W):

EntrainmentRadiationLatent heat fluxSensible heat fluxHorizontal advection Diffusion(all in W/m^2)

Control Enhanced wind mixing

Dhc

Qw)TT(HvTuT

t

T

pesyx

Heat budget from PIRATA

Seasonal cycle Tropical Atlantic

Colors SPEEDOContours Da Silva

Seasonal cycle SST along equator(colors SPEEDO, contours Da Silva)

Change in winds and rain in JAS

Idem for JFM

MEAN STATE: Observations

SST Temperature along equator

Standard deviation of SST

: obs

:COLA model

: SPEEDO

SST error along equator in coupled models

Davey et al 2001

Dominant patterns of covariability betweenSST and early winter 500 hPa geopotential height when SST leads by 4 months. (Czaja and Frankignoul 2001)

Users shell for MICOM developed by

Dutch Climate Research Center (CKO)

Camiel Severijns severijn@knmi.nl

Can be downloaded from website: www.knmi.nl/onderzk/CKO

• NetCDF based

• Easy to use: basin, resolution, parameter choice.

• Coupling implemented as a library, using SCRIP tools for

interpolation

• Analysis tools (Bausteinen) for EOF, SVD, spectra etc.