interannual variability Teleconnections and interannual variability of the atmosphere Franco Molteni European Centre for Medium-Range Weather Forecasts

Outline Teleconnection patterns: Definition from observational datasets Dynamical origin and interpretation Explaining interannual variability through teleconnections: Partition of variance: effect of time and ensemble averaging Are teleconnection patterns generated by internal variability and boundary forcing/coupling the same? Predictability of teleconnection/low fr. variability indices Can we predict tropical rainfall? Linear indices: examples from the ECMWF System-4 re-forecasts

Teleconnections: definitions and examples Relationship between anomalies of opposite sign (and/or different variables) in different regions of the world Teleconnection pattern: Regional or planetary scale pattern of correlated anomalies Teleconnections may be induced either by internal atmospheric dynamics or by coupling with other components of the climate system (mainly the tropical oceans)

Teleconnections: definitions and examples The Southern Oscillation (and its links to El Niño) Walker and Bliss (1932); Bjerknes (1969) SOI: Tahiti – Darwin SLP Nino3.4 SST

ENSO teleconnections: rainfall and temperature

Teleconnections: definitions and examples The North Atlantic Oscillation Walker and Bliss (1932) Van Loon and Rogers (1978) Positive NAO phase Negative NAO phase

Teleconnections: definitions and examples The Indian Ocean Zonal Mode (or I.O. Dipole) Saji et al. (1999), Saji and Yamagata (2003) Webster et al. (1999)

Seminal papers: Wallace and Gutzler 1981 The Pacific / North American (PNA) pattern: correlations

Seminal papers: Wallace and Gutzler 1981 The Pacific / North American (PNA) pattern: composites

Seminal papers: Wallace and Gutzler 1981 Eastern Atlantic (EA) pattern: composites

Seminal papers: Horel and Wallace 1981 Correlation of 700hPa height with a) PC1 of Eq. Pacific SST c) SOI index Schematic diagram of tropical-extratropical teleconnections during El Niño

Seminal papers: Gill 1980 Atmospheric response to near-equatorial heating a) Near-surface wind and vertical motion b) Near-surface wind and sea-level pressure c) Motion in x-z plane along the Equator

Seminal papers: Hoskins and Karoly 1981 Response to a tropical heat source (15N): height (lon, z) at 18N 300hPa vorticity 300hPa height

Seminal papers: Simmons Wallace Branstator 1983 First barotropic normal mode at four times during a half-cycle: East. Atlantic phase PNA phase

Seminal papers: Sardeshmukh and Hoskins 1988 Definition of Rossby wave source generated by near-equatorial heating

Seminal papers: Sardeshmukh and Hoskins 1988 Response to heating in trop. West Pacific Rossby wave source and abs. vorticity

Non-linear models: Charney and DeVore 1979 Multiple steady states of very-low-order barotropic model with wave-shaped bottom topography

The new ECMWF Seasonal fc. system (Sys-4) IFS 36R4 0.7 deg (T255) 91 levels NEMO 1/1-0.3 d. lon/lat 42 levels OASIS-3 H-TESSEL Initial Con. Ens. Forecasts 4-D variational d.a. Gen. of Perturb. System-4 CGCM 3-D v.d.a. (NEMOVAR)

ECMWF System 4: main features Operational forecasts 51-member ensemble from 1st day of the month released on the 8th 7-month integration Experimental ENSO outlook 13-month extension from 1st Feb/May/Aug/Nov 15-member ensemble Re-forecast set 30 years, start dates from 1 Jan 1981 to 1 Dec 2010 15-member ensembles, 7-month integrations

Variability in an ensemble of time-evolving fields F (t, m) = F (t’, j, m) t’ = time within year j = 1, N (no. of years) m = 1, M (no. of ens. members) Climatology: Fcl (t’) = {F (t’, j, m) }j, m Anomaly: A (t’, j, m) = F (t’, j, m) - Fcl (t’) Seasonal mean anomaly: As (j, m) = {A (t’, j, m) }t’ Ens./seas. mean anomaly: Aes (j) = {A (t’, j, m) }t’, m for an observation/analysis dataset, M =1 ! in a “perfect” model environment , the average correlation between Aes and As is equal to the ratio of their standard deviations.

Variability of Z 200hPa in DJF from seasonal ensembles Standard deviation from 11-member ensembles, DJF 1981/2005 5-day means Seasonal- ensemble Seasonal means St.dev. ratio

‘Forced’ vs. internal variability in seasonal ensembles 11-member ensembles, DJF 1981/2005 Z 200 hPa El Niño composite La Niña 5-day-mean EOF 1 EOF 2

Can we predict tropical rainfall teleconnections? Tropical convection tends to occur over warm SST and warm-and-wet land surface, and it induces a number of feedbacks: Convective clouds decrease solar radiation at the surface On tropical oceans, anomalous divergent winds tend to increase evaporation to the west/east of the convective region when the climatological near-surface zonal winds are westerly/easterly The strength of the feedback depends on the heat capacity of the “interactive” surface layer and the phase of the day

Variability of tropical rainfall: EOF comparison GPCP S3 S4

Variability of tropical rainfall: EOF comparison GPCP S3 S4

Predictability of teleconnections in Sys4: Nino3.4, IOD SST DJF IOD SON

Predictability of telecon. in Sys4: South Africa rain (DJF)

Predictability of telecon. in Sys4: East Africa rain (SON)

Predictability of teleconnections in Sys4: Sahel rain (JJA)

Predictability of teleconnections in Sys4: PNA, NAO (DJF)

Predictability of telecon. in Sys4: Europe T_2m (MAM)

Conclusions Teleconnections may be induced either by internal atmospheric dynamics or by coupling with other components of the climate system (mainly the tropical oceans). Although a linear interpretation of teleconnection is adequate in most cases, non linear effects should not be neglected. Prediction of trop. rainfall teleconnections requires accurate modelling of heat-flux feedbacks from the ocean mixed layer and land surface. In the ECMWF seasonal fc. System-4, teleconnections related to ENSO are well modelled in the tropical and extra-trop. Pacific region, fairly well over Africa and South America, less so over South Asia. Predictability over Europe/Atlantic: mostly limited during winter, higher in other seasons when internal variability is reduced.