ENSO complexity:

A Monitoring and Forecasting Perspective

Slide 1

M.A. Balmaseda , Michael Mayer, Tim Stockdale, Antje Weisheimer and colleagues.

Current Ocean Monitoring Capabilities: ENSO energetics

Progress and challenges in ENSO prediction (at ECMWF)

Outlook: Coupled Century long reanalyses and reforecasts

Observations

Oce

an

Reanalysis = maps without gaps

Reanalyses

• “Optimal” combination of information from model and observations

• Dynamical interpolation

• There are known deficiencies• Temporal consistency is a challenge

• Dynamical balance in specific regions is difficult

• Large variety of ocean reanalyses (ORAs)• ORA-IP intercomparison project

• Multi-ORA real time monitoring (NOAA-BoM)

• European GREP by CMEMS

• Continuous improvements

Real Time Monitoring with multiple ocean reanalyses

Temperature and ENSO monitoring, NOAA, Xue et al 2017

http://www.cpc.ncep.noaa.gov/products/GODAS/multiora_body.html

4MAGDALENA A. BALMASEDA IV INTERNATIONAL ENSO CONFERENCE- GUAYAQUIL 15-18 OCTOBER 2018

Beyond Monitoring: characterizing ENSO energetics

October 29, 2014

Contributors to tropical Pacific OHC evolution

5EUROPEAN CENTRE FOR MEDIUM-RANGE WEATHER FORECASTS

• Marked differences in Indonesian Throughflow heat transport and surface heat flux

• Differences in surface fluxes related to increased absorbed solar radiation in 2015/16

Map of 0-300m 2-yearly OHC changes (in 109Jm-2) and accumulated heat

fluxes (in ZJ) during El Nino events From Mayer et al 2017

October 29, 2014

Indonesian Throughflow transports

6EUROPEAN CENTRE FOR MEDIUM-RANGE WEATHER FORECASTS

• Record-low ITF heat transport during 2015/16

• Mainly driven by reduced volume fluxes

• Extreme ITF behaviour in 2015/16 mainly explained by

Indian Ocean - shift towards positive SLA in Indian Ocean

from ~2008

From Mayer et al 2017

• Recent OHC evolution consistent with SLA anomalies

• Rapid Indian Ocean warming has been attributed to

negative state of Pacific Decadal Oscillation

October 29, 2014

Radiation at top-of-the-atmosphere

9EUROPEAN CENTRE FOR MEDIUM-RANGE WEATHER FORECASTS

• ASR and OLR anomalies extended far less to the east in 2015/16

• Positive ASR anomalies in subtropics in 2015/16, but not present in 1997/98

• ASR anomalies dominate subtropical net radiation anomalies

All fluxes positive downward

October 29, 2014

Advancing the forecast capabilities: new ECMWF SEAS5

10EUROPEAN CENTRE FOR MEDIUM-RANGE WEATHER FORECASTS

SEAS5 Contributes to the Copernicus Climate Change Service (C3S) seasonal multi-model.

Johnson et al 2018, submitted

Stockdale et al 2018, in preparation

RMS errorReduced cold-tongue bias,

improved ENSO variability

increased forecast skill

attributed to better

atmospheric model and

increased ocean resolution

https://climate.copernicus.eu/seasonal-forecasts

ACC

Variability ration (model/obs)SEAS5 SEAS4

Steady and significant progress

Persistence

SEAS2

ensemble spread

RMS error of Nino3 SST anomalies

EUROSIP 2005

SEAS5Bayesian Calibration

2002 - SEAS2

2017 – SEAS5

Prediction of the onset of ENSO has benefited from the reduction of model biases – more realistic air

coupled feedbacks – better simulation of westerly wind bursts.

ENSO and intraseasonal variability

20 years or progress in ENSO prediction at ECMWF

and contribution of ocean observations

13

2.83

3.53.8

4.5

S1 S2 S3 S4 S5

Le

ad

Tim

e (

mo

nth

s)

Forecast lead month for correlation above 0.9 in NINO3.4 SST anomalies

• S1 was the first ECMWF seasonal forecasting system. Implemented as a pilot in 1997

• SEAS5 is the latest ECMWF seasonal forecasting system. Implemented in November 2017.

Contributes to Copernicus Climate Change Services C3S.

2.8 33.5

3.84.5

3.2

S1 S2 S3 S4 S5 S5-NoOobs

Gain about 2 months in

ENSO prediction

What if we did not have

ocean observations?

We would lose about 15

years of progress.

1997

• System 1

2002

• System 2

2006

• System 3

2011

• System 4

2017• SEAS5

SEAS5 SST normalized forecast error

verifying in August from May starts

But worrying overprediction in latest years continues

This flow dependent error appears related with low

frequency variability in cross-equatorial winds

Trend in ORAS5 meridional wind stress

(normalized anomalies)

Differences in Trends: SEAS5- ORAS5

MAGDALENA A. BALMASEDA

ERA5

Current reanalysis products at ECMWF

15

ERA-Interim

ORAS4

LAND

ATMOSPHERE

WAVE

OCEAN

Centennial

ERA-20CM/20C

Experimental reanalyses: Centennial and Coupled

LAND

ATMOSPHERE

WAVE

LAND

ATMOSPHERE

WAVE

ORAS5

OCEAN ICE

ORA-20C

OCEAN ICE

Coupled CentennialCERA-20C

LAND

ATMOSPHERE

WAVE

OCEAN ICE

Coupled High-ResCERA-SAT

LAND

ATMOSPHERE

WAVE

OCEAN ICE

Recent period

October 29, 2014

16MAGDALENA A. BALMASEDA GOOS WEBMINAR 9-JULY- 2018

Centennial reanalyses provide experimental records to study decadal modulations of ENSO

16

From Laloyaux et al 2018

Multivariate ENSO Index Equatorial Pacific Ocean Heat Content

(upper 300m)

The ECMWF Coupled Reanalyses effortsensemble of earth system data assimilation

Observation uncertainties

Model uncertainties

Model forcing

uncertainties

Reanalysis uncertainties

Ocean temperature obs

assimilated in CERA-20C

1901

1950

2010

Laloyaux et al 2018) De Boisseson et al 2017

ERA-20CCERA-20C

TIWs in Coupled/Uncoupled

reanalysis of the 20th-Century

CERA-20C

CERA-SAT

1900s

1940s

1970s

2000s

ORA-20C CERA-20CSolar radiationDecadal variations

of uncertainty

October 29, 2014

Seasonal Reforecast data set spanning 1900-2010.

19

10 ens

24 months

May starts

1950 -1980

May starts

1981 -2010

October 29, 2014

Summary

• Increasing monitoring capabilities, enabling the characterization of ENSO energy budget

Tropical Pacific end of 2016 was warmer than before the 2015/16 El Nino – in stark contrast to cooling associated

with earlier El Ninos

– Indonesian Throughflow, pre-conditioned by strong Indian Ocean warming in recent decade

– Absorbed solar radiation, related to unusually warm subtropical Pacific SSTs

• SEAS5: Continuous and Steady progress in ENSO forecast: resolution+physics+ocean initialization

– Overprediction in recent years associated with inability to capture the meridional asymmetry in Eastern Pacific

– SEAS5 available in C3S database

• Centennial reanalyses: an asset for understanding/attributing/predicting ENSO

– Uncoupled ERA-20C/ORA-20C and Coupled CERA-20C

– SEAS5-20c: A set of extended seasonal (24m) reforecast spanning the XX-Century with the same model

version as operational SEAS5 (at low resolution)

20EUROPEAN CENTRE FOR MEDIUM-RANGE WEATHER FORECASTS

EXTRA SLIDES

21

October 29, 2014

Coupled Data Assimilation:It all started with some pilot projects; CERA and ERA-CLIM2

CERA-20C: A coupled reanalysis of the 20th century

CERA-SAT: A coupled reanalysis at higher resolution

Produce global reanalyses to reconstruct the past climate/weather of the earth system