© crown copyright met office clivar climate of the 20 th century project adam scaife, chris...

© Crown copyright Met Office

CLIVAR Climate of the 20th Century Project

Adam Scaife, Chris Folland, Jim Kinter, David Fereday

January 2009


• Aim

• Characterize and understand variability and predictability of climate over the past ~130 years associated with slowly varying forcing functions including SST

• History

• Initiated by Hadley Centre in 1993

• Now jointly lead by Hadley Centre (Folland) and COLA - Center for Ocean Land Atmosphere studies (Kinter)

• CLIVAR project & reports to WMO/CAS/WGNE

• Workshops: Hadley 1994, COLA 2002, ICTP 2004, Hadley 2007


• Experimental Design

• Initially focused on ensembles of AGCM simulations, at least 4 members

• All forced with same HadISST sea surface temperature and sea ice analysis

• Longer timescale than other intercomparisons such as AMIP: 1871 onwards

• Focus is on climate variability and predictability rather than model evaluation


• Experimental Design cont….

• Expanded to include other forcing data sets, including greenhouse gases, ozone, volcanic aerosols and solar variability

• Recent extensions:

• “Pacemaker” experiments with coupled models in order to more accurately simulate variability that is inherently coupled

• Land surface forcing, interaction with LUCID Land Use

and Climate – IDentification of robust impacts (De Noblet et al)

• More highly resolved SST to be available later this year: HadISST2


Reproducing climate variability

Land Surface Temperature Sahel Rainfall

Southern Oscillation North Atlantic Oscillation


Multi-Model Comparisons e.g. Evaluation of Climate Events:

20th Century ClimateEvent

(e.g. surface T trend)

Consistent with ensemble means?

Potentially predictable,“forced” and well

modelled

Consistent withensemble members?

Unpredictable internal variation but well

modelled

Poorly modelled in thisexperiment:

missing process/forcing

YES

NO

NO

YES


Predictable interdecadal trends?

Ensemble Means

Ensemble Members

Land Surface T: 1970-2000 Sahel Rainfall: 1950-1980

Sahel Rainfall: 1950-1980

NAO: 1965-1995

NAO: 1965-1995


Selected results: Increase in predictability of boreal winter land temperature, using two models -

mainly caused by decadal changes in ENSO variability

Kang et al, 2006, GRL, highlighted


Selected results: Winter NAO and the stratosphere

Change in NAO index Change in surface pressure

Model also forced with HadISST and all known major forcings in C20C mode.

Full NAO and surface climate change 1965-95 reproduced

Scaife et al, 2005, GRL


North Atlantic Oscillation in Summer

SNAO pattern

• Biggest single atmospheric circulation influence on summer climate in N W Europe/UK.

• Related to summer storm track – like 2007/8 flood or 1976 drought in UK.

• Related to ENSO SSTs, West African Monsoon and climate worldwide on decadal time scales, perhaps via AMO

• Current phase: (a) better understand mechanisms of SNAO links to atmospheric circulation and forcings, e.g. W. African summer monsoon. (b) Investigate SNAO seasonal predictability.

Rainfall correlations1900-1998

CFFolland et al. 2008


Selected results: Simulating Indian Monsoon Rainfall (IMR) and causes of its decadal variations

Interannual ensemble means, ENSL (1902-1999; black) and CRU (red), mm/day

Decadal IMR of CRU (red) and the ensemble means of C20C (black), mm/day

Kucharski et al. 2008

© Crown copyright Met Office Schubert et al. 2004

Selected results: Simulating Dust Bowl era drought


Atlantic hurricanes in C20C simulations

Storm counts (obs, model)


Pacemaker Experiments

• Motivation:

• On seasonal time scales, there is large-scale atmosphere-ocean covariability (e.g. ENSO-monsoon)

• There is also local atmosphere-ocean coupling

• Latent heat flux – SST

• Rainfall – SST

• Lag-lead relationships

• None of these processes are well represented (often wrong sign) in typical AGCM simulations with global prescribed SST

• CGCM Pacemaker Strategy

• Specify SST only where it drives the atmosphere, and model the ocean (slab or dynamic) elsewhere

• Main example: prescribe SST in tropical eastern Pacific (Lau and Nath, 2003)

• Allow for coupled feedbacks outside region of specified SST

• Test importance of thermodynamic vs. dynamic coupling

• Some experiments with mixed-layer (slab) ocean models

• Some experiments with dynamic ocean models


Pacemaker Strategy: Overcoming Shortcomings of AGCMS and Coupled Models

The “Pacemaker” strategy permits a consistent air-sea energy balance while simultaneously including the time sequence of climate-driver events, such as ENSO.

Teleconnections from the eastern tropical Pacific to remote tropical and extratropical regions are well represented in pacemaker runs, e.g., phenomena that are at once driven by and independent of ENSO,

like the Asian monsoon.

DJF SST Composite (El Nino - La Nina)

Observed

Pacemaker

PacemakerObserved

JJA Rainfall Composite (El Nino - La Nina)

Cash et al. 2007

Pacemaker design: specified SST

regions


Evolving C20C Experimental Design

• Pacemaker

• Specified SST in limited region (e.g. eastern tropical Pacific or north Atlantic)

• Thermodynamic ocean (slab or mixed layer formulation with Q-flux)

• Dynamical ocean models in some basins

• Land Use and Change

• Coordination with LUCID

• Phenomena-Focused Experiments

• subsets of C20C group

• West African Monsoon Modeling and Evaluation (WAMME)

• Asian monsoon

• Influence of the stratosphere on seasonal predictability


Some Lessons Learned• Collaborative data analysis sometimes works better

than large on-line databases

• Beware of normalised indices

• Normalised, ensemble mean anomalies can give the impression of reproducible and potentially predictable anomalies, when members do not even span the observations.

Absolute Sahel Rainfall Normalised Sahel Rainfall


Available Diagnostics

• Large selection of data available:• PMSL, T, RH, Z, precip, U, V, w, cloud, heat flux, wind stress, soil moisture

• All monthly and some daily diagnostics

• Data available on line from COLA, HADLEY, SNU, GSFC

• http://www.iges.org/c20c/sharing_data.html


Link with WGSIP on potential predictability?

• C20C forcing datasets available

• Use C20C data as a limit to predictability?

• Decadal climate events

• Earlier hindcasts? • Pre-1979• Atm. analyses from 1891(Compo et al)

© crown copyright met office clivar climate of the 20 th century project adam scaife, chris...

Documents