Atlantic Multi-decadal Oscillation (AMO)

Teaser image
Atlantic Multi-decadal Oscillation (AMO)
Main content

The Atlantic Multi-decadal Oscillation (AMO) has been identified as a coherent mode of natural variability occurring in the North Atlantic Ocean with an estimated period of 60-80 years. It is based upon the average anomalies of sea surface temperatures (SST) in the North Atlantic basin, typically over 0-80N. To remove the climate change signal from the AMO index, users typically detrend the SST data at each gridpoint or detrend the spatially averaged timeseries. Trenberth and Shea (2006) recommend that the detrending be done by subtracting the global-mean SST anomaly timeseries from the spatially averaged timeseries. See the Expert Guidance by Dr. Kevin Trenberth for the rationale for the global-mean detrending approach. The Expert Guidance by Dr. Rong Zhang discusses the impacts and mechanisms of the AMO.

Key Strengths

Key Strengths

The SST-based AMO index provides a simple, concise way to describe multidecadal climate variability in the North Atlantic

Associated with important climate impacts, such as the multidecadal variability of Atlantic Hurricane activity, North American and European summer climate, northern hemispheric mean surface temperature, and Arctic sea ice anomalies

The AMO pattern is robust across different datasets (i.e. HadISST, ERSST, Kapan SST)

Key Limitations

Key Limitations

Instrumental SST data are short in length compared to the multidecadal timescale of the AMO

The SST-based AMO index does not directly capture aspects of the AMO related to coherent variations in salinity, subsurface temperature and ocean-driven turbulent heat fluxes

Quantifying the relative importance of large scale ocean circulation (especially AMOC) vs. external radiative forcing in causing the AMO is challenging

Please cite data sources, following the data providers' instructions
Suggested Data Citation
Dataset DOIs
None
Hosted Climate Index Files
  1. AMO Index monthly (10-yr lowpass) based on Trenberth and Shea (2006), HadISST1

    Missing Value
    -999

    Units
    degC

    Updated Through
    2021-03-01

    Next Update
    2022-03-01

    Start Date
    1870-01-01

    File Notes
    HadISST1 is the input SST dataset. The anomalies are with respect to the entire time period (1870-present). The Trenberth and Shea (2006) method has been applied to the monthly data. A 10-year running mean is applied to the monthly AMO Index to form the low-pass index. Due to the smoothing, missing values occur near the beginning and end of the low-pass index. This index will only be updated occasionally (approximately yearly).

    CAS Citation
    AMO Index Data provided by the Climate Analysis Section, NCAR, Boulder, USA, Trenberth and Shea (2006). Updated yearly. Accessed DD Month YYYY [list date you accessed the data].

  2. AMO Index monthly (raw) based on Trenberth and Shea (2006), HadISST1

    Missing Value
    -999

    Units
    degC

    Updated Through
    2021-03-01

    Next Update
    2022-03-01

    Start Date
    1870-01-01

    File Notes
    HadISST1 is the input dataset Trenberth and Shea (2006) method has been applied Anomalies, in degC, are relative to the entire time period of the input dataset. This index will only be updated occasionally (approximately yearly).

    CAS Citation
    AMO Index Data provided by the Climate Analysis Section, NCAR, Boulder, USA, Trenberth and Shea (2006). Updated yearly. Accessed DD Month YYYY [list date you accessed the data].

Usage Restrictions
None

Expert User Guidance

Expert User Guidance

The following was contributed by Dr. Rong Zhang, March, 2016:

The observed large-scale multidecadal fluctuations in the Atlantic sea surface temperature (SST), has been referred to as the Atlantic Multidecadal Oscillation (AMO) (Kerr, 2000) to emphasize the “multidecadal” character of this oceanic phenomenon and to distinguish it from the inter-annual variability associated with the atmospheric North Atlantic Oscillation (NAO) (Enfield et al. 2001). The AMO index is defined as an area average of detrended low-pass filtered North Atlantic SST anomalies to reflect the Atlantic low frequency variability at time scales longer than a decade (Enfield et al., 2001; Sutton and Hodson, 2005; Knight et al 2005; Ruiz-Barradas et al., 2013).

The AMO pattern (Figure 2a) is derived by regressing detrended SST anomalies on the AMO index (Figure 1) (Sutton and Hodson, 2005; Knight et al., 2005; Ruiz-Barradas et al., 2013).  The positive AMO phase corresponds with positive SST anomalies over most of the North Atlantic, with stronger anomalies in the subpolar region and weaker anomalies in the tropics (Figure 2a). The warm AMO phases occurred during the middle of the 20th century and the recent decades since 1995, and cold phases occurred during the early 20th century and during 1964-1995 (Figure 1). The monopolar AMO pattern associated with this multidecadal variability should be distinguished from the tripolar SST pattern induced by the NAO at the inter-annual time scale (Figure 2b), and hence the AMO pattern should not be derived by regressing on the detrended “unfiltered” North Atlantic basin averaged SST index (see Figure 2).

The AMO has significant regional and hemispheric climate impacts, such as modulating multidecadal variations in U.S. rainfall and drought frequency (Enfield et al., 2001; McCabe et al. 2004), summer climate over North America and Europe (Sutton and Hodson, 2005, 2007; Sutton and Dong, 2012), Atlantic Hurricane activity and India/Sahel summer rainfall (Folland et al., 1986; Goldenberg et al. 2001; Knight et al., 2006; Zhang and Delworth, 2006; Latif et al., 2007), northern hemispheric mean surface temperature (Zhang et al. 2007; Semenov et al., 2010), and Arctic sea ice (Mahajan et al. 2011; Day et al. 2012; Miles et al. 2014; Zhang, 2015; Yeager et al. 2015). Zhang and Delworth (2007) found that the AMO can contribute to the Pacific Decadal Oscillation (PDO) and the associated Pacific/North America (PNA) pattern at the multidecadal time scale, and suggested that a North Pacific regime shift (opposite to the 1976–77 shift) might occur several years after the mid 90’s shift in the observed AMO. The AMO is also found to have had a substantial influence on the equatorial Pacific decadal variability (Kucharski et al. 2015) and have modulated the El Niño Southern Oscillation (ENSO) variability at  low frequency (Dong and Sutton, 2007; Kang et al. 2014). Observational and modeling studies suggest that multidecadal variations of the atmospheric NAO are modulated by the AMO, with the positive AMO phase leading the negative winter NAO phase by a few years (Gastineau and  Frankignoul, 2012; Hodson et al. 2014). Recent modeling studies show that the simulated multidecadal winter NAO response to the AMO can be significantly improved by resolving the stratosphere and enhancing stratosphere/troposphere coupling (Omrani et al., 2014, 2016).  

The AMO is often thought to be driven by the variability of the Atlantic Meridional Overturning Circulation (AMOC) (Kushnir, 1994; Delworth and Mann, 2000; Knight et al., 2005; Latif et al., 2006), although some have suggested that the AMO is mainly driven by changes in anthropogenic radiative forcing (Mann and Emanuel, 2006). Various approaches have been proposed for the quantitative attribution of the AMO to an anthropogenic radiatively forced part and a part arising from natural variability (Trenberth and Shea, 2006; Kravtsov and Spannagle, 2008; Ting et al. 2009; DelSole et al. 2011; Wu et al. 2011). Recently it has been suggested that anthropogenic aerosols are a prime driver of the AMO using climate model simulations incorporating aerosol indirect effects (Booth et al., 2012). However, there are major discrepancies between the simulations in Booth et al. (2012) and observations in the North Atlantic upper-ocean heat content, in the spatial pattern of multidecadal SST changes within and outside the North Atlantic, and in the subpolar North Atlantic sea surface salinity (SSS), due to overestimated aerosol indirect effects (Zhang et al. 2013). Besides, the aerosol effects cannot account for the observed anticorrelation between detrended multidecadal surface and subsurface temperature variations in the tropical North Atlantic. These discrepancies cast considerable doubt on the claim that aerosols drive the bulk of the AMO. On the other hand, independent AMOC fingerprints derived from the observed subsurface ocean temperature indicate that the past AMOC variations are coherent with the observed AMO (Zhang, 2007, 2008; Wang and Zhang, 2013), bringing more evidence that the observed AMO is linked to AMOC variations rather than merely a 20th century artifact of changes in radiative forcing.

The SST-based definition of the AMO index often leads to the incomplete understanding of the AMO only in terms of North Atlantic SST anomalies. In contrast, the AMO actually reflects coherent multivariate low frequency variability observed in the Atlantic, including correlated variations in the subpolar North Atlantic heat content, salt content, and ocean-driven surface turbulent heat fluxes, as well as anticorrelated variations in the tropical North Atlantic subsurface temperature (Zhang 2007, 2008; Wang et al. 2010, Robson et al. 2012; Zhang et al. 2013; Gulev et al. 2013; Wang and Zhang, 2013). Mechanisms proposed for the AMO have to account for the observed coherent multivariate low frequency variability in the Atlantic, in addition to the low frequency North Atlantic SST variations. It is crucial to use multivariate metrics to understand the mechanisms causing the AMO.##

Expert Developer Guidance

Expert Developer Guidance

The following was contributed by Dr. Kevin Trenberth, August, 2015:

The increasing human influence on climate, mainly through changes in atmospheric composition, produce global warming.  Many variables and climate indices are therefore influenced by associated climate changes and any analysis of variance can be dominated by the recent trends.  It is therefore desirable to remove the trends to examine the underlying variability and associated patterns.  One approach has been to fit a linear trend to a time series of an index, but this is a seriously flawed procedure because (i) it depends on the length of the time series; (ii) it implies the trend goes on indefinitely into the future and into the past; and (iii) there is no such trend evident in the earlier part of the 20th Century.   A relatively simple approach instead is to use the global mean time series as a primary indicator of the non-stationary component, and remove that from each grid point prior to carrying out an analysis.  This approach has been shown to work effectively by using model results, where the cause of the non-stationarity is known.   A good example is the AMO, which is based on an index of area average SST over the North Atlantic from 0 to 60N (see Trenberth and Shea (2006)).  The PDO is also derived from SSTs where the global mean has been removed, and a determination of the first mode of variability for 20-70N over the North Pacific (see Trenberth and Fasullo (2013)).##

Key Figures

AMO index timeseries.

Observed AMO index, defined as detrended 10-year low-pass filtered annual mean area-averaged SST anomalies over the North Atlantic basin (0N-65N, 80W-0E), using HadISST dataset (Rayner et al. 2003) for the period 1870-2015.

AMO SST pattern compared with NAO SST pattern

(a) Observed AMO SST pattern, derived by regressing detrended North Atlantic annual mean SST anomalies on the observed AMO index (Figure 1), using HadISST dataset (Rayner et al. 2013) for the period 1870-2015. The regression corresponds to 1 standard deviation of the observed AMO index. (b) Observed NAO SST pattern (inverted), derived by regressing detrended North Atlantic annual mean SST anomalies (HadISST dataset) on the inverted detrended observed winter NAO index (Hurrell Station-Based DJFM NAO Index) for the period 1870-2015. The regression corresponds to 1 standard deviation of the inverted observed NAO index.

AMO index based on Trenberth and Shea (2006) and HadISST

Atlantic Multi-decadal Oscillation (AMO) 1870-2011 derived from HadISST and using the Trenberth and Shea (2006) definition. [Climate Data Guide; D. Shea]

Other Information

Earth system components and main variables
Type of data product

Years of record
to
Data time period extended
Yes, data set is extended
Timestep
Monthly, Decadal
Formats:
Input Data
None
Vertical Levels:
Missing Data Flag
None
Ocean or Land
Ocean Only
Spatial Resolution
None
Model Resolution (reanalysis)
None
Data Assimilation Method
None
Model Vintage (reanalysis)
None

Key Publications
Not available...