Multivariate ENSO Index (MEI)

Last update: 6 April 2009

The views expressed are those of the author and do not necessarily represent those of NOAA.

El Niño/Southern Oscillation (ENSO) is the most important coupled ocean-atmosphere phenomenon to cause global climate variability on interannual time scales. Here we attempt to monitor ENSO by basing the Multivariate ENSO Index (MEI) on the six main observed variables over the tropical Pacific. These six variables are: sea-level pressure (P), zonal (U) and meridional (V) components of the surface wind, sea surface temperature (S), surface air temperature (A), and total cloudiness fraction of the sky (C). These observations have been collected and published in COADS for many years. The MEI is computed separately for each of twelve sliding bi-monthly seasons (Dec/Jan, Jan/Feb,..., Nov/Dec). After spatially filtering the individual fields into clusters (Wolter, 1987), the MEI is calculated as the first unrotated Principal Component (PC) of all six observed fields combined. This is accomplished by normalizing the total variance of each field first, and then performing the extraction of the first PC on the co-variance matrix of the combined fields (Wolter and Timlin, 1993). In order to keep the MEI comparable, all seasonal values are standardized with respect to each season and to the 1950-93 reference period. The MEI is extended during the first week of the following month based on near-real time marine ship and buoy observations (courtesy of Diane Stokes at NCEP) summarized into COADS-compatible 2-degree monthly statistics at NOAA-ESRL PSD. Caution should be exercised when interpreting the MEI on a month-to-month basis, since the input data for updates are not as reliable as COADS, and the MEI has been developed mainly for research purposes. Negative values of the MEI represent the cold ENSO phase, a.k.a.La Niña, while positive MEI values represent the warm ENSO phase (El Niño).

You can find the numerical values of the MEI timeseries under this link, and historic ranks under this related link. You are welcome to use any of the figures or data from the MEI websites, but proper acknowledgment would be appreciated. Please refer to the (Wolter and Timlin, 1993, 1998) papers (NOW available online as pdf files!), and/or this webpage.

If you have trouble getting the data, please contact me under (Klaus.Wolter@noaa.gov)

How does the 1998-2000 La Niña event compare against the seven previous biggest La Niña events since 1949? Only strong events (with a peak value of at least -1.2 sigma) are included in this figure. Note that some events last through the full three years shown here (for instance, 54-56), while others revert to "normal" or El Niño conditions by the second or third year (especially in 64-66). The 1998-2000 La Niña does not resemble any previous event in this comparison figure. It started late (about three months later than the previous latest case), and it featured a superimposed annual cycle (peaking around May and troughing around November) that does not match the other events displayed in this figure. However, the weak La Niña period after the 1982-83 El Niño had similar characteristics. Click on the "Discussion" button below to find a comparison of recent MEI conditions against a large subset of these La Niña events.

  • Discussion and comparison of recent conditions with historic El Niño events

    • How does the 2002-04 El Niño event compare against the seven previous biggest El Niño events since 1949? Aside from 2002-04, only strong events (with a peak value of at least +1.4 sigma) are included in this figure. The 2002-03 El Niño event peaked below that threshold, with just over +1.2 sigma in early 2003. Overall, I would rank it just barely in the top 10 El Niño events of the last half century. In its evolution, it bears some resemblance to the 1965-67 event (highest temporal correlation), but shared with 1991-93 its reluctance to drop below the zero line once it had run its course. The most recent El Niño event of 2006-07 reached a similar peak as the 2002-03 event, but lacked 'staying power', and collapsed rather early in 2007. I have discontinued its comparison plot against other weaker events in favor of a new comparison with other La Niña events below (click on "Discussion" to jump to the bottom of the page).

    The six loading fields show the correlations between the local anomalies and the MEI time series. Land areas are flagged in green, and typically noisy regions with no coherent structures and/or lack of data are shown in grey. Each field is denoted by a single capitalized letter and the explained variance for the same field in the Australian corner.

    The sea level pressure (P) loadings show the familiar signature of the Southern Oscillation: low pressure anomalies in the west and high pressure anomalies in the east correspond to negative MEI values, or La Niña-like conditions. Consistent with P, U has positive loadings along and just south of the Equator, corresponding to easterly anomalies near the dateline. The meridional wind field (V) features negative loadings north of the Equator across most of the Pacific basin, denoting the northward shift of the ITCZ so common during La Niña conditions, juxtaposed with high positive loadings northeast of Australia.

    Both sea (S) and air (A) surface temperature fields exhibit the typical ENSO signature of a wedge of positive loadings stretching from the Central and South American coast to the dateline, or cold anomalies during a La Niña event. They are flanked by a horse-shoe pattern of negative loadings (warm anomalies) to their north- and southwest. During La Niña conditions, total cloudiness (C) tends to be decreased over the central equatorial Pacific, as opposed to increased cloudiness over the Philippines and north of Australia.

    The MEI continues its retreat from its peak explained variance (now just under 26%) of all six fields in the tropical Pacific from 30N to 30S. The loading patterns shown here resemble the seasonal composite anomaly fields of Year 1 in Rasmusson and Carpenter (1982).

    Consistent with re-emerging La Niña conditions, there are several of the observed key anomalies in the MEI component fields that exceed or equal one standard deviation, or one sigma (compare to loadings figure) that flag typical La Niña features, whle the opposite camp (El Niño) is not represented by a single such anomaly. Significant negative anomalies (coinciding with high positive loadings) denote strong easterly anomalies (U) along the Equator and near the dateline, as well as northerly wind anomalies (reduced southerly flow) east of Australia (V). Significant positive anomalies (coinciding with high negative loadings) flag anomalously high sea level presssure over the eastern equatorial Pacific (P), and positive sea surface and air temperature (S and A) anomalies over the subtropical central South Pacific and east of the Philippines. Compared to last month, the number and strength of these significant anomalies is reduced, but not enough to change the overall MEI value.

    Go to the discussion below for more information on the current situation.

    If you prefer to look at anomaly maps without the clustering filter, check out the climate products map room.

    Discussion and comparison of recent conditions with historic El Niño events

    In the context of La Niña-like conditions for much of the last two years, this section displays a comparison figure of historic La Niña events since 1949 vs. current conditions. Compared to an earlier version of this plot, I picked 'analog' cases to include all La Niña events that lasted at least two boreal winter seasons, with at least one winter reaching moderate intensity. The most recent (February-March) MEI value has dropped by 1/100 to reach -.74. This is just slightly above the 1973-75 'analog' case, which remains the most similar stretch to 2007-09 in the MEI record, correlating at +0.88 for the last 27 months. In 1975, this was followed by continued La Niña conditions lasting for another year. The most recent MEI rank has not changed (15th lowest), still exceeding the weak La Niña threshold in the MEI rankings for this season. The MEI has thus continued the status of weak La Niña conditions that it resumed in August-September 2008.

    Moderate negative SST anomalies cover most of the equatorial Pacific from just west of the dateline to the South American coast. However, pockets of warmer-than-average SST keep reappearing closer to the South American coast, including one equatorial patch near 90W, at least as reflected in the latest weekly SST map.

    For an alternate interpretation of the current situation, I highly recommend reading the latest NOAA ENSO Advisory which represents the official and most recent Climate Prediction Center opinion on this subject. In its latest update (5 March 2009), current La Niña conditions are expected "to gradually weaken" over the next few months. While the latest weekly Niño 3.4-based data confirm this weakening for the SST-based ENSO indices, the longer-term MEI data has shown remarkable stability for the last several months.

    There are several other ENSO indices that are kept up-to-date on the web. Several of these are tracked at the NCEP website that is usually updated around the same time as the MEI, just in time for this go-around. Niño regions 3 and 3.4 showed persistent temperature anomalies of -1.3C or lower from September (region 3) and October 2007 (region 3.4) through February 2008, but hovered much closer to 0C anomalies between about June and November 2008. A dramatic cool-off ensued, leading to monthly SST anomalies of around -1C in Niño region 3.4, and Niño region 3 anomalies of near -0.6C during January 2009. Interestingly, the latter cooled off further by March (-0.8C), while Niño 3.4 SST have rebounded to -0.55C. For extended Tahiti-Darwin SOI data back to 1876, and timely monthly updates, check the Australian Bureau of Meteorology website. This index belatedly caught on to the La Niña event in November 2007, reaching +12 to +14 (1.4 sigma) in December, January and March, and higher yet (+21) in February 2008. The average for February-March (+16.8) was ranked fourth highest for that season (since 1950, but this ranking is true for the full record as well). It then hovered near 0 from April through July, while the August SOI rebounded right up to +9, followed by values at or above that value right through February 2009. However, the March value of just above 0 (+0.2) constitutes a drop of almost 1.5 standard deviations compared to the previous month. Until last month, a remarkably strong run in La Niña territory, easily the index with the strongest La Niña footprint this season. An ever longer Tahiti-Darwin SOI (back to 1866) is maintained at the Climate Research Unit of the University of East Anglia website, however with less frequent updates (finally updated through December 2008). Extended SST-based ENSO data can be found at the University of Washington-JISAO website, currently updated through September 2008.

    Stay tuned for the next update (by May 9th) to see where the MEI will be heading next. We have moved into the time of year when rapid changes of the MEI are more common than earlier. The majority of ENSO-forecasts from last month indicate weakening La Niña conditions during the upcoming months (based on SST only). It will be interesting to see which way the MEI will go - based on the current ranking alone, three of the 10 closest-ranked analog years remained stuck in La Niña mode, while three other analog cases evolved into El Niño events, and the remaining four ending up ENSO-neutral. Interestingly, all three La Niña cases started out slightly stronger than the current MEI value, while all three El Niño cases started out slightly weaker. In the near-term, I would expect the MEI to follow the lead of the SOI and relax towards 'normal' in the next few months. This does not exclude the return of more La Niña-like conditions later this year. Note that I have discontinued my monthly e-mail announcements in favor of keeping the discussion right here on this webpage.

    REFERENCES

    • Rasmusson, E.G., and T.H. Carpenter, 1982: Variations in tropical sea surface temperature and surface wind fields associated with the Southern Oscillation/El Niño. Mon. Wea. Rev., 110, 354-384. Available from the AMS.
    • Wolter, K., 1987: The Southern Oscillation in surface circulation and climate over the tropical Atlantic, Eastern Pacific, and Indian Oceans as captured by cluster analysis. J. Climate Appl. Meteor., 26, 540-558. Available from the AMS.
    • Wolter, K., and M.S. Timlin, 1993: Monitoring ENSO in COADS with a seasonally adjusted principal component index. Proc. of the 17th Climate Diagnostics Workshop, Norman, OK, NOAA/NMC/CAC, NSSL, Oklahoma Clim. Survey, CIMMS and the School of Meteor., Univ. of Oklahoma, 52-57. Download PDF.
    • Wolter, K., and M. S. Timlin, 1998: Measuring the strength of ENSO events - how does 1997/98 rank? Weather, 53, 315-324. Download PDF.

    Questions about the MEI and its interpretation should be addressed to:
    (Klaus.Wolter@noaa.gov), (303) 497-6340.