Multivariate ENSO Index (MEI)

Last update: 3 May 2008

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 CDC. 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 at 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 near the Equator across the Central 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 just east of the dateline, or cold anomalies during a La Niña event. During La Niña conditions, total cloudiness (C) tends to be decreased over the central and eastern equatorial Pacific, as opposed to increased cloudiness from the Phillippines to Hawaii.

    The MEI continues its retreat from its peak explained variance (now just 20%) 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 moderating La Niña conditions, there are still quite a few observed key anomalies in the MEI component fields that exceed or equal one standard deviation, or one sigma (compare to loadings figure), with most of them flagging La Niña conditions, and lower anomalies in many cases than last month. Significant positive anomalies (coinciding with high negative loadings) denote anomalously high sea level pressure (P) from Tahiti to Galapagos, southerly wind anomalies (V) north of New Guinea, and increased cloudiness (C) near the Philippines. Significant negative anomalies that flag La Niña in regions with high positive loadings are easterly wind anomalies (U) near the equatorial dateline (still in excess of TWO sigma), northerly wind anomalies east of Australia and over the Philippines, and negative temperature anomalies (S and A) in and near the equatorial cold tongue (Central Eastern Pacific). Positive SST anomalies (S) near the coast of South America and enhanced cloudiness (C) in the same region are more typically associated with El Niño rather than La Niña.

    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 recent La Niña conditions, this section displays a comparison figure of historic La Niña events since 1949 vs. current conditions. The most recent (March-April) MEI value has decreased its (negative) amplitude dramatically from last month's -1.55 down to -.89, a rise in the MEI that was exceeded only in 1951 and 1997 (for this time of year). Note that those two cases heralded a transition to El Niño conditions within a few months, while the 4th biggest one-month rise in 2000 ended up being followed by renewed La Niña conditions later that year. The MEI rank has dropped from being the 3rd strongest La Niña in February-March to now being the 12th strongest, right at the moderate La Niña threshold for this season.

    Negative SST anomalies cover the equatorial Pacific from just east of New Guinea to just west of Galapagos, but have weakened considerably over the last few months, while positive anomalies near South American coast indicate what could be considered a LOCAL El Niño situation in the latest weekly SST map. Basin-wide La Niña conditions have been in place since about August 2007.

    For an alternate interpretation of the current situation, I 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 (April 10, 2008), moderate La Niña conditions were diagnosed and expected to continue, albeit at a weaker levvel, through at least June 2008.

    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 (not in time for this month). Niño regions 3 and 3.4 have shown persistent temperature anomalies of -1.3C or lower from September (region 3) and October (region 3.4) through February. The last time region 3.4 featured negative anomalies as big as the February value (-1.9C) was in January 1989. However, Niño 3 and 3.4 have both warmed up by about 0.7C in terms of their anomalies, now at -0.6C and -1.1C, respectively. Weekly April SST data show further warming, to about -.2C and -.8C, respectively. For extended Tahiti-Darwin SOI data back to 1876, and timely monthly updates, check the Australian Bureau of Meteorology website. This index caught on to the current La Niña event last November, reaching +12 to +14 (1.4 sigma) in December, January and March, and higher yet (+21) in February. The average for February-March (+16.8) was ranked fourth for that season (since 1950 which happens to also hold for the full record). However, the April SOI weakened to under +5. The SOI is now in sync with other ENSO indices in terms of assessing this event, following an extended period of apparently erratic behavior from about the middle of 2006 to the end of 2007. 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 (currently more than one year behind). Extended SST-based ENSO data can be found at the University of Washington-JISAO website, currently updated through April 2007.

    Stay tuned for the next update (by June 7th) to see where the MEI will be heading next. While persistence has been hard to beat for much of the fall and early winter, the dramatic drop earlier this year, and rapid rebound last month are consistent with weakening persistence this time of year. Continued La Niña conditions are still more likely than not, at least into northern summer, but a switch to El Niño is not out of the question by the fall season. As I pointed out in February, in those cases when the MEI has reached a top-10 La Niña ranking in the past, La Niña cases have often persisted for more than one year: from early 1950 into early 1951, mid-1954 into early 1957, early 1962 into early 1963, mid-1970 into early 1972, mid-1973 into early 1976, mid-1988 into mid-1989, and early 1999 into early 2001. By comparison, only 1964 and 1968 featured similar La Niña rankings (top 10) that did not result in La Niña conditions (rank 19 or lower) for more than a year. On average, analogous statistics for El Niño show shorter 'life expectancies'. So, while this event could be replaced by El Niño later this year, chances are at least as good that we will see continued/renewed La Niña conditions right into next winter. Time will tell. 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/N MC/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.