Generalized linear modeling of the El Niño/Southern Oscillation with application to seasonal forecasting and climate change projections

Authors

  • S. Stevenson,

    Corresponding author
    1. International Pacific Research Center, University of Hawaii, Honolulu, Hawaii, USA
    2. Department of Atmospheric and Oceanic Sciences, University of Colorado Boulder, Boulder, Colorado, USA
    • Corresponding author: S. Stevenson, International Pacific Research Center, University of Hawaii, Honolulu, HI 96822, USA. (slgs@hawaii.edu)

    Search for more papers by this author
  • B. Rajagopalan,

    1. Cooperative Institute for Research in the Environmental Sciences, University of Colorado Boulder, Boulder, Colorado, USA
    2. Department of Civil, Environmental and Architectural Engineering, University of Colorado Boulder, Boulder, Colorado, USA
    Search for more papers by this author
  • B. Fox-Kemper

    1. Department of Atmospheric and Oceanic Sciences, University of Colorado Boulder, Boulder, Colorado, USA
    2. Cooperative Institute for Research in the Environmental Sciences, University of Colorado Boulder, Boulder, Colorado, USA
    Search for more papers by this author

Abstract

[1] A new framework for simulating the El Niño/Southern Oscillation (ENSO) using a generalized linear model (GLM) is provided. The GLM provides a versatile and computationally inexpensive method for investigating ENSO dynamics, by conditioning an ENSO index on an arbitrary set of input variables. Here the system state (El Niño/neutral/La Niña) at previous times is combined with the first few principal components of sea surface temperature (SST) and thermocline depth. Despite having relatively few degrees of freedom, the model accurately reproduces 20th century SST time series, seasonal variance, power spectra, and autocorrelation functions for both the eastern and western Pacific. The GLM also has good overall forecast skill, especially at subyearly lead times; performance is competitive with models currently used for operational ENSO forecasting. The model is then used to examine changes to El Niño/La Niña statistics under CO2 increases, by using the GLM to represent simulations run with the National Center for Atmospheric Research (NCAR) Community Climate System Model (CCSM), versions 3.5 and 4. GLM simulations of 21st century CCSM4 changes to El Niño/La Niña magnitudes show insignificant results, despite a slight increase in El Niño persistence. GLM fits conditioned on millennial stabilized CCSM3.5 simulations with varying CO2 levels, however, show a weakening and shortening of El Niño events as CO2 concentration increases, whereas La Niña events become markedly stronger and do not change significantly in length. The reduction in El Niño persistence in CCSM3.5 is consistent with previous results showing that at higher CO2 levels, a stronger seasonal cycle creates a Southern Hemisphere “seasonal footprint” leading to more efficient El Niño termination.