CODA Lab Logo

A Statistical ENSO Prediction Model

Computational Oceanography and Dynamics of Air-sea Interaction Laboratory

Model Forecast

Our physical-based ENSO prediction model (EPM) employs a multivariate linear regression framework utilizing four key predictors:

  • EPISST: The Niño3.4 index derived from OISST.
  • EPIWWV: A warm water volume index, calculated from D20 in the tropical Pacific.
  • EPIOA: An Ocean-Atmosphere coupling index that integrates zonal wind and warm water volume in the tropical Pacific.
  • EPISLP: Defined by extratropical sea level pressure patterns, including NPO-like variability in North Hemisphere, and PSA-like variability in South Hemisphere.

Our EPM allows for the evaluation of individual factor contributions to ENSO events by analyzing different combinations of predictors. The model has been trained on data from 1982 to the most recent year of observations. For more information, please see our publications below.

Plume of Model ENSO Predictions
Plume of Model ENSO Predictions

Two types of ENSO Preditions

Niño3 (orange line) and Niño4 (yellow line) indices using the same approach previously. Note that the North and South hemisphere of EPISLP are used separately.

SST reconstucttion

Reconstructed SST Patterns

SST patterns reconstructed using a multivariate linear regression model based on ERSST and the Niño3 and Niño4 indices from different months. Future SST patterns are derived from our predicted Niño3 and Niño4 indices using this regression model.

Current Status (latest observation)

Nino Index time series

ENSO Index Time Series

Niño3.4 (black line) and Niño1+2 (gray line) indices derived from OISST (climo.: 1991-2020), 3-month running mean

R-O phase diagram

SST-WWV Phase Diagram

  • WWV: D20 anomaly in 5°S–5°N, 120°E–80°W, 3-month running mean
  • Triangle: January; Red star: present
  • Color: blue (2 years ago) to green (present)
SST / Wind / SLP anomaly

SST / Wind / SLP anomaly

  • SST: OISST (climo.: 1991-2020), shading
  • Wind: NCEP-NCAR Reanalysis (climo.: 1980-2024), quiver
  • SLP: NCEP-NCAR Reanalysis (climo.: 1980-2024), contour (solid line=[1,3,5...]; dashed line=[-1,-3,-5...]). Green boxes indicate the location used to define EPISLP

Time Series

EPISST, EPIOA and EPISLP. EPISLP includes the North Hemisphere (NH, solid line) and South Hemisphere (SH, dashed line) partitions.

Equatorial Pacific profile and D20 anomaly

Equatorial (2°S-2°N) Profile of Temperature anomaly

  • Temperature: GODAS (climo.: 1980-2024), shading
  • 20°C isothermal: present (dashed line) and climatology (solid line)

D20 / Wind anomaly

  • D20: GODAS (climo.: 1980-2024), shading
  • Wind: NCEP-NCAR Reanalysis (climo.: 1980-2024), quiver

Time Series

EPIWWV and EPIOA.

Hovmöller Diagram of SST

Hovmöller Diagram of SST and Zonal wind anomaly

  • SST: OISST (climo.: 1991-2020), shading
  • Zonal Wind: NCEP-NCAR Reanalysis (climo.: 1980-2024), contour

All variables are averaged between 5°S and 5°N.

Hovmöller Diagram of D20

Hovmöller Diagram of D20 and Zonal wind anomaly

  • D20: GODAS (climo.: 1980-2024), shading
  • Zonal Wind: NCEP-NCAR Reanalysis (climo.: 1980-2024), contour

All variables are averaged between 5°S and 5°N.

Validation

ENSO Hindcast

ENSO Hindcast

  • Observed Niño3.4 index derived from OISST (climo.: 1991-2020), 3-month running mean.
  • Predicted Niño3.4 index from the EPM. Colored lines represent different ENSO events: first-year El Niño (red), second-year El Niño (purple), first-year La Niña (blue) and second/third-year La Niña (green). Each prediction is trained on historical data from 1982 up to the year immediately preceding the forecast.
Prediction Correlation

Correlation Analysis

Correlation coefficients between 3-month running mean observed and predicted Niño3.4 index.

Publications

Chen, H. C., Tseng, Y. H., Huang, J. H., & Juang, P. H. (2025). Understanding the driving mechanisms behind triple-dip La Niñas: insights from the prediction perspective. npj Climate and Atmospheric Science, 8(1), 143. [doi link]

Tseng, Y. H., Huang, J. H., & Chen, H. C. (2022). Improving the predictability of two types of ENSO by the characteristics of extratropical precursors. Geophysical Research Letters, 49(3), e2021GL097190. [doi link]

Chen, H. C., Tseng, Y. H., Hu, Z. Z., & Ding, R. (2020). Enhancing the ENSO predictability beyond the spring barrier. Scientific Reports, 10(1), 984. [doi link]

Tseng, Y. H., Hu, Z. Z., Ding, R., & Chen, H. C. (2017). An ENSO prediction approach based on ocean conditions and ocean–atmosphere coupling. Climate Dynamics, 48, 2025-2044. [doi link]