Skip to main content
U.S. flag

An official website of the United States government

Publication Date
25 May 2024

Exploring the Relative Importance of the MJO and ENSO to North Pacific Subseasonal Predictability

Print / PDF
Powerpoint Slide

Using an interpretable neural network, we are able to decompose the relative contributions of the MJO and ENSO to boreal winter subseasonal predictability of atmospheric circulation over the North Pacific. We find that the state of ENSO is more important for subseasonal predictability of circulation over this region than the MJO in the CESM2 preindustrial simulations for a variety of lead times and predictand temporal averaging windows. However, we also identify specific types of MJO events useful for prediction in this region when ENSO is in a neutral state.


The Madden-Julian Oscillation (MJO) can influence mid-latitude circulation on subseasonal timescales by exciting Rossby waves manifesting as a PNA-like signal, where the Aleutian Low limb of the PNA is particularly responsible for downstream effects on North American weather. Beyond influencing subseasonal variability through the MJO, the El Nino Southern Oscillation (ENSO) also influences midlatitude subseasonal variability over the Aleutian Low region through its teleconnections during
boreal winter due to the within-season evolution of the background flow (mid-latitude jet). This raises the question about the relative importance of these two tropical modes of variability in subseasonal predictability over the North Pacific. In the CESM2 preindustrial simulation, we find that the state of ENSO alone is overall a more important contributor to the prediction skill of the network than the MJO in this region, demonstrating that ENSO is a particularly valuable predictor for subseasonal lead times as well.


We explore the importance of the state of two tropical modes of variability (the MJO and ENSO) on the prediction of midlatitude atmospheric circulation over the North Pacific on subseasonal lead times in the CESM2 preindustrial simulations. During network-identified forecasts of opportunity (periods of enhanced predictability), the model more frequently uses the ENSO network to correctly predict the sign of the circulation anomaly than the MJO network on the majority of subseasonal lead times. Scrambling information about the state of ENSO also reduces prediction skill of the network 2 to 6 times more than when information about the MJO is shuffled. Nevertheless, during ENSO neutral conditions, we find specific MJO states that provide useful information for predicting atmospheric circulation over the North
Pacific, including persistent and anomalously strong MJO events.

Point of Contact
Kirsten Mayer
National Center for Atmospheric Research (NCAR)
Funding Program Area(s)