top of page
  • Moetasim Ashfaq

ENSO overreliance could be limiting seasonal predictability

The El Niño-Southern Oscillation (ENSO) stands out as the most influential naturally occurring climate force ever encountered by humankind. Its profound impact on global weather patterns is attributed to its extensive reach across the equatorial Pacific, spanning over 7,000 kilometers and marked by distinct sea surface temperature anomalies. ENSO reigns supreme among natural climate variability drivers, emerging as the undisputed leader in this race—an Earth system equivalent of the unparalleled Usain Bolt among natural forcings.

The daily sea surface temperature anomaly in the NOAA/NCEI 1/4 Degree Daily Optimum Interpolation Sea Surface Temperature (OISST) Analysis, Version 2.1, between January 01 and November 12, 2023.

Accurate representation of ENSO and its global teleconnections is crucial for all climate models. A model's reputation in the Climate Modeling League for future climate projections, also known as CMIP, depends on its ability to generate ENSO variability and teleconnections with reasonable accuracy. Only a few models achieve this level of accuracy. However, since the key focus in the CMIP League is understanding the Earth System response to anthropogenic forcing, a "model democracy" criterion takes the preference, and every model is equally allowed to contribute to the Intergovernmental Panel on Climate Change (IPCC) Assessment Reports. Performance-based model selection is primarily left to downstream users.

Things work differently in the Climate Modeling League for sub-seasonal to seasonal forecasting (S2S). Natural forcings in the Earth system have a more significant role at the shorter timescale of months to seasons. Therefore, accurately simulating ENSO characteristics is crucial for predicting seasonal outlooks. Only a few models that qualify for the CMIP League can meet the higher skill requirements needed to be eligible for the S2S League. This situation is similar to the League of Cricket-playing nations, where only a handful of teams can play all formats due to their exceptional performance.

Consequently, the most effective S2S forecasting systems are developed by modeling institutes that produce top-performing models in the CMIP League. However, despite the higher ability of climate models explicitly tuned for S2S predictions, the forecasting skill at the S2S scale is generally only "average." Why do the most capable modeling systems struggle to provide high fidelity at the S2S scale even when accurately modeling ENSO forcing?

To answer this question, we need to consider the Earth system in a way that acknowledges ENSO as the primary natural forcing while also recognizing the presence of other less impactful, yet still capable, naturally occurring forcings in the oceans and atmosphere. If, at any given location, a significant percent of the climate variability results from the combined influence of ENSO and these forcings, an accurate representation of ENSO alone cannot reliably predict the climate variability at the S2S scale.

Tropical Modes of Variability

At times, ENSO may be accompanied by other oceanic forcings, with ENSO having a minor or no role in their occurrences. Currently, a strong Indian Ocean Dipole (IOD) is co-occurring with a strong ENSO in the Pacific Ocean. IOD, a contrasting east-west sea surface temperature anomaly in the tropical Indian Ocean, typically peaks in October-November. However, an unusually strong IOD may persist through December this year. IOD's influence is well-known along the east coast of Africa during boreal fall to winter, but its impact on seasonal climates elsewhere is yet to be robustly understood. An IOD presence well into the winter may lead to uncertainties in S2S predictions of those modeling systems that need more skill in simulating the impacts of their (ENSO and IOD) compound forcing beyond the east coast of Africa.

The sea surface temperature anomaly on Nov 12, 2023. Strong positive phases of ENSO and IOD can be noted [].

Additionally, other less influential but capable tropical forcings, such as those related to sea surface temperature variability in the Atlantic Ocean, can co-exist with ENSO forcing and have a compounding or compensating effect on ENSO teleconnections [1]. Therefore, seasonal forecasts will continue going off the charts without accurately representing these inter-basin tropical interactions and their compound influences.

Precipitation Hotspots in Tropics

The remote influence of tropical forcing propagates through eastward-propagating Rossby waves in higher latitudes. Atmospheric diabatic heating due to SST-forced precipitation anomalies dictates the origin and strength of this wave activity. Most emphasis in S2S modeling systems is on the accurate predictability of SST variability, while accuracy in producing the correct precipitation distribution takes a secondary preference. Less skillfulness in simulating precipitation distribution in the tropics is an issue that has consequences. Even if two models accurately represent SST variability in the tropics, they may lead to substantially different Rossby wave responses and remote influences over distant regions. At the S2S scale, this uncertainty in global teleconnection of tropical forcings is less due to chaos in the Earth system and more due to their differences in precipitation distribution in the tropics (and resulting diabatic forcing) in response to SST anomalies. Therefore, accurately simulating precipitation hotspots in the tropics must take preference to reduce uncertainty in seasonal predictability.

Tropical-Extratropical Interactions

The strongest ENSO forcing occurs typically in November- December; therefore, ENSO-driven predictability is generally higher in the boreal winter season. However, this is also the high time for many modes of atmospheric variability in the Northern Hemisphere. There are almost a dozen known named modes of atmospheric variability [2]. Many of these modes may not have a substantial circumglobal impact, but they can cause a significant impact on local to regional climate in the immediate vicinity. For instance, the North Atlantic Oscillation (NAO) - the leading mode of atmospheric variability in the Northern Hemisphere - is Europe's primary cause of boreal winter precipitation and temperature variability.

Each of the atmospheric modes represents a specific wave pattern in the atmosphere. Unlike oceanic modes, these modes can have a phase shift within a season. When propagating through higher latitudes in subtropics, tropical-forcings-driven Rossby waves inevitably interact with these atmospheric modes. The nature of this tropical-extratropical interaction varies with time and depends on many factors, including the atmospheric mode that dominates at the time of these interactions and the origin of the Rossby wave forcing.

ENSO displays diversity in its characteristics, which means that the origin of ENSO-driven Rossby wave activity can differ in each ENSO occurrence. This spatial diversity implies that ENSO-forcing can be in-phase with an atmospheric mode, such as NAO, in one instance, and out-of-phase in another, even when both are in the same phase. Additionally, two ENSO forcings with the same origin will interact differently with extratropical atmospheric forcing when the latter changes phase between episodes. This is a prevalent scenario because while ENSO forcing will likely remain in the same phase through winter, many Northern Hemisphere atmospheric modes will undergo a phase shift. Therefore, the interaction between tropical and extratropical forcings changes within one season.

The impact of tropical-extratropical interactions differs depending on the region. In areas like the southeastern United States, where extratropical influence on ENSO teleconnection is minimal and ENSO-based predictability is high at several months lead, the changing nature of these interactions is not a major concern. However, in areas where extratropical atmospheric variability greatly affects seasonal climate, it is crucial to simulate tropical-extratropical interactions with precision to achieve skillful seasonal predictability. Central Southwest Asia is an example of such a region where internal atmospheric variability significantly affects precipitation variability, and ENSO's teleconnection varies sub-seasonally due to the changing nature of tropical-extratropical interactions [3, 4, 5].

Unfortunately, the current S2S models do not adequately represent the interannual variability of the Northern Hemisphere atmospheric modes and how tropical-extratropical interactions shape global teleconnections. Without improving these deficiencies, there is no clear path to achieving a breakthrough in model-based seasonal predictability.

During a strong ENSO year, S2S predictability is typically more accurate than usual, which may overshadow seasonal prediction systems' inherent limitations this winter. This is because ENSO forcing may be more dominant than other oceanic forcings, and its interaction with extratropical atmospheric variability may be less critical in shaping its teleconnections. However, as we enter the era of AI-based deep learning, the shortcomings of these systems will inevitably become more apparent soon.


bottom of page