Impact of Annual Cycle on ENSO Variability and Predictability

Abstract

Low-order Linear Inverse Models (LIMs) have been shown to be competitive with comprehensive coupled atmosphere-ocean models at reproducing many aspects of tropical oceanic variability and predictability. This paper presents an extended cyclo-stationary Linear Inverse Model (CS-LIM) that includes the annual cycles of the background state and stochastic forcing of tropical sea surface temperature (SST) and sea surface height (SSH) anomalies. Compared to a traditional stationary LIM that ignores such annual cycles, the CS-LIM is better at representing the seasonal modulation of ENSO-related SST anomalies and their phase locking to the annual cycle. Its deterministic as well as probabilistic hindcast skill is comparable to the skill of the North American Multi-Model Ensemble (NMME) of comprehensive global coupled models. The explicit inclusion of annual-cycle effects in the CS-LIM improves the forecast skill of both SST and SSH anomalies through SST-SSH coupling. The impact on the SSH skill is particularly marked at longer forecast lead times over the western Pacific and in the vicinity of the Pacific North Equatorial Countercurrent (NECC), consistent with westward propagating oceanic Rossby waves that reflect off the western boundaries as eastward propagating Kelvin waves and influence El Niño development in the region. The higher CS-LIM skill is thus associated with the improved representation of both ENSO phase-locking and Pacific NECC variations. These improvements result not only from explicitly accounting for the annual cycle of the background state, but also that of the stochastic forcing.

The Influence of the Trend, Basin Interactions, and Ocean Dynamics on Tropical Ocean Prediction

“…The full LIM has considerable skill in predicting SST and SSH anomalies at a 6‐month lead in portions of the three basins, consistent with previous LIM forecasts (e.g., Newman & Sardeshmukh, 2017; Shin et al., 2021). AC values exceed 0.6 for SST in the central tropical Pacific and Indian oceans and the western Atlantic.…”

“…Including Atlantic and Indian Ocean interactions enhance SST forecast skill in the eastern Pacific and substantially increase the SSH prediction skill west of South America around 15°S. The underlying dynamics for the latter source of skill is unclear but it may be related to the trend rather than direct interbasin connections (see Figure S8 in Supporting Information ) or result from $$\textbf{\textsf{L}}$$ not varying with the seasons (Shin et al., 2021). Removing SSH variability in the Indian and Atlantic Oceans has limited impact on the 6‐month SST forecast skill throughout the tropics and on 6–12 months ENSO forecasts, suggesting an important role for thermodynamic processes within the Indian and Atlantic oceans and in basin interactions.…”

“…The matrix $$\textbf{\textsf{L}}$$ can be estimated from the concurrent and lagged covariances, $$\textbf{\textsf{C}}(0)$$ and $$\textbf{\textsf{C}}({\tau }_{0})$$, $$\textbf{\textsf{L}}={\tau }_{0}^{-1}\hspace*{.5em}\mathrm{ln}[\textbf{\textsf{C}}({\tau }_{0})\textbf{\textsf{C}}{(0)}^{-1}]$$, where the lag $${\tau }_{0}$$ is set to 1 month in this study. Given the necessity of having long data records to accurately represent seasonally varying $$\textbf{\textsf{L}}$$ and $$\textbf{\textsf{S}}$$ values (Shin et al., 2021), we assumed that they are state‐independent; and thus, the ensemble mean forecasts at lead $$\tau $$ can be obtained from $$$\mathbf{x}(t+\tau )=\mathbf{G}(\tau )\mathbf{x}(t),$$$ where $$\textbf{\textsf{G}}(\tau )=\mathrm{exp}(\textbf{\textsf{L}}\tau )$$ is the linear system propagator.…”

“…, where the lag 𝐴𝐴 𝐴𝐴0 is set to 1 month in this study. Given the necessity of having long data records to accurately represent seasonally varying 𝐴𝐴 ▲ and 𝐴𝐴 ❙ values (Shin et al, 2021), we assumed that they are state-independent; and thus, the ensemble mean forecasts at lead 𝐴𝐴 𝐴𝐴 can be obtained from…”

An Optimal Precursor of Northeast Pacific Marine Heatwaves and Central Pacific El Niño Events

“…The onset times (September‐to‐February) of six of the seven MHWs identified in our record (Figure 1a) are consistent with the seasonal development of the Pacific Meridional Mode, and the subsequent evolution of CP El Niño events, as discussed in a recent study that used a seasonally varying LIM (Vimont et al., 2022), suggesting that some seasonal aspects of MHW and CP ENSO evolution may be implicitly captured by our LIM. A seasonally varying LIM (Shin et al., 2021) will be considered in future studies.…”