Past Seminars

Abstract

The present study comprehensively investigates the practical and intrinsic predictability of sea surface temperature (SST) in the Northern Tropical Atlantic (NTA) using 138-year coupled hindcasts from a recently developed seasonal ensemble prediction system. The system produces skillful deterministic predictions of prominent warm and cold events at least six months in advance. It is particularly effective in providing probabilistic predictions of below- and above-normal events, rather than neutral events.

NTA SST predictability shows pronounced seasonal variation, with predictability peaks for April and October targets, regardless of lead time. For April, preceding remote forcing from ENSO in the tropical Pacific, together with local signals, contributes to phase locking of SST variability and the seasonality of the signal component over the NTA. For October, predictability is associated with additional sources that have received comparatively little attention in previous studies.

Similar to the Indian Ocean Dipole, ENSO and the system’s signal-to-noise ratio contribute substantially to predictability beyond persistence at long lead times for spring NTA SST. This suggests that future improvements in ENSO prediction may create considerable room for improving current NTA SST forecasts.

Abstract

Tropical Pacific precipitation regulates the global energy balance and exerts far-reaching influences on weather and climate through atmospheric teleconnections. In the tropics, sea surface temperature and precipitation are linked by a pronounced nonlinear relationship that plays a central role in shaping ENSO and its climatic impacts. Despite its importance, the extent to which this nonlinearity governs ENSO characteristics and the underlying physical mechanisms remains poorly understood.

This study addresses the problem from three complementary perspectives. First, a parameterization explicitly representing SST–precipitation nonlinearity is incorporated into the Zebiak–Cane model. ENSO amplitude and periodicity emerge as the characteristics most sensitive to changes in this nonlinearity, while several statistical properties exhibit strongly nonlinear responses when interactions among atmospheric parameters are considered.

Second, the study identifies the sources of intermodel spread in simulations of the present-day SST–precipitation nonlinearity. Differences in negative precipitation anomalies associated with varying La Niña intensity, together with differences in mean-state precipitation generated by convective parameterization schemes, jointly modulate the simulated relationship.

Third, the present-day cold-SST bias averaged over the equatorial cold tongue is shown to be a key source of uncertainty in projected future changes. An emergent-constraint approach substantially reduces intermodel uncertainty and indicates that SST–precipitation nonlinearity will weaken under future warming.

Abstract

Understanding atmospheric responses to sea surface temperatures is essential for seasonal prediction because the atmospheric response represents the best possible prediction and determines how fully potential predictability can be realized as forecast skill. Comparisons of atmospheric responses during four strong El Niño winters, inferred from a large ensemble of atmospheric-model simulations, show that the connection between atmospheric response and observed SST forcing remains challenging to understand.

The 2023/2024 El Niño produced the weakest California precipitation response among the four recent strong El Niño events. SST anomalies differed in the amplitude of eastern Pacific warming, the broader warming of the global oceans, and the presence of exceptionally warm tropical Atlantic SSTs. Sensitivity experiments were used to determine which differences contributed to the weak California response.

The results indicate that although the global SST warming trend contributed to reduced California rainfall, the exceptional tropical Atlantic warming—larger than could be explained by the global SST trend alone—played a key role. The findings underscore the need for a systematic, community-based approach to understanding variations in atmospheric responses to changes in SST forcing.

Abstract

Tropical high-altitude mountain ice cores provide a unique perspective on climate variability in the tropical middle and upper troposphere. However, whether the ice-core oxygen isotope ratio (δ¹⁸O_ice) primarily reflects temperature fluctuations or monsoon precipitation has remained a subject of long-standing debate.

By integrating transient simulations from the isotope-enabled iTRACE model with satellite observations and radiative-convective equilibrium theory, this study shows that δ¹⁸O_ice quantitatively records tropical middle-to-upper-tropospheric temperature changes since the Last Glacial Maximum. The results identify this region as one of high climate sensitivity and indicate that tropical δ¹⁸O_ice captures a global mean surface cooling of approximately −5.85 ± 0.51°C during the Last Glacial Maximum relative to the present.

The long-term decline in tropical δ¹⁸O_ice during the Holocene cannot be explained by annual-mean temperature alone. High-resolution iCESM simulations, adjusted so that model-grid elevations match actual ice-core sites, indicate that roughly half of the Holocene decline was driven by seasonal precipitation shifts associated with orbital insolation. The remaining signal reflects a combination of local wet-season precipitation intensity and dry-season temperature variations.

The study demonstrates that the climatic meaning and sensitivity of δ¹⁸O_ice change across forcing regimes and timescales. Reliable reconstruction of past climate therefore requires careful understanding of how the isotope signal responds to different climate processes.