Atmospheric and Oceanic Sciences

OBJECTIVE: The objective of this scoping review is to systematically map the extent and nature of the literature on the use of imaging for assessing cervicothoracic junction anatomical variation in patients with neurogenic thoracic outlet syndrome. INTRODUCTION: Neurogenic thoracic outlet syndrome is a complex, disputed diagnosis that lacks a definitive gold-standard test, despite being the most common thoracic outlet syndrome subtype. Imaging is essential for identifying anatomical variations, such as cervical ribs, and excluding competing diagnoses. This review aims to clarify how imaging modalities are utilized to describe, classify, and assess these anatomical variations, and how these findings are explicitly linked to diagnosis or management. SELECTION CRITERIA: This review will consider studies that include patients with neurogenic thoracic outlet syndrome and evaluate the role of imaging in assessing anatomical variations at the cervicothoracic junction. All study designs, including reviews, quantitative, and descriptive studies, will be considered. This review will exclude studies where the primary focus is the diagnosis of arterial or venous thoracic outlet syndrome using vascular imaging modalities. METHODS: The search strategy will employ a three-step process, developed in consultation with a medical librarian, utilizing a final comprehensive search across multiple major databases (e.g., MEDLINE, CINAHL, Embase, Scopus) and gray literature sources. Studies published in the English language and from database inception will be included. Following screening by two independent reviewers, data will be extracted, synthesized descriptively, and presented in tabular and narrative format to map the identified evidence. REVIEW REGISTRATION: Open Science Framework https://osf.io/hybg2.

[This corrects the article DOI: 10.1371/journal.pone.0341172.].

Reduced audit quality behavior is prevalent in audit practice and poses a serious threat to overall audit quality. With the increasing adoption of digital auditing, auditors' digital literacy has become essential not only for adapting to the modern audit environment but also for improving audit quality and curbing inappropriate audit practices. This study examines the effect of auditors' digital literacy on reduced audit quality behavior by introducing digital self-efficacy and task performance as mediating variables and constructing a chain mediation model, while also exploring the moderating role of prosocial behavior. Empirical analysis based on 480 valid questionnaires revealed that digital literacy negatively influences reduced audit quality behavior. Digital self-efficacy and task performance partially mediate this relationship. Additionally, digital self-efficacy positively enhances task performance, resulting in a chain mediating effect. Prosocial behavior moderates these relationships, significantly affecting the mediating roles of digital self-efficacy and task performance. Accordingly, fostering a digitally enabled audit environment that enhances auditors' analytical efficiency, judgment accuracy, and procedural compliance requires auditors to effectively leverage digital technologies to mitigate audit quality-threatening behavior.

BACKGROUND: The pneumococcal conjugate vaccine (PCV) and rotavirus vaccine (RVV) have been introduced to Ethiopia's expanded childhood immunization program in 2011 and 2013, respectively. METHODS: A cross-sectional study among 2,055 children aged 12-35 months was employed. Data were extracted from the Kids Record file of the 2019 Ethiopian Mini Demographic and Health Survey. Spatial regression models were fitted and compared using corrected Akaike Information Criteria, Bayesian Information Criteria and adjusted R2. Spatial predictors were determined to be statistically significant if their p-value was < 0.05. RESULTS: Incomplete uptake of recently introduced immunizations was observed in nearly half (48.83%) of children aged 12-35 months. Its distribution throughout Ethiopia's regions shows significant spatial clustering, with the eastern part of SNNPR and Somali regions having hot spots. A total of 3 significant clusters, located in southern Oromia, east and west Hararge, the entire SNNPR, and the majority of the Somali region, with a high rate of incomplete uptake of newly introduced vaccines, were identified during SaTScan analysis. Not having vaccination cards, household size more than 5 members, home delivery, not having postnatal care, and less than 18 maternal age at first birth are positive significant spatial factors while parents not being head of the household were identified as negative significant spatial factors. CONCLUSION: There is high incomplete uptake and spatial disparities of PCV and RVV. To improve vaccination coverage among children aged 12-35 months, policymakers and health planners should prioritize targeted interventions in hotspot areas, strengthen maternal health service utilization, and enhance vaccination tracking systems. Promoting community awareness and improving access to essential health services will be critical to ensuring equitable immunization uptake across the country.

PURPOSE: Type 2 diabetes mellitus (T2DM) is a common comorbidity among patients with heart failure (HF). This study aims to describe how T2DM relates to HF hospitalization patterns across demographic groups and HF subtypes. METHODS: We conducted a population-based retrospective cohort study using 19 years of clinical data from Cerner Health Facts®, a nationwide electronic health record (EHR) database. Adult patients hospitalized with HF were identified using ICD-9/ICD-10 diagnosis codes and HF-related medications and were stratified by T2DM status. Measures included HF-related hospitalization count, length of first HF hospitalization, and age at first HF hospitalization. Comparisons were examined across HF subtypes (historically termed systolic, diastolic, other). Additionally, sensitivity analyses were conducted using alternative inclusion criteria to assess the robustness of the findings. RESULTS: We identified 137,785 HF patients from the EHR database, among whom 29.5% had T2DM. Overall HF hospitalization was more common in men than in women; however, diastolic HF was more prevalent among women and presented at older ages whereas systolic HF was more prevalent among men. Compared with patients without T2DM, patients with T2DM experienced higher HF hospitalization count (mean: 2.81 vs. 2.42, p < 0.001), and both longer stay (mean: 7.03 vs. 6.85 days, p < 0.001) and earlier age (mean: 68.9 vs. 70.4 years, p < 0.001) of initial HF hospitalization. Across HF subtypes, T2DM was more prevalent among patients with diastolic HF (31.4% diastolic vs. 29.5% systolic vs. 28.7% other), and patients with T2DM were younger at first HF hospitalization than those without T2DM, with the largest difference observed for diastolic HF (mean: 70.4 vs. 73.1 years, p < 0.001). CONCLUSIONS: In this large nationwide EHR cohort, T2DM was associated with more intensive HF hospitalization patterns and was more prevalent in the diastolic HF subtype. These findings highlight the relevance of diabetes status and HF subtype, in addition to demographic factors, in shaping HF-related healthcare utilization.

Prescription of prosthetic ankle-foot devices is constrained by imprecise clinical guidelines and inconsistent scientific evidence, hindering optimal device selection for individuals with lower limb loss. This multisite, prospective, randomized crossover study aimed to identify patient-reported, performance-based, and biomechanical parameters sensitive to ankle-foot device type, providing a foundation for more objective and individualized prescription practices. Ninety-one individuals with unilateral transtibial limb loss completed the crossover trial, and 13 control participants without musculoskeletal impairment were enrolled to provide normative reference data. Participants were fitted with duplicate sockets and randomized to trial three ankle-foot device types: energy storing and returning, articulating, and powered. Participants were heterogeneous in demographic characteristics, including veterans, service members, and civilians. After one week of acclimation per device, participants completed performance-based (6-minute walk, Timed Up and Go, Four Square Step Test, Stair and Hill Assessment Indices, Amputee Mobility Predictor) and patient-reported (Prosthesis Evaluation Questionnaire, 12-Item Short Form Health Survey, Orthotics and Prosthetics Users' Survey) assessments; a subset (n = 29 completed) underwent full-body gait analysis to capture detailed biomechanical parameters. Biomechanical outcomes demonstrated the greatest sensitivity to device type, with 19 distinct parameters, primarily at the ankle, highlighting ankle mechanics as a key determinant of differences among prosthetic devices. Five Prosthesis Evaluation Questionnaire subscales were sensitive to device type, while performance-based measures showed no significant effects. Results revealed a dichotomy between biomechanical and patient-reported outcomes: Biomechanical parameters were more similar to control values for powered devices, whereas patient-reported outcomes favored non-powered devices. Linear discriminant analysis identified key gait features, including peak plantarflexion during preswing and peak ankle moment, which most strongly contributed to group separation and clinical discrimination. These findings identified distinct biomechanical and patient-reported parameters sensitive to ankle-foot device type and highlight the need for evidence-based, individualized prosthetic prescription to optimize device selection and improve patient outcomes.

Abstract Climate change is intensifying compound flooding in coastal regions, caused by the co‐occurrence of coastal and fluvial floods. This trend increases the probability of disasters beyond historical precedent, posing growing threats to densely populated deltas. Existing studies, however, cannot fully characterize how a warming climate affects multiple flood drivers and their interactions, which may lead to biased projections of future compound extreme events. In this study, we develop a physics‐based method that dynamically integrates the impact of climate change on both tropical cyclone (TC) activities and rainfall‐runoff processes. We apply this method to project coastal‐fluvial compound flood risk in the Pearl River Delta (PRD). An ensemble of 100,000 storm events is generated to characterize possible compound extreme events under the historical (1974–2014) and future (2060–2100; SSP5‐8.5) climate scenarios. At a representative estuarine location, the projected results reveal a significant increase in the probability of coastal and fluvial extremes, with the return period decreasing from 55.5 to 15.5 years. These projections indicate that compound flood events unprecedented in the historical record are more likely to occur in the future. Our findings highlight the need for adaptive coastal hazard management strategies to address the escalating compound flood risk under climate change.

Abstract The global terrestrial water and carbon cycles have evolved rapidly during recent decades, particularly in regions with intensive anthropogenic interventions and vulnerable ecohydrological systems. However, current land surface models (LSMs) and global hydrological models (GHMs) often treat natural ecohydrological processes and human water use in isolation, leaving a critical gap in the holistic understanding of coupled human‐natural systems at global scale. To bridge this gap, we developed the Conjunctive Surface–Subsurface Process Model version 3 (CSSPv3), a global land surface‐ecology‐hydrology coupled model that innovatively integrates human activities including irrigation and reservoir operation, dynamic water‐carbon couplings, and glacier freeze–thaw cycles within a water–energy balanced framework. A global offline simulation for 1991–2020 using CSSPv3 with a spatial resolution of 0.25° is validated against extensive observations, demonstrating that CSSPv3 outperforms its predecessor, CSSPv2, particularly for simulations of streamflow over human and glacier influenced basins, and evapotranspiration over irrigation fields. Moreover, the CSSPv3‐generated global water product demonstrates advantages for representing irrigation amount, streamflow, surface soil moisture, evapotranspiration, and glacier mass change as compared with 26 global products. The CSSPv3‐generated global carbon product is consistent with global ensemble medians from other products. These results establish CSSPv3 as a robust and comprehensive platform for simulating coupled ecohydrological processes under both natural climate variability and human influence, providing a powerful tool for advancing land surface modeling in the Anthropocene.

Reliable drought prediction is crucial for early warning and decision-making regarding drought-related disasters. Flash droughts, characterized by their rapid onset and severe impacts, are challenging to predict and present critical challenges to water resource management. Specifically, streamflow flash droughts (SFDs) are significant as they directly influence water supply for irrigation, industry, and domestic use. Traditional hydrologic models and drought early warning systems are primarily designed to capture slowly evolving drought conditions and often fail to represent rapid transitions in streamflow that characterize the onset of SFDs. Despite increasing attention to the identification and characterization of SFDs, their prediction remains unexplored. This study evaluates the ability of different deep learning architectures, especially the Temporal Fusion Transformer (TFT), in comparison to the widespread baseline Long Short-Term Memory (LSTM) model and a two-source LSTM incorporating static catchment attributes, to predict SFDs based on streamflow percentiles. The model training and performance evaluation were conducted using hydroclimatic datasets from 671 Catchment Attributes and Meteorology for Large-sample Studies (CAMELS) catchments in the contiguous United States. Performance was evaluated using standard accuracy and event detection metrics. Results show that the TFT achieved the highest overall performance, with a median KGE of 0.87, RMSE of 0.13, MAE of 0.10, and a correlation coefficient of 0.91, outperforming the LSTM with static features (median KGE = 0.81) and the baseline LSTM (KGE = 0.78). In terms of detecting SFDs, the TFT showed the highest median detection rate and lowest miss rate across all models. These improvements were consistent across most catchments and hydroclimatic regions, indicating robust performance gains relative to LSTM-based approaches, although some regions with complex hydrologic behavior remained challenging for all models. While model performance varied across hydroclimatic regions, particularly in data-sparse or hydrologically complex areas, the integration of static attributes and attention-based mechanisms consistently improved predictive skill. These findings demonstrate that Transformer-based models can improve prediction of rapid streamflow changes associated with the onset of SFDs; however, the results are based on deterministic predictions and do not explicitly quantify predictive uncertainty. These findings contribute new insights into the spatial patterns and physical drivers of model performance under non-stationary climate conditions.

Abstract Predictability analysis, which focuses on perturbation growth dynamics, is a key problem in both weather and climate prediction. Among all perturbations, the conditional nonlinear optimal perturbation (CNOP) explores maximum uncertainties in forecasts, which is fundamentally important for theoretical studies and applications. Traditionally, CNOPs are solved through iterative optimization of numerical weather prediction (NWP) systems. Their large computational demands pose significant challenges to long‐term predictability analysis. In our study, using a fast and accurate Artificial Intelligence (AI) model, i.e. FuXi, a low‐cost optimization framework for solving five‐day tropical cyclone (TC) CNOP is developed. For the first time, CNOPs that achieve the optimal (i.e., fastest) nonlinear development of long‐term TC forecast errors are solved, with their optimality and physical explainability verified. Results demonstrate that perturbations with specific spatial structures undergo significant development. In both AI and NWP models, AI‐based CNOPs exhibit rapid and physically consistent error growth across diverse TC cases, faster compared to random and lagged forecast perturbations. Furthermore, sensitivity analysis reveals that far‐environment systems and processes are more crucial for long‐term TC forecasts. Structural analyses of the CNOP emphasize the interactions between TC internal and external processes for rapid perturbation growth. The successful derivation of AI‐based CNOPs, with their rapid growth and physical explainability verified in both AI and NWP models, suggests that AI models can capture the most rapidly growing perturbation patterns and their subsequent nonlinear evolution. Thus, the potential of AI models is highlighted for advancing atmospheric predictability studies, including theoretical analysis, targeting observations and ensemble forecasts.

Abstract. The climate in Europe is warming faster than the global average, raising concerns about how climate change will affect extreme fire events. In this study, we use ERA5-Land reanalysis data and an ensemble of 33 high-resolution regional climate models (RCMs) from the EURO-CORDEX framework to compute the Canadian Forest Fire Weather Index (FWI) and investigate both recent and projected changes in atmospheric conditions favorable for wildfires across Europe. Historical trends (1950–2023) based on ERA5-Land data reveal statistically significant increases in the frequency and intensity of extreme fire weather in regions such as the Iberian Peninsula, Central Europe, and parts of Eastern Europe. All RCM input fields were bias-adjusted prior to FWI calculation using Quantile Delta Mapping, resulting in improved FWI representation relative to unadjusted simulations. Projections based on the bias-adjusted EURO-CORDEX ensemble indicate that future extreme fire weather will become more frequent, more intense, and more widespread across Europe as global warming progresses. The strongest signals are projected for southern Europe, with a northward expansion of fire-prone conditions under higher global warming levels (GWLs). At 3 °C GWL, the spatial extent of robust changes in extreme fire weather metrics nearly doubles compared to 2 °C, with one metric increasing fivefold. Relative increases in frequency-based metrics generally exceed those in magnitude-based metrics. These changes coincide with rising vapor pressure deficit, suggesting that thermodynamic processes play a key role through atmospheric drying. The projected intensification of extreme fire weather in Europe highlights the growing need for coordinated climate action along with proactive mitigation strategies.

Abstract Permafrost regions store vast amounts of soil carbon, and thaw under global warming enhances microbial decomposition and CO 2 release, strengthening the permafrost carbon feedback. While this feedback has been extensively studied for continued warming, the processes controlling its magnitude and persistence remain uncertain. Using the Community Earth System Model version 2 (CESM2), we analyze permafrost carbon cycle dynamics under both warming and subsequent mitigation and quantify the drivers of net biome production (NBP). We show that soil moisture is a primary regulator of these dynamics. Thaw‐induced increases in liquid soil water stimulate plant photosynthesis but more strongly enhance microbial decomposition, shifting ecosystems toward net carbon loss. The sensitivity to soil moisture is highly heterogeneous, with the strongest response in central Siberia, where high litter carbon coincides with relatively dry background conditions, and weaker sensitivity in wetter North America. Elevated soil moisture also delays permafrost refreezing even after zero emission are reached, sustaining high heterotrophic respiration and prolonging net carbon loss. These results demonstrate that soil moisture amplifies and prolongs permafrost carbon losses from warming through mitigation, underscoring the need for improved representation of coupled hydrological and permafrost processes in carbon‐budget assessments.

Abstract Antarctic sea ice projected to have large reduction associated with increasing greenhouse gases, with potential impacts extending beyond the Southern Hemisphere. However, its influence on East Asian climate is not fully understood. Using Earth system model experiments and multi‐model simulations, we show that Antarctic sea‐ice loss by the end of the 21st century leads to a significant increase in summer precipitation over central and southern China. This precipitation response is further amplified with enhanced sea‐ice loss under quadrupled CO 2 forcing. Additional atmospheric model experiments indicate that atmospheric pathways alone cannot transmit the Antarctic signal to East Asia. Instead, Antarctic sea‐ice loss shapes a hemispherically asymmetric pattern of tropical sea surface warming through anomalous northward cross‐equatorial heat transport and southward of the intertropical convergence zone driven by interhemispheric energy imbalance, with the warming further amplified by positive low‐cloud shortwave feedbacks. The resulting warming in the tropical Atlantic and Indian Oceans induces anomalous northeasterly winds over eastern China and southwesterly flow over southern China through Rossby wave trains and modulation of the Hadley circulation. These circulation changes enhance upward motion and precipitation over central and southern China. Antarctic sea‐ice loss accounts for roughly 40% of the projected summer precipitation increase in this region under the high‐emission scenario, highlighting its potential role in shaping future East Asian climate.

This study aimed at (i) developing a multi-station deep learning rainfall–runoff–operation (DL-RRO) model for cascading dam systems and (ii) applying the model through the large ensemble dataset for Policy Decision-Making for Future Climate Change (d4PDF) to assess flood characteristic changes. The DL-RRO model was developed using a Long Short-Term Memory network that coupled rainfall–runoff processes and mimicked existing dam operation rules by learning operational logic from observed data. Dynamic water level–storage curves were generated to account for time-varying increases in sediment storage. The proposed method was applied to the upper Chikugo River Basin (CRB), where the Shimouke and Matsubara Dams play a critical role in flood control in Kyushu, Japan. The model showed high performance when trained, validated, and tested with 40 years (1986–2025) of hourly hydrological observations (NSE = 0.89, RMSE = 18.90 m 3 /s). Subsequently, d4PDF rainfall data for the future scenarios (2031–2090 and 2051–2110; 1,440 ensemble-year) were imposed to the DL-RRO model. Analysis of 8,725 simulated flood events revealed increased flood frequency and notable changes in hydrograph characteristics. Projected 100-year return flood may increase by +53 % at Shimouke and +165 % at Matsubara under future scenarios, exceeding existing spillway design capacities. Under the synergistic impacts of climate change and sediment dynamics, amplified flood characteristics may intensify flood-driven sediment deposition, leading to further reductions in flood control capacity of up to +2.9 % at the Shimouke and +4.1 % at the Matsubara. These findings highlighted the urgent need to upgrade dams, revise operations and sedimentation management strategies in the upper CRB.

Konservat-Lagerstätten provide exceptional records of past ecosystems, yet the environmental and climatic processes governing their formation in extreme climatic regimes remain a significant, unresolved conundrum. Here we present a novel, process-based mechanism for the Villaggio del Pescatore Konservat-Lagerstätte (northeastern Italy; Late Cretaceous, 80 Ma), which preserves a diverse terrestrial and aquatic biota. Integrating high resolution synchrotron micro-analyses, bulk geochemistry, detailed taphonomy and palaeoclimate simulations we identify a dynamic, monsoon-paced tidal flat ecosystem within a carbonate coastal margin. Field and core records comprise metrescale packages of laterally persistent light–dark laminae; microstratigraphic, spectral and statistical analyses indicate a semidiurnal tidal regime with neap–spring modulation. MicroXRF/XRD mapping reveals that individual microlaminae within couplets share the same elemental inventory refuting a seasonal varve origin. Laminae chronometry, tied to the tidal interpretation, constrains net accumulation of the fossil bearing rhythmites to decadal (~40 year) timescales. Geochemical proxies record alternating terrigenous rich, high TOC intervals and drier, better oxygenated intervals. Palaeoclimate model ensembles diagnose a vigorous, summer dominated monsoon domain at ~29° N under greenhouse boundary conditions. We document that episodic monsoon enhanced flood and tropical cyclone pulses delivered sediment, nutrients, carcasses and plant debris to a microbially active, carbonate tidal flat; microbial mats rapidly stabilised and promoted early cementation, enabling exceptional soft tissue preservation. We propose a genetic model in which monsoon–cyclone forcing, tidal trapping and microbial sealing combine to produce decadal, daily resolved vertebrate Lagerstätten in greenhouse settings, offering predictive targets for similar deposits and deep time analogues for modern coastal change.

Abstract Europe has recently experienced a series of hot and dry summers. Previous case studies indicate that, for a given atmospheric circulation, heatwave temperatures may increase by 1.5–3 K per degree of global warming, but the magnitude of this amplification varies substantially across regions and events, and its underlying mechanisms remain insufficiently constrained. Here, we investigate this variability using circulation-nudged regional climate storylines for the summers 2018–2022, dynamically downscaled over Europe and spanning climates from pre-industrial conditions to +4 K global warming. The present-day simulations realistically reproduce observed temperature, soil moisture, and surface fluxes. We find that warming amplification is spatially heterogeneous and seasonally dependent, with the strongest amplification over Central Europe occurring in August, indicating a disproportionate intensification of late-summer heatwaves. Using the evaporative fraction-soil moisture framework, we show that warming amplification is closely linked to changes in the evapotranspiration regime. The strongest amplification occurs when warming induces a transition from energy-limited to moisture-limited conditions, triggering rapid soil moisture depletion, a pronounced reduction in evaporative fraction, and enhanced sensible heating. In contrast, amplification is weaker when conditions are already moisture-limited in the pre-industrial climate, due to the reduced potential for further soil drying, and weakest when the regime remains energy-limited. These findings demonstrate that spatial and event-to-event differences in warming amplification can partially be attributed to the sensitivity of land-atmosphere coupling to soil moisture changes. Accounting for these feedbacks is essential for robust projections of future heat extremes and for identifying regions most vulnerable to disproportionate increases in heatwave intensity.

ABSTRACT Output from dynamically downscaled, convection‐permitting regional climate simulations is used to examine projected changes in the frequency, timing and spatial distribution of synthetic severe convective storm (SCS) detections based on peril proxies across three 15‐year epochs—1990–2005, 2040–2055 and 2085–2100—under intermediate and pessimistic emissions scenarios, including the first mid‐century projections of SCS activity, providing a rare assessment of how SCS climatology may evolve throughout the 21st century. An explicit multi‐hazard approach is employed to define severe convective perils using permutations of simulated upward vertical velocity and lowest model level reflectivity thresholds, allowing SCS activity to be assessed collectively for tornado, wind and hail perils. In response to enhanced greenhouse forcing and related changes in fundamental severe storm ingredients, projected synthetic SCS activity exhibits spatiotemporal changes relative to the historical baseline, including lengthening of the severe season, increases in the overall frequency of SCS days and an eastward shift of event frequency maxima. In addition to changes in mean frequency, results illustrate enhanced variability and increases in relatively high‐activity days across all future epochs, driven by more frequent spring and early summer occurrences. Regionally, increases in SCS days are projected for south‐central and Gulf Coast states, the southern Great Plains and the Midwest in the spring and the Midwest, Northeast and Southeast during the summer, while decreases are projected for the Great Plains during the summer. These changes are accompanied by a broadening geographic footprint of SCS activity, indicating a redistribution of severe storm risk under warming. Results provide physically grounded insight into the potential evolving severe storm landscape, which stakeholders, policymakers and the public may use to mitigate and build resilience against future events.

El Niño-Southern Oscillation (ENSO) is one of the key drivers of global climate anomaly, and its evolution during geological history is crucial for understanding the natural variability of the climate system. The Middle Miocene Climatic Optimum (MMCO, 16.9–14.7 Ma) was a period of relatively warm global climate. During this period, the contraction and closure of tropical seaways exerted profound impacts on global ocean circulation and climate. Using idealized sensitivity experiments with the FGOALS-g3 model, we artificially close the Indonesian Seaway to isolate its mechanistic influence on ENSO dynamics. Although this extreme scenario does not reflect the actual open-gateway conditions of the Miocene, it allows us to clearly identify the seaway’s role in modulating tropical climate variability. The results demonstrate that the closure of the Indonesian Seaway exerts a significant impact on ENSO, increasing ENSO amplitude by 140%, an impact far exceeding that of the Panama and Tethys Seaways. Diagnostics based on the mixed-layer heat budget equation reveal that the enhancement of ENSO amplitude is primarily driven by the strengthened zonal advection feedback. The closure of the Indonesian Seaway drives the formation of an “El Niño-like” mean state in the tropical Pacific, which facilitates the eastward propagation and amplification of sea surface temperature anomalies and wind stress anomalies. These findings underscore the critical role of the Indonesian Seaway in modulating ENSO variability during the MMCO and provides mechanistic insights into the potential amplifying effect of the weakening Indonesian Throughflow on ENSO under future global warming scenarios.

Abstract Tropical instability waves (TIWs) generate mesoscale sea surface temperature (SST) fluctuations in the eastern equatorial Pacific and influence the evolution of the El Niño‐Southern Oscillation (ENSO). Yet satellite‐based prediction of TIW‐related SST anomalies remains largely SST‐centric and makes limited use of sea surface salinity (SSS) observations. Here we show that satellite SSS provides a practically useful constraint on TIW‐related SST forecasts. We develop a dual‐branch deep learning framework trained on Copernicus Marine Service SST and SSS products, in which a ConvLSTM‐UNet3D trend branch represents slowly varying evolution and a frame‐difference branch highlights transient TIW disturbances. A Mean Deviation Loss strengthens learning in regions of strong anomalies, including the TIW triangular zone and the cold tongue. SST is the sole prognostic output, while SSS is prescribed at each rollout step under three protocols with distinct operational meanings: a deployable, leakage‐free SST‐recursive forecast using training‐derived monthly SSS climatology; a diagnostic upper bound using observed SSS at the corresponding forecast dates; and a constant‐field Zero‐map stress test. Compared with the SST‐only configuration, incorporating time‐varying SSS reduces the one‐step 5‐day SST RMSE from 0.30 to 0.27°C, while the deployable climatological‐SSS setting attains 0.28°C, within 0.01°C of the observed‐SSS upper bound at the single‐step horizon and closely tracking it under multi‐step rollouts. Cross‐spectral and time‐lagged regression analyses further reveal that coherent leading SSSA modes can lead SSTA by 5–12 days on TIW timescales, indicating that seasonal‐background and event‐scale SSS carry physically meaningful, complementary predictive information for TIW‐related SST evolution.