Marine heat waves (MHWs) are prolonged periods of elevated ocean temperatures that can devastate marine ecosystems, fisheries, and coastal communities. Skillfully predicting these events with sufficient lead time is crucial for mitigating their adverse effects. This study presents a probabilistic subseasonal MHW forecast tool using a U-Net-based neural network architecture, with a focus on the Northern Indian Ocean and the Arabian Sea. The model was trained using sea surface temperature and sea surface height reanalysis data. The U-Net-based forecast tool demonstrated significant predictive skill up to 10 weeks in advance across various deterministic and probabilistic skill metrics. The model outperformed persistence and climatology-based benchmarks, especially in the tropical warm pool. Future applications of explainable artificial intelligence (XAI) methods have the potential to identify the sources of predictive skill, inform understanding of underlying dynamics, and improve dynamic subseasonal to seasonal forecast models.

Estimating ocean heat content (OHC) with reliable uncertainties is critical for understanding and monitoring the evolution of Earth's climate, as the ocean has stored most of the energy accumulated in the climate system due to Earth Energy Imbalance. Here, we use Argo profiling float data from 2004-2022 to map OHC. As fewer Argo observations are available deeper in the water column, previous studies have partitioned the ocean into at least two pressure layers and mapped each separately, which complicates the estimation of uncertainties when the maps are summed to get the total OHC. In this work, we consider the case of two pressure layers and propose an improved mapping and uncertainty quantification method using bivariate locally stationary Gaussian processes and conditional simulations to map the two sections jointly while accounting for the correlation between them. We find that modeling this correlation results in improved OHC anomaly mapping and up to a 15 percent reduction of global OHC anomaly uncertainties in comparison to mapping the two layers separately without accounting for their dependence. These estimated uncertainties are essential to analyze the statistical significance of OHC anomalies on both regional and global scales, which we demonstrate using several climatological case studies.

Argo profiling floats measure seawater temperature and salinity in the upper 2000 meters of the ocean. These floats are uniquely capable of measuring the global Ocean Heat Content (OHC), a quantity that is of central importance for understanding Earth Energy Imbalance. Yet, producing Argo-based OHC estimates with reliable uncertainties is statistically challenging due to the complex structure and large size of the Argo dataset. Here we present an end-to-end mapping and uncertainty quantification framework for Argo-based OHC estimation using state-of-the-art methods from spatio-temporal statistics. The framework is based on modeling vertically integrated Argo temperature profiles as a locally stationary Gaussian process defined over space and time. This enables us to produce computationally tractable OHC anomaly maps based on data-driven decorrelation scales estimated from the Argo observations. Our modeling choices are validated using statistical cross-validation, which demonstrates the importance of including a climatological time trend in the mean field and accounting for time in the covariance function. We quantify the uncertainty of these maps using local conditional simulation ensembles, a novel approach that leads to principled spatially and temporally correlated uncertainty quantification. A new paired cross-validation technique is presented to validate these uncertainties. The mapping framework is implemented in an open-source codebase that is designed to be modular, reproducible and extensible. To demonstrate the mapping and uncertainty quantification capabilities of this approach, we present new Argo OHC maps with uncertainties for 2004-2022 and report on various downstream climatological estimates and their uncertainties.

Abstract. Marine heatwaves (MHWs) have emerged as a very active area of research due to the devastating impacts of these events on marine ecosystems across different trophic levels. Yet, a clear understanding of the local drivers of these extreme ocean conditions is still limited at a global scale. Observations of the terms needed to constrain ocean heat budgets are very sparse, ocean reanalysis products are generally non-conservative and inadequate to conduct accurate heat budget analyses, and the fidelity of climate models in simulating MHWs is still unclear. In this study, we make use of Argo floats observations, a satellite-based sea surface temperature product, and the Estimating the Circulation and Climate of the Ocean (ECCO) state estimate to assess MHW characteristics over the global ocean. ECCO is then used to evaluate local MHW drivers. ECCO assimilates observations using the adjoint methodology, which optimizes the system trajectory given the observational constraints in a conservative fashion, making it an ideal product for the estimation of heat budgets. The representation of MHWs in ECCO is overall consistent with observations, although ECCO tends to underestimate MHW frequency and intensity and overestimate duration, relative to the observational products. Atmospheric forcing emerges as the dominant contributor to MHW onset and decline across most regions, while ocean dynamics, including advective and diffusive convergence of heat, play crucial roles in the equatorial regions, specific extra-tropical zones (e.g., western boundary currents such as the Gulf Stream and Kuroshio), and the Southern Ocean. Regional analyses in the Northeast Pacific, Southwest Pacific, and Tasman Sea, show diversity in leading dynamical mechanisms for MHW onset and decline both across regions and across events in the same regions: while air-sea exchanges of heat may contribute most frequently to MHW onset and decline, other mechanisms can also often provide dominant contributions and at times be the main driver. A more complete understanding of MHWs and their drivers is crucial for predicting their initiation, duration, intensity and decline, to ultimately inform the development of mitigation and adaptation strategies for affected communities.

Recent advances in explainable artificial intelligence (XAI) methods show promise for understanding predictions made by machine learning (ML) models. XAI explains how the input features are relevant or important for the model predictions. We train linear regression (LR) and convolutional neural network (CNN) models to make 1-day predictions of sea ice velocity in the Arctic from inputs of present-day wind velocity and previous-day ice velocity and concentration. We apply XAI methods to the CNN and compare explanations to variance explained by LR. We confirm the feasibility of using a novel XAI method [i.e., global layerwise relevance propagation (LRP)] to understand ML model predictions of sea ice motion by comparing it to established techniques. We investigate a suite of linear, perturbation-based, and propagation-based XAI methods in both local and global forms. Outputs from different explainability methods are generally consistent in showing that wind speed is the input feature with the highest contribution to ML predictions of ice motion, and we discuss inconsistencies in the spatial variability of the explanations. Additionally, we show that the CNN relies on both linear and nonlinear relationships between the inputs and uses nonlocal information to make predictions. LRP shows that wind speed over land is highly relevant for predicting ice motion offshore. This provides a framework to show how knowledge of environmental variables (i.e., wind) on land could be useful for predicting other properties (i.e., sea ice velocity) elsewhere.

Marine heatwaves (MHWs)-extreme ocean temperature anomalies-are increasing in frequency and intensity globally, yet their vertical structure and drivers remain underexplored in the Caribbean Sea (CS), a region of critical ecological and socioeconomic importance. Using high resolution climate model simulations and reanalysis fields, we show that MHW intensity in the CS peaks in the subsurface, where vertical gradients of temperature are stronger, and that MHWs are associated with the passage of Rossby waves. Projections under the RCP8.5 scenario reveal that while the ocean continues to warm to the end of the century, subsurface MHW intensity weakens. This decline is associated with a weaker thermocline, reduced ocean currents and eddy kinetic energy, and less energetic Rossby waves. Reducing uncertainties in climate projections will be essential to improve our understanding of how marine heat extremes may evolve and affect Caribbean ecosystems.

Atmospheric rivers (ARs) are elongated structures in the atmosphere that transport water vapor from the tropics to mid‐ and high‐latitudes in highly episodic events. ARs are a major source of precipitation in the western United States and many other regions of the world and play a role in air‐sea exchanges of heat and freshwater in the subtropics. For the first time, an experiment was designed to observe (open) ocean changes during AR precipitation, using profiling floats in the northeast subtropical Pacific. Results show upper ocean freshening and cooling with AR precipitation (lasting several hours), especially when surface winds are low ( m/s). ARs contribute 70%–90 of the total winter precipitation in the subtropical northeast Pacific: a salt budget analysis (using ECCO) shows that precipitation over winter is the main driver of ocean freshening in the top 25–30 m and contributes by to the integrated changes in the top m.

The current ecosystem of Internet-facing search and distribution tools for ocean data has limited options suitable for supporting the fast queries needed to underwrite applications that update their data frequently and on demand, such as visualization websites, interactive educational activities, and analyses constructed as living documents. These applications are best served by a Representational State Transfer (RESTful) application programming interface (API) with incisive search capabilities over an appropriately indexed database of ocean datasets represented with consistent encoding. For this purpose, a new Argovis API was developed and released, along with a web-facing frontend and a collection of Jupyter notebooks that leverage it and demonstrate its capabilities. This paper reviews the key engineering decisions and reference architecture used by Argovis to create a responsive, FAIR ocean data service for Argo and ship-based profiles, derived gridded fields and other products, observations from the Global drifter program, tropical cyclone track data, an atmospheric river climatology, and weekly gridded fields of sea surface temperature, sea level anomaly, and surface winds based on satellite data. We also tour some of the use cases and applications of this data service, both by the Argovis team and third-party consumers.

Research on marine heat waves (MHWs) in the Red Sea has focused on the surface signatures of these extreme warm events, such as the sea surface temperature (SST). This focus may potentially neglect the detrimental effects of subsurface MHWs. The unprecedented coral bleaching event observed in the southern Red Sea in 2015, despite less intense SSTs than in the MHW in 2002, highlights this oversight. A high-resolution regional reanalysis of the Red Sea reveals that 2015 and 2002 were characterized by subsurface heat content anomalies of opposite signs at depths up to 100 m, with positive anomalies in 2015 and negative anomalies in 2002. A heat budget analysis suggests that the primary heat source is advection from the southern boundary connecting with the Gulf of Aden (GoA). The advection of negative temperature anomalies from the GoA contributed to decreased subsurface heat in 2002, and the advection of positive temperature anomalies from the GoA contributed to increased subsurface heat in 2015. The total increase in the subsurface heat observed in 2015 is linked to the reduction in Red Sea surface water (RSSW) and GoA intermediate water (GAIW). The higher sea surface height (SSH) and deeper 25σ isopycnal in GoA during 2015 resulted in horizontal pressure differences between the southern Red Sea and GoA, corresponding to the reduced flow of RSSW and GAIW that year. The primary factor contributing to the elevated SSH and deeper 25σ isopycnal is the presence of an anticyclonic eddy (Somali current ring) along the western shores of the GoA. The probable cause for the stronger anticyclonic eddy in 2015 compared to 2002 is the decreased intensity of the westward-propagating upwelling Rossby waves that originated from as far away as the Arabian Sea and the western coasts of India.

Marine heatwaves (MHWs) have emerged as a very active area of research due to the devastating impacts of these events on marine ecosystems across different trophic levels. However, a clear understanding of the local drivers of these extreme ocean conditions is still limited at a global scale. Observations of the terms needed to constrain ocean heat budgets are very sparse, ocean reanalysis products are generally non-conservative and inadequate to conduct accurate heat budget analyses, and the fidelity of climate models with respect to simulating MHWs is still unclear. In this study, we make use of Argo float observations, a satellite-based sea surface temperature product, and the Estimating the Circulation and Climate of the Ocean (ECCO) state estimate to assess MHW characteristics over the global ocean. ECCO is then used to evaluate local MHW drivers. ECCO assimilates observations using the adjoint methodology, which optimizes the system trajectory given the observational constraints in a conservative fashion, making it an ideal product for the estimation of heat budgets. The representation of MHWs in ECCO is generally consistent with observations, although ECCO tends to underestimate MHW frequency and intensity and overestimate duration, relative to the observational products. Atmospheric forcing emerges as the dominant contributor to MHW onset and decline across most regions, while ocean dynamics, including the advective and diffusive convergence of heat, play crucial roles in the equatorial regions, specifically in the extratropical zones (e.g., western boundary currents, such as the Gulf Stream and Kuroshio) and the Southern Ocean. Regional analyses in the northeastern Pacific, southwestern Pacific, and Tasman Sea show diversity in leading dynamical mechanisms for MHW onset and decline both across regions and across events in the same regions: while air-sea exchanges of heat may contribute most frequently to MHW onset and decline, other mechanisms can also often provide dominant contributions and, at times, be the main driver. A more complete understanding of MHWs and their drivers is crucial for predicting their initiation, duration, intensity, and decline, to ultimately inform the development of mitigation and adaptation strategies for affected communities.

As the Earth's climate warms, the intensity and rain rate of tropical cyclones (TCs) is projected to increase. TCs intensify by extracting heat energy from the ocean; hence, a better understanding of upper-ocean changes with the TC passage is helpful to improve our understanding of air-sea interactions during and after the event. This work uses Argo float observations and the HYbrid Coordinate Ocean Model (HYCOM) ocean reanalysis to describe characteristics of upper-ocean changes with hurricane-strength wind events. We study the association of these changes with the vertical structure of the salinity profile before the event, i.e., increasing versus decreasing with depth. We also study the contribution of changes in salinity to upper-ocean density changes in each case. Results show that in regions where pre-event salinity increases with depth there is a corresponding statistically significant increase in upper-ocean salinity; vice versa, we observe a significant decrease in upper-ocean salinity in regions where pre-event salinity decreases with depth. Consistent with previous studies, temperature decreases in both regions. As near-surface temperature decreases, upper-ocean density increases, and the increase is larger where pre-event salinity increases with depth. Changes in upper-ocean properties from Argo and HYCOM are overall consistent with wind-driven vertical mixing of near-surface waters with colder and higher-salinity (or lower-salinity) waters below. Resulting changes in ocean stratification have implications for air-sea interactions during and after the event, with potential impacts on weather events that follow.

This study uses an oceanic energy budget to estimate the ocean heat transport convergence in the North Atlantic during 2005–2018. The horizontal convergence of the ocean heat transport is estimated using ocean heat content tendency primarily derived from satellite altimetry combined with space gravimetry. The net surface energy fluxes are inferred from mass-corrected divergence of atmospheric energy transport and tendency of the ECMWF ERA5 reanalysis combined with top-of-the-atmosphere radiative fluxes from the clouds and the Earth’s radiant energy system project. The indirectly estimated horizontal convergence of the ocean heat transport is integrated between the rapid climate change-meridional overturning circulation and heatflux array (RAPID) section at 26.5°N (operating since 2004) and the overturning in the subpolar north atlantic program (OSNAP) section, situated at 53°–60°N (operating since 2014). This is to validate the ocean heat transport convergence estimate against an independent estimate derived from RAPID and OSNAP in-situ measurements. The mean ocean energy budget of the North Atlantic is closed to within ± 0.25 PW between RAPID and OSNAP sections. The mean oceanic heat transport convergence between these sections is 0.58 ± 0.25 PW, which agrees well with observed section transports. Interannual variability of the inferred oceanic heat transport convergence is also in reasonable agreement with the interannual variability observed at RAPID and OSNAP, with a correlation of 0.54 between annual time series. The correlation increases to 0.67 for biannual time series. Other estimates of the ocean energy budget based on ocean heat content tendency derived from various methods give similar results. Despite a large spread, the correlation is always significant meaning the results are robust against the method to estimate the ocean heat content tendency.

Earth’s energy imbalance (EEI) is a fundamental metric of global Earth system change, quantifying the cumulative impact of natural and anthropogenic radiative forcings and feedback. To date, the most precise measurements of EEI change are obtained through radiometric observations at the top of the atmosphere (TOA), while the quantification of EEI absolute magnitude is facilitated through heat inventory analysis, where 90

A new Community Earth System Model/Modular Ocean Model 6 (CESM/MOM6) ocean-only regional 1/12 degrees configuration of the Caribbean Sea is presented and validated. The model configuration was developed as a response to the rising need for high-resolution models for climate impact applications. The configuration is validated for the period covering 2000-2020 against ocean reanalysis and a suite of observation-based datasets. Particular emphasis is paid to the configuration's ability to represent the dynamical regime and properties of the region across sub-seasonal, seasonal, and inter-annual timescales. Near-surface fields of temperature, salinity, and sea surface height are well represented. In particular, the seasonal cycle of sea surface salinity and the spatial pattern of the low salinity associated with the Amazon and Orinoco river plumes is well captured. Surface speeds compare favorably against reanalysis and show that the mean flows within the Caribbean Sea (CS) are well represented. The model reproduces known pathways and timing for river plume waters intruding the region. The seasonal cycle of the mixed-layer depth is also well represented, with biases of <3 m when comparing to ocean reanalysis. The vertical structure and stratification across the water column is represented favorably against ship-based observations, with the largest simulated biases in the near-surface water mass and the sub-surface salinity maximum associated with the sub-tropical underwater mass. The temperature and salinity variability in the vertical structure is well represented in the model solution. We show that mean ocean mass transport across the multiple passages in the eastern Caribbean Sea compares favorably to observation-based estimates, but the model exhibits smaller variability and underestimates the mean Yucat & aacute;n Channel transport when compared to observations and ocean reanalysis estimates. Furthermore, a brief comparison against a 1 degrees CESM global ocean configuration shows that the higher-resolution regional model better represents the extent and seasonality of the Amazon River plume and hence better represents near-surface salinity and mixed-layer depth in the CS. Overall, the regional model reproduces the processes within the Caribbean Sea to a good degree and opens the possibility of regional ocean climate studies in support of decision-making within CESM.

Tropical cyclones (TCs), driven by heat exchange between the air and sea, pose a substantial risk to many communities around the world. Accurate characterization of the subsurface ocean thermal response to TC passage is crucial for accurate TC intensity forecasts and for an understanding of the role that TCs play in the global climate system. However, that characterization is complicated by the high-noise ocean environment, correlations inherent in spatio-temporal data, relative scarcity of in situ observations, and the entanglement of the TC-induced signal with seasonal signals. We present a general methodological framework that addresses these difficulties, integrating existing techniques in seasonal mean field estimation, Gaussian process modeling, and nonparametric regression into an ANOVA decomposition model. Importantly, we improve upon past work by properly handling seasonality, providing rigorous uncertainty quantification, and treating time as a continuous variable, rather than producing estimates that are binned in time. This ANOVA model is estimated using in situ subsurface temperature profiles from the Argo fleet of autonomous floats through a multi-step procedure, which (1) characterizes the upper ocean seasonal shift during the TC season; (2) models the variability in the temperature observations; (3) fits a thin plate spline using the variability estimates to account for heteroskedasticity and correlation between the observations. This spline fit reveals the ocean thermal response to TC passage. Through this framework, we obtain new scientific insights into the interaction between TCs and the ocean on a global scale, including a three-dimensional characterization of the near-surface and subsurface cooling along the TC storm track and the mixing-induced subsurface warming on the track's right side.

Monthly Ocean Heat Content Anomalies (OHCA) in the top 2000 dbar of the ocean are calculated (during 2004-2022, equatorward of 65 degree latitude) subtracting the mean over the period 2004-2022 from the monthly time series of OHC. Yearly OHCA time series are then calculated that include two points per year, i.e., from averaging Jan to Dec and Jul to Jun, respectively. OHC fields are mapped using locally stationary Gaussian processes (defined over space and time) with data-driven decorrelation scales (Kuusela and Stein, 2018). A linear time trend was included in the estimate of the mean field (along with spatial terms and harmonics for the annual cycle). Mapping is done separately for different vertical sections: 15-20 dbar, 15-300 dbar, 300-700 dbar, 700-1850 dbar, 1800-1850 dbar. The 15-20 dbar (1800-1850 dbar) section is used to estimate OHCA for 0-15 dbar (1850-2000 dbar), where observations are sparser. Different vertical sections are combined to estimate global OHCA time series for 0-2000 dbar. The attribute "area" is included in the netcdf files and it tells the corresponding surface area for the estimates. Regions of the ocean that are shallower than 300 m or are not sufficiently well sampled by the Argo array are not included. Ocean Heat Uptake is also calculated from the monthly OHCA and then averaged as described above to produce a yearly time series and included in the file. For the uncertainty at each time point, the standard deviation of each OHCA/OHU value in the time series is included. When plotting a time series, the user may consider, e.g., shading plus/minus 1* or 1.96*standard deviation (corresponding to a confidence level of 68% or 95% respectively). These standard deviations in the file are estimated using spatially and temporally dependent conditional simulations of monthly gridded anomalies. When combining different layers, the standard deviation of the sum is conservatively estimated as the sum of the standard deviations. Finally, OHCA/OHU trends are estimated via a least-squares fit and reported in the variable metadata with uncertainties (confidence level of 68%). Trend uncertainties are estimated by repeating the fit for each member of the conditional simulation ensemble described above.

The Earth climate system is out of energy balance, and heat has accumulated continuously over the past decades, warming the ocean, the land, the cryosphere, and the atmosphere. According to the Sixth Assessment Report by Working Group I of the Intergovernmental Panel on Climate Change, this planetary warming over multiple decades is human-driven and results in unprecedented and committed changes to the Earth system, with adverse impacts for ecosystems and human systems. The Earth heat inventory provides a measure of the Earth energy imbalance (EEI) and allows for quantifying how much heat has accumulated in the Earth system, as well as where the heat is stored. Here we show that the Earth system has continued to accumulate heat, with 381±61 ZJ accumulated from 1971 to 2020. This is equivalent to a heating rate (i.e., the EEI) of 0.48±0.1 W m−2. The majority, about 89 %, of this heat is stored in the ocean, followed by about 6 % on land, 1 % in the atmosphere, and about 4 % available for melting the cryosphere. Over the most recent period (2006–2020), the EEI amounts to 0.76±0.2 W m−2. The Earth energy imbalance is the most fundamental global climate indicator that the scientific community and the public can use as the measure of how well the world is doing in the task of bringing anthropogenic climate change under control. Moreover, this indicator is highly complementary to other established ones like global mean surface temperature as it represents a robust measure of the rate of climate change and its future commitment. We call for an implementation of the Earth energy imbalance into the Paris Agreement's Global Stocktake based on best available science. The Earth heat inventory in this study, updated from von Schuckmann et al. (2020), is underpinned by worldwide multidisciplinary collaboration and demonstrates the critical importance of concerted international efforts for climate change monitoring and community-based recommendations and we also call for urgently needed actions for enabling continuity, archiving, rescuing, and calibrating efforts to assure improved and long-term monitoring capacity of the global climate observing system. The data for the Earth heat inventory are publicly available, and more details are provided in Table 4.

Monthly Ocean Heat Content Anomalies (OHCA) in the top 2000 dbar of the ocean are calculated (during 2005-2022, in the North Atlantic, north of 20N) subtracting the time mean over the period 2005-2021 from the monthly time series of OHC. OHC fields are mapped using a locally stationary Gaussian process (defined over space) with data-driven decorrelation scales (Kuusela and Stein, 2018). A linear time trend was included in the estimate of the mean field (along with spatial terms and harmonics for the annual cycle). In this product, mapping is done in latitude and longitude with monthly subsets of data. Mapping is done separately for different vertical sections: 15-20 dbar, 15-300 dbar, 300-700 dbar, 700-1850 dbar, 1800-1850 dbar. The 15-20 dbar (1800-1850 dbar) section is used to estimate OHCA for 0-15 dbar (1850-2000 dbar), where observations are sparser. Different vertical sections are combined to estimate global OHCA for 0-2000 dbar. Regions of the ocean that are shallower than 300 m or are not sufficiently well sampled by the Argo array are not included.

Physics-based simulations of Arctic sea ice are highly complex, involving transport between different phases, length scales, and time scales. Resultantly, numerical simulations of sea-ice dynamics have a high computational cost and model uncertainty. We employ data-driven machine learning (ML) to make predictions of sea-ice motion. The ML models are built to predict present-day sea-ice velocity given present-day wind velocity and previous-day sea-ice concentration and velocity. Models are trained using reanalysis winds and satellite-derived sea-ice properties. We compare the predictions of three different models: persistence (PS), linear regression (LR), and convolutional neural network (CNN). We quantify the spatio-temporal variability of the correlation between observations and the statistical model predictions. Additionally, we analyze model performance in comparison to variability in properties related to ice motion (wind velocity, ice velocity, ice concentration, distance from coast, bathymetric depth) to understand the processes related to decreases in model performance. Results indicate that a CNN makes skillful predictions of daily sea-ice velocity with a correlation up to 0.81 between predicted and observed sea-ice velocity, while the LR and PS implementations exhibit correlations of 0.78 and 0.69, respectively. The correlation varies spatially and seasonally; lower values occur in shallow coastal regions and during times of minimum sea-ice extent. LR parameter analysis indicates that wind velocity plays the largest role in predicting sea-ice velocity on one-day time scales, particularly in the central Arctic. Regions where wind velocity has the largest LR parameter are regions where the CNN has higher predictive skill than the LR.