Fusion of ocean data from multiple sources using deep learning: Utilizing sea temperature as an example

: The feasibility of global ocean weather prediction was just emerging as the Global Ocean Data Assimilation Experiment (GODAE) began in 1997. Ocean weather includes phenomena such as meandering currents and fronts, the surface mixed layer and sea surface temperature (SST), waves, upwelling of cold water, all influencing ocean variables such as temperature (T), salinity (S), currents, and sea surface height (SSH). Adequate real-time data input, computing power, numerical ocean models, data assimilation capabilities, atmospheric forcing, and bathymetric/boundary constraints are essential to make such prediction possible. The ocean models dynamically interpolate data in conjunction with data assimilation, convert atmospheric forcing into oceanic responses, and forecast the ocean weather. The results are substantially influenced by ocean model simulation skill and it is advantageous to use an ocean model that is eddy-resolving, not just eddy-permitting. Because the most abundant ocean observations are satellite surface data, and subsurface data are very sparse, downward projection of surface data is a key challenge in ocean data assimilation. The need for accurate prediction of ocean features that are inadequately observed places a major burden on the ocean model, data assimilation, and atmospheric forcing. The sensitivity of ocean phenomena to atmospheric forcing and the time scale for response affect the oceanic data requirements and prediction system design. Outside of surface boundary layers and shallow regions, forecast skill is about one month globally and over many subregions, and is only modestly reduced by using climatological forcing. In addition, global ocean prediction systems must demonstrate the ability to provide initial and boundary conditions to nested regional and coastal models that enhance their predictive skill.

In this study, the accuracy of ocean analysis data, which are produced from the Korea Meteorological Administration (KMA) Nucleus for European Modelling of the Ocean/Variational Data Assimilation (NEMO/NEMOVAR, hereafter NEMO) system and the HYbrid Coordinate Ocean Model/Navy Coupled Ocean Data Assimilation (HYCOM/NCODA, hereafter HYCOM) system, was evaluated using various oceanic observation data from March 2015 to February 2016. The evaluation was made for oceanic thermal environments in the tropical Pacific, the western North Pacific, and the Korean peninsula. NEMO generally outperformed HYCOM in the three regions. Particularly, in the tropical Pacific, the RMSEs (Root Mean Square Errors) of NEMO for both the sea surface temperature and vertical water temperature profile were about 50% smaller than those of HYCOM. In the western North Pacific, in which the observational data were not used for data assimilation, the RMSE of NEMO profiles up to 1000 m (0.49℃) was much lower than that of HYCOM (0.73℃). Around the Korean peninsula, the difference in RMSE between the two models was small (NEMO, 0.61oC; HYCOM, 0.72oC), in which their errors show relatively big in the winter and small in the summer. The differences reported here in the accuracy between NEMO and HYCOM for the thermal environments may be attributed to horizontal and vertical resolutions of the models, vertical coordinate and mixing scheme, data quality control system, data used for data assimilation, and atmosphere forcing. The present results can be used as a basic data to evaluate the accuracy of NEMO, before it becomes the operational model of the KMA providing real-time ocean analysis and prediction data.

Contribution of thermal expansion to present-day sea-level change revisited

: The goal of this project is to improve the forecast performance of shallow-water and limited-area ocean forecast systems through improvement in ocean data analysis and assimilation techniques. The Shallow Water Analysis and Forecast System (SWAFS) is the operational ocean nowcast/forecast system at the Naval Oceanographic Office (NAVOCEANO). The objective of this project is to improve the forecast skill of SWAFS by integrating into the NAVOCEANO system the ocean data assimilation component developed for the Coupled Ocean Atmosphere Mesoscale Prediction System (COAMPS). The ocean data quality control, analysis, and assimilation components of SWAFS will be upgraded.

Globally, 39% of the world's population lives within 100 km of the coast. Eight of the top ten largest cities in the world are located along the coast. In South Korea, 27% of the population also resides in coastal areas. Ocean waves, induced by sea winds, potentially endanger offshore infrastructure and threaten low-lying ecosystems and communities due to coastal erosion and flooding. More accurate prediction of ocean wave patterns, or sea states, is crucial for informed decision-making in mitigating these risks. Recently, many National Meteorological and Hydrological Services (NMHs) have been endeavoring to enhance the accuracy of ocean wave predictions by assimilating in-situ and remote sensing observations. The Korea Meteorological Administration (KMA) also initiated an ocean wave data assimilation system in 2021. However, this system has thus far only been adapted to a global-scale model. This study aims to install and evaluate the impact of ocean wave data assimilation on a regional scale. The regional ocean wave data assimilation system employs optimal interpolation based on WAVEWATCH-Ⅲ version 6.07 with a spatial resolution of 1/30° targeted for the East Asian region. Significant wave height data from five satellites in polar orbit, ocean data buoys, and coastal wave buoys were utilized. Numerical experiments for summer (2023JJA) and winter (2023/24DJF) reveal that the use of data assimilation reduced the root mean square error by 49.2% and 38.6%, respectively, for the initial fields of regional wave models. The assimilated initial fields improved ocean wave predictions by 12 hours in the KMA regional ocean wave model, which is consistent with previous research in the field. Particularly during Typhoon KHANUN in August 2023, there was a tendency to overestimate sea winds, which are input variables of the ocean wave model. Despite the use of overestimated sea winds, the regional ocean wave model utilizing data assimilation reduced the error by up to 94 cm.

The coupled data assimilation (CDA) system consisting of ocean, sea‐ice, and atmosphere data assimilation components with the Beijing Climate Center (BCC) Climate System Model has been developed to provide reliable analyses of the atmosphere, ocean, and sea‐ice states. It incorporates ocean temperature/salinity profiles, sea surface temperature, sea level height, and sea‐ice concentration observations at a daily frequency, and atmosphere reanalysis at a 6‐hourly frequency. Results show that the system is capable of realistically reproducing the climatology and variability of ocean, sea‐ice, and atmosphere. The performances in analyzing ocean component are comparable to those of well‐known ocean reanalyses that were once used to initialize the BCC model for climate predictions. A series of experiments with and without sea‐ice observations in the CDA framework are designed to explore the role of sea‐ice data assimilation (DA). The addition of sea‐ice DA exerts very small influence to the analysis of upper ocean temperature over the Arctic area, but leads to an evident reduction of temperature error in the upper 1000 m of ocean south of 60°S. Particularly, only the inclusion of sea‐ice DA can make the ocean/ocean‐atmosphere DA effective in providing skillful analysis in the high‐latitude Southern Ocean. Furthermore, we find that only on the basis of ocean DA, can the addition of sea‐ice concentration assimilation improve the analysis of tropical tropospheric atmosphere, along with a better analysis of mid‐ and high‐latitude stratospheric atmosphere. These results address the importance of coordination of sea‐ice observations and ocean observations in CDA.

Ocean acoustic propagation models used for sonar performance prediction often rely on global or regional ocean modeling and data assimilation, or monthly climatologic averages, for estimating the 4D ocean temperature and salinity (T/S) field. Ocean models rely heavily on satellite-based ocean surface measurements, e.g., sea surface temperature (SST) and height (SSH), plus typically vastly less resolved in space and time vertical water column measurements derived using ship-based and/or ARGO float instruments. Systems also exist that create 3D synthetic T/S fields from satellite SST/SSH measurements, e.g., the Modular Ocean Data Assimilation System (MODAS) and the more recent Improved Synthetic Ocean Profile (ISOP) system. It is unclear, however, how well these synthetic T/S fields capture significant ocean ducting conditions. Similarly, climatologic averages combine historical vertical T/S profile data in spatial cells by month; but the averaging process tends to blur out important features. This talk will address the use of 2021 New England Shelf Break Acoustics (NESBA) experiment and ARGO vertical profile data for: (1) testing the degree to which ocean models or climatology capture important acoustic features under a range of conditions; and (2) providing more realistic sonar performance prediction means and uncertainties. [Work supported by the Office of Naval Research].

In the previous study, the authors have produced an eddy-resolving ocean ensemble analysis product called the local ensemble transform Kalman filter (LETKF)-based ocean research analysis (LORA) over the western North Pacific and Maritime Continent regions using an ocean data assimilation system driven by the Japanese operational atmospheric reanalysis dataset known as the JRA-55. However, the LORA includes warm biases in sea surface temperatures (SSTs) in coastal regions during the boreal winter. In this study, we perform sensitivity experiments with atmospheric forcing using an ocean forcing dataset known as the JRA55-do, which adjusts the JRA-55 to high-quality reference datasets to reduce biases and uncertainties. The results show that the nearshore warm SST biases are significantly improved by the JRA55-do. During the boreal autumn, the improvement comes from mainly two factors: (i) enhancement of surface cooling by latent heat releases caused by removing contamination of weak winds at the land grid cells, and (ii) weakening surface heating by downward shortwave radiation through the adjustment in the JRA55-do.During the boreal winter, enhanced cooling by the analysis increments suppresses the growth of the warm SST biases when the JRA55-do is used. However, if the JRA-55 dataset is used, the adaptive observation error inflation (AOEI) scheme acts negatively to keep the nearshore SST biases in winter. Based on the innovation statistics, the AOEI inflates the observation errors when the differences between the squared observation-minus-forecast innovations and the squared forecast ensemble spreads are larger than the prescribed observation error variance, and improves the accuracy in the open ocean, especially around the frontal regions. However, when substantial warm SST biases are formed in the previous season, AOEI's observation error inflation makes the analysis increments smaller and cannot suppress the warm biases.We also validate the analysis accuracy using various data such as sea surface height and horizontal velocities and find that the JRA55-do has significant advantages. Therefore, continuous maintenance and development of ocean forcing datasets are essential for ocean modeling and data assimilation.

Predicting subsurface thermohaline structure from remote sensing data based on long short-term memory neural networks

A coupled atmosphere–ocean data assimilation system, the Meteorological Research Institute‐Coupled Data Assimilation System Version 1 (MRI‐CDA1), was developed based on the coupled atmosphere–ocean general circulation model and separate atmosphere and ocean analysis routines operated by the Japan Meteorological Agency (JMA). To implement coupled atmosphere–ocean data assimilation, 6‐hr data assimilation cycles with the coupled model in the outer loop are adopted in atmospheric data assimilation, whereas incremental analysis updates with 10‐day data assimilation cycles are adopted in ocean data assimilation. A coupled data assimilation (reanalysis) experiment (CDA‐Exp) is conducted using MRI‐CDA1 along with an uncoupled reanalysis experiment (UCPL‐Exp) in which the same atmospheric component of the coupled model is forced by prescribed sea surface temperature (SST) similarly to the JMA reanalysis, JRA‐55. Climatological precipitation and variations in precipitation and sea surface air temperature (SAT) are better represented in CDA‐Exp than in JRA‐55, which is brought about by the modification of the atmosphere model physics of the coupled model to better represent climate states. In CDA‐Exp, the SST adjustment to the atmosphere amplifies the lead/lag correlations between subseasonal variations of SST and precipitation. The atmosphere–ocean coupling generates SST variations associated with tropical instability waves, and the SAT field responds to SST variations. The SST–precipitation and SST–SAT relationships on the weather timescale are also recovered in CDA‐Exp, although they are hardly seen in UCPL‐Exp. The coupled model physics generates weather‐timescale SST variations consistent with the atmospheric state, and the atmospheric parameters respond to the SST variations through the coupled model physics. This study suggests that the benefits of coupled data assimilation would be more evident if the ability of MRI‐CDA1 to represent SST variations and its interaction with the atmosphere are improved.

[1] Annual rankings of global temperature are an important component of climate monitoring. However, there is some degree of uncertainty for every yearly value in the global temperature time series, which leads to uncertainty in annual rankings as well. This study applies a Monte Carlo uncertainty analysis to the National Oceanic and Atmospheric Administration (NOAA) National Climatic Data Center's global land-ocean surface temperature (NOAATMP) time series. Accounting for persistence between years does not materially affect the results versus presuming statistical independence. The highest probabilities for the warmest year analysis (1880–2012) are associated with the years 2010 (~36%), 2005 (~28%), and 1998 (~11%). The current separation among the warmest observed years is relatively small compared to the standard errors of the NOAATMP time series. However, each year between 1997 and 2012 was warmer than the vast majority of all other years since 1880 at the 95% confidence level.

During the 2023-24 El Niño, the July 2023–June 2024 average global mean surface temperature peaked at 1.58 °C above pre-industrial levels, associated with a record 0.36 °C year-over-year rise. Here we use statistical models and a Green’s Function method to explore the causes for this rise. We show that sea surface temperature accounts for ~92% of the interannual warming, highlighting the critical role of El Niño diversity. Unlike typical El Niños, dual tropical Pacific warming centers in 2023-24 and an eastward-extended North Atlantic jet stream synergistically amplified sea surface temperatures across tropical basins, accelerating the pace of global warming. From an energy balance perspective, ocean heat content accumulated during the preceding La Niña and a continuous increase in absorbed shortwave radiation over the ocean drove sea surface temperature rise. Accelerated warming may push the climate system closer to critical tipping points, emphasizing the need for enhanced monitoring, mitigation, and adaptation strategies. The 2023-24 El Niño event led global mean surface temperatures to surpass 1.5 degrees above pre-industrial levels, according to analysis of surface air temperature and ocean heat content data.

Surface winds from atmospheric reanalysis lead to contrasting oceanic forcing and coastal upwelling patterns

<p>The purpose of employing data assimilation methods in operational ocean forecasting systems is to provide good initialization to the models and is dependent on good quality ocean observations being assimilated. Using or accepting erroneous data can result in an inaccurate analysis and alternatively, rejecting extreme or valid data can result in missing important events.</p><p>In this study two ocean-sea ice coupled systems are considered: HYCOM-CICE4 and MOM6-CICE6 at ¼-deg horizontal resolution and 41 vertical layers. The two coupled models are initialized from the World Ocean Atlas 2018 (WOA) temperature and salinity climatology for a period of 20 years. Both models are forced with GEFS (Global Ensemble Forecast System created by the National Centers for Environmental Prediction: NCEP). The data assimilation is performed on a 24-hr cycle using RTOFS-DA (Real-time Ocean Forecasting system-DA; 3DVAR) for HYCOM-CICE4 and SOCA (Sea ice Ocean Coupled Assimilation; 3DVAR) for MOM6-CICE6 to compare the data Quality Control (QC) methods. The ocean data being assimilated include satellite sea surface temperature (SST) and sea surface salinity (SSS), in-situ temperature & salinity, absolute dynamic topography (ADT), sea ice concentration.</p><p>The QC in RTOFS-DA and SOCA are fully automated and are performed through various filters applied (e.g., land-sea area fraction to eliminate satellite data near the coast, temperature inversion elimination in in-situ profile data, etc). The various QC methods in both DA systems are described. The results of the analysis and 24-forecast are compared against independent observations and statistics of the data accepted and rejected between the two DA systems are presented and discussed.</p>

Estimating diurnal variations of Sea Surface Temperature (SST) is important for studying air-sea heat exchange. Existing operational diurnal SSTs are derived from numerical models incorporating satellite data, and assimilated with in-situ measurements, which are very accurate. However, numerical model-based methods incur significant computational costs for identifying the diurnal cycle of SST from various heat flux sources (i.e., sensible, latent heat). In this study, we first proposed a Generative Adversarial Network (GAN) method to reconstruct high-resolution diurnal SST using satellite observations as an actual diurnal signal from the ocean surface layer. A generator in the GAN model was trained using the diurnal variability-related variables, including the hourly SSTs and shortwave radiation measurements from Himawari-8 geostationary satellite observations, to estimate diurnal SSTs. The discriminator in the GAN model was learned to reduce the difference in spatiotemporal variability of diurnal SSTs between a satellite data-assimilated numerical model product (Global Ocean OSTIA Diurnal Skin Sea Surface Temperature; Copernicus marine service) and estimated SST from the generator. The results showed that the reconstructed SST had a better spatial distribution of ocean phenomena such as front and eddy than compared with the numerical model-derived SST. It implied that the GAN model could simulate a high spatial variability of SSTs using satellite-based data with a spatial resolution of 2km. The proposed GAN model produced high validation accuracy, resulting in the coefficient of determination of 0.99, bias of -0.2℃, and root mean square errors of 0.58℃ when compared with in situ SST Quality Monitor drifting buoy data. Since we use geostationary satellite data, the proposed model can capture real diurnal variability of SST more frequently than existing numerical model data using analysis data. In addition, the proposed deep learning model is much more computationally efficient than the numerical models.