Deep Learning-Based Statistical Downscaling of Sea Surface Temperature Using a Residual Corrective Neural Network

OceanBench: A Benchmark for Data-Driven Global Ocean Forecasting systems

OceanBench is a benchmark designed to evaluate and accelerate global short-range data-driven ocean forecasting, constructed from a curated dataset comprising first-guess trajectories, nowcasts, and atmospheric forcings from operational physical ocean models, typically unavailable in public datasets due to assimilation cycles.

Advancing oceanographic forecasts in the Bay of Bengal: an AI-based spatiotemporal prediction for SST and MLD

Deep learning for multi-step forecasting of stratospheric wind and temperature profiles over the Tibetan plateau

Accurate prediction of stratospheric wind and temperature profiles is essential for understanding regional atmospheric dynamics over the complex terrain of the Tibetan Plateau. However, conventional numerical weather prediction models are computationally expensive, while statistical time-series models are limited in capturing the nonlinear evolution of atmospheric variables across both temporal and vertical dimensions. To address these limitations, MTPV-HDRNet is proposed as a fixed-site, multi-level intelligent forecasting model driven by ECMWF ERA5 pressure-level reanalysis data, with the fixed site defined as a selected ERA5 grid cell rather than an observational station. MTPV-HDRNet jointly predicts zonal wind (U), meridional wind (V), and temperature (T) for the next 24 h across 11 ERA5 pressure levels from 100 to 1 hPa, which correspond to heights of approximately 16–48 km. The model adopts a hybrid encoder–decoder architecture that explicitly represents temporal evolution and vertical stratification in a decoupled but complementary manner, thereby enhancing its ability to capture multi-lead-time profile evolution and cross-level dependencies. The model was trained using ERA5 data from 2016 to 2022, validated using data from 2023, and independently tested using data from 2024 at the fixed site in the central Tibetan Plateau. Results demonstrate that MTPV-HDRNet consistently outperforms the baseline models at all forecast lead times. On the 2024 test set, the mean RMSEs are 3.48 m s − 1 for U wind, 3.54 m s − 1 for V wind, and 1.62 K for temperature. Compared with the strongest baseline, ConvLSTM, MTPV-HDRNet reduces the RMSE by 11.9%, 11.3%, and 7.4% for U wind, V wind, and temperature, respectively. The model also maintains strong correlation performance at longer lead times, demonstrating its robustness in fixed-site atmospheric profile forecasting over complex terrain.

DLESyM-Ocean: A Deep Learning Probabilistic Global Model for Simulating Present-Day Upper Ocean and Sea Ice

While AI has shown remarkable promise in atmospheric and meteorological forecasting, accurately simulating other components of the Earth system with AI remains an active frontier. We present DLESyM-Ocean, a Deep Learning Earth System Model that simulates global present-day sea ice and upper ocean conditions. Unlike conventional probabilistic models optimized via diffusion objectives or losses such as continuous-ranked probability score, DLESyM-Ocean is trained using a patch energy score loss. When driven by atmospheric forcing, DLESyM-Ocean produces a well-calibrated, spatially coherent, and skillful ensemble of sea ice and upper ocean conditions with minimal bias relative to reanalysis products. DLESyM-Ocean is stable when autoregressively run for multi-year simulations and produces a climatology and variability with minimal bias compared with reanalysis. We evaluate case studies including a recent sea ice extreme, a severe marine heatwave, the 2023 El Ni\~no transition, and the 2023 spike in global mean temperature. In all of these case studies, DLESyM-Ocean produces realistic surface and subsurface trajectories and ample ensemble diversity in response to common atmospheric forcing, suggestive of learned autoregressive ocean dynamics. When coupled with other Earth system components, such as the atmosphere, the computational efficiency of DLESyM-Ocean makes it a promising tool for subseasonal to seasonal forecasting.

LangYa: a large AI model for global ocean forecasting.

Ocean forecasting is crucial for both scientific research and societal benefits. Large artificial intelligence (AI)-based models have recently boosted forecasting efficiency and accuracy. However, it remains challenging to develop a comprehensive AI-driven ocean forecasting system capable of integrating cross-spatiotemporal and atmospheric forcing. This study introduces LangYa, a cross-spatiotemporal and atmospheric forcing ocean forecasting system featuring: (1) a large-language-model-based (LLM-based) time embedding to explicitly represent forecast lead times, (2) an asynchronous cross-iterative random sampling strategy to represent the impacts of atmospheric forcing on ocean processes, (3) an ocean self-attention module to enhance network stability and accelerate training convergence, and (4) an adaptive loss function to capture ocean dynamics in the thermocline, at depths ranging from tens of meters to about 300 m. LangYa is trained on 27 years of global ocean data from the Global Ocean Reanalysis and Simulation, version 12 (GLORYS12). Using reanalysis and observational data, compared to existing open-source AI-based forecasting systems and numerical models, LangYa enables a single model to produce forecasts with lead times of 1 to 7 d (1/12°, daily) and achieves 7 d RMSEs below 0.0736 m/s, 0.0701 m/s, 0.4376 ℃, and 0.1302 psu for global currents, temperature, and salinity respectively. These quantitative results indicate that LangYa provides clear advantages in forecast accuracy, lead-time robustness, and stability for global OSV forecasting, demonstrating its potential for real-time operational deployment.

Global reanalysis from observations alone with machine learning

Earth system reanalysis datasets are foundational for weather and climate research and provide the gridded training data used by most machine learning weather prediction systems. Here we show results from a prototype system that suggest that machine learning models trained only on Earth system observations can potentially be used to generate multi-decade global reanalyses without using physics-based numerical models. The resulting gridded fields capture large-scale atmospheric structure and variability across multiple timescales, while exhibiting signs of physical coherence in several key dynamical diagnostics. Evaluations of the prototype against held-out independent atmospheric observations indicate that the root mean square vector error of upper-level winds is close to that of ERA5 when compared at a consistent resolution, and that the standard deviation of the error at the surface is between that of 4th- and 5th-generation ECMWF reanalyses (ERA-Interim and ERA5). Furthermore, while traditional reanalysis production is computationally expensive, typically taking several years to produce, the reanalysis presented here was generated during the course of a single working day. These results suggest that observation-trained machine learning models offer a promising new approach for reanalysis production from observations alone.