19 July 2025 in Nature Communications
Weather forecasting traditionally relies on numerical weather prediction (NWP) systems that integrate global observations, data assimilation (DA), and physics-based models. However, further advances are increasingly constrained by high computational costs, the underutilization of vast observational datasets, and challenges in obtaining finer resolution. Recent advances in machine learning present a promising alternative, but still depend on the initial conditions generated by NWP systems. Here, we introduce FuXi Weather, a machine learning-based global forecasting system that assimilates multi-satellite data and is capable of cycling DA and forecasting. FuXi Weather generates reliable 10-day forecasts at 0.25° resolution using fewer observations than conventional NWP systems. It demonstrates the value of background forecasts in constraining the analysis during DA. FuXi Weather outperforms the European Centre for Medium-Range Weather Forecasts high-resolution forecasts beyond day one in observation-sparse regions such as central Africa, highlighting its potential to improve forecasts where observational infrastructure is limited.
14 November 2024 in Geophysical Research Letters
Recent machine learning (ML)‐based weather forecasting models have improved the accuracy and efficiency of forecasts while minimizing computational resources, yet still depend on traditional data assimilation (DA) systems to generate analysis fields. Four dimensional variational data assimilation (4DVar) enhances model states, relying on the prediction model to propagate observation to the initial field. Consequently, the initial fields from traditional DA are not optimal for ML‐based models, necessitating a customized DA system. This paper introduces an ensemble 4DVar system integrated with the FuXi model (FuXi‐En4DVar), which can independently generate accurate analysis fields. It utilizes automatic differentiation to compute gradients, and demonstrates the equivalence of these gradients with those derived from adjoint models. Experimental results indicate that this system preserves the physical balance of the analysis field and exhibits flow‐dependent characteristics. These features enhance the propagation and assimilation of observation into the initial analysis field, thereby improving the accuracy of the analysis fields.
22 September 2023 in IEEE Transactions on Multimedia
Zero-shot action recognition (ZSAR) aims to recognize unseen action categories in the test set without corresponding training examples. Most existing zero-shot methods follow the feature generation framework to transfer knowledge from seen action categories to model the feature distribution of unseen categories. However, due to the complexity and diversity of actions, it remains challenging to generate unseen feature distribution, especially for the cross-dataset scenario when there is potentially larger domain shift. This paper proposes a De confounding Ca usa l GAN (DeCalGAN) for generating unseen action video features with the following technical contributions: 1) Our model unifies compositional ZSAR with traditional visual-semantic models to incorporate local object information with global semantic information for feature generation. 2) A GAN-based architecture is proposed for causal inference and unseen distribution discovery. 3) A deconfounding module is proposed to refine representations of local object and global semantic information confounder in the training data. Action descriptions and random object feature after causal inference are then used to discover unseen distributions of novel actions in different datasets. Our extensive experiments on C ross- D ataset Z ero- S hot A ction R ecognition (CD-ZSAR) demonstrate substantial improvement over the UCF101 and HMDB51 standard benchmarks for this problem.
6 December 2024 in Journal of Geophysical Research: Atmospheres
20 April 2026 in Journal of Advances in Modeling Earth Systems
Accurate modeling of hydrometeor bulk scattering properties (BSPs) is essential for the effective assimilation of satellite microwave observations in cloudy and precipitating conditions. Nevertheless, current radiative transfer models use oversimplified hydrometeor BSP parameterizations, leading to significant simulation errors and biases that limit the full potential of all‐sky assimilation. To address this challenge, this study developed PhySCAT‐Net, a physics‐informed deep learning (DL) framework that integrates forward and Jacobian operators of a physical radiative transfer model into the neural network training, enabling efficient optimization of the BSP models against satellite observations. Applied to vertically and horizontally polarized radiances from the Global Precipitation Measurement Microwave Imager 166.5 GHz channels, the framework selectively fine‐tunes the snow BSP model while temporarily fixing other hydrometeor types. Results demonstrate that the optimized DL model substantially improves agreement between simulated and observed brightness temperatures. Across most regions globally, mean observation‐minus‐background (O‐B) biases are reduced to within ±1K, and the Jensen‐Shannon divergence decreases by orders of magnitude. The error distributions, which were previously highly skewed and therefore problematic for data assimilation, are now roughly symmetrical. Furthermore, PhySCAT‐Net enables the DL model to extract polarimetric information of non‐spherical ice particles directly from observed radiances, demonstrating superior performance compared to existing empirical schemes. It successfully reproduces the distributions of polarization differences and their non‐monotonic relationship with brightness temperature.