Selected work

Representative Papers

FlowCast-ODE: Continuous Hourly Weather Forecasting with Dynamic Flow Matching and ODE Integration

Sep 18, 2025

To address rapid error accumulation in autoregressive hourly weather forecasting and temporal discontinuities arising from ERA5’s 12-hour data assimilation cycle, this paper proposes a continuous-time modeling framework integrating dynamic flow matching with ordinary differential equations (ODEs). Methodologically, we design a conditional flow path formulation coupled with a low-rank AdaLN-Zero modulation mechanism, trained via a coarse-to-fine strategy that reduces model parameters by 15% without sacrificing accuracy. Experiments demonstrate significant improvements over strong baselines in RMSE, energy conservation, and fine-grained feature fidelity. The approach effectively mitigates assimilation-induced discontinuities, enhancing short-term forecast stability and temporal coherence. Moreover, it achieves state-of-the-art performance in predicting extreme events—particularly tropical cyclones—surpassing existing methods in both trajectory accuracy and intensity evolution.

0 citationsRead paper

CNCast: Leveraging 3D Swin Transformer and DiT for Enhanced Regional Weather Forecasting

Mar 16, 2025

Current nowcasting systems suffer from insufficient accuracy for short-term forecasts (1 hour to 5 days) and low-resolution precipitation diagnostics at regional scales. Method: This paper proposes a high-accuracy, hourly regional forecasting framework that integrates physical constraints with generative modeling. We introduce the first coupling of a 3D Swin Transformer with a latent diffusion transformer (DiT) to construct a multi-scale meteorological sequence generator; incorporate numerical weather prediction (NWP) boundary-condition embeddings to enforce physical consistency; and establish a novel 5-km-resolution, hourly precipitation diagnostic paradigm. Results: Experiments demonstrate that our model significantly outperforms the global benchmark Pangu-Weather across multiple key meteorological variables. It achieves substantial improvements in regional forecast accuracy and temporal stability, offering a scalable, physically interpretable technical pathway for high-resolution, short-term forecasting.

0 citationsRead paper

Skillful High-Resolution Ensemble Precipitation Forecasting with an Integrated Deep Learning Framework

Jan 06, 2025

To address the low accuracy and challenging uncertainty quantification in small-scale stochastic weather and extreme precipitation forecasting, this paper proposes a physics-informed deterministic–probabilistic dual-path deep learning framework for high-resolution (0.05°×0.05°) ensemble precipitation prediction. Methodologically, the deterministic branch employs a 3D Swin Transformer to model mesoscale precipitation structures, while the probabilistic branch innovatively embeds physical priors into a latent-space conditional diffusion model to characterize convective-scale residual uncertainties. Our key contribution is the first dual-path coupled architecture enabling interpretable, unbiased, and reliable ensemble forecasts. Experiments demonstrate significant improvements in the Critical Success Index (CSI) and spatial detail fidelity; rank histograms confirm ensemble reliability; and case studies of intense rainfall in South China show superior performance over ERA5, with robust 5-day real-time forecasting capability.

0 citationsRead paper

Recent publications

Latest Papers

FlowCast-ODE: Continuous Hourly Weather Forecasting with Dynamic Flow Matching and ODE Integration

Sep 18, 2025

To address rapid error accumulation in autoregressive hourly weather forecasting and temporal discontinuities arising from ERA5’s 12-hour data assimilation cycle, this paper proposes a continuous-time modeling framework integrating dynamic flow matching with ordinary differential equations (ODEs). Methodologically, we design a conditional flow path formulation coupled with a low-rank AdaLN-Zero modulation mechanism, trained via a coarse-to-fine strategy that reduces model parameters by 15% without sacrificing accuracy. Experiments demonstrate significant improvements over strong baselines in RMSE, energy conservation, and fine-grained feature fidelity. The approach effectively mitigates assimilation-induced discontinuities, enhancing short-term forecast stability and temporal coherence. Moreover, it achieves state-of-the-art performance in predicting extreme events—particularly tropical cyclones—surpassing existing methods in both trajectory accuracy and intensity evolution.

0 citationsRead paper

CNCast: Leveraging 3D Swin Transformer and DiT for Enhanced Regional Weather Forecasting

Mar 16, 2025

Current nowcasting systems suffer from insufficient accuracy for short-term forecasts (1 hour to 5 days) and low-resolution precipitation diagnostics at regional scales. Method: This paper proposes a high-accuracy, hourly regional forecasting framework that integrates physical constraints with generative modeling. We introduce the first coupling of a 3D Swin Transformer with a latent diffusion transformer (DiT) to construct a multi-scale meteorological sequence generator; incorporate numerical weather prediction (NWP) boundary-condition embeddings to enforce physical consistency; and establish a novel 5-km-resolution, hourly precipitation diagnostic paradigm. Results: Experiments demonstrate that our model significantly outperforms the global benchmark Pangu-Weather across multiple key meteorological variables. It achieves substantial improvements in regional forecast accuracy and temporal stability, offering a scalable, physically interpretable technical pathway for high-resolution, short-term forecasting.

0 citationsRead paper

Skillful High-Resolution Ensemble Precipitation Forecasting with an Integrated Deep Learning Framework

Jan 06, 2025

To address the low accuracy and challenging uncertainty quantification in small-scale stochastic weather and extreme precipitation forecasting, this paper proposes a physics-informed deterministic–probabilistic dual-path deep learning framework for high-resolution (0.05°×0.05°) ensemble precipitation prediction. Methodologically, the deterministic branch employs a 3D Swin Transformer to model mesoscale precipitation structures, while the probabilistic branch innovatively embeds physical priors into a latent-space conditional diffusion model to characterize convective-scale residual uncertainties. Our key contribution is the first dual-path coupled architecture enabling interpretable, unbiased, and reliable ensemble forecasts. Experiments demonstrate significant improvements in the Critical Success Index (CSI) and spatial detail fidelity; rank histograms confirm ensemble reliability; and case studies of intense rainfall in South China show superior performance over ERA5, with robust 5-day real-time forecasting capability.

0 citationsRead paper