The key role of the planetary boundary layer height (PBLH) in pollution, climate, and model forecasting has long been recognized. However, the observed PBLH has rarely been used to evaluate numerical weather prediction models in China. We compared the temporal and spatial characteristics of the bias in the PBLH in China predicted by the CMA-GFS model with vertical high-resolution sounding data and Global Positioning System occultation data from 2019 to 2020. We found that: (1) The PBLH in East China is systematically underestimated by the CMA-GFS model. The bias mainly results from the underestimation of the wind shear in the boundary layer, a smaller sensible heat flux near the surface, and a lower surface temperature. The combined effects of these factors inhibit the boundary layer from developing to a higher height, although the most important contributor is the small sensible heat flux. (2) There is a systematic overestimation of the PBLH over the Tibetan Plateau throughout the year. The bias is mainly a result of the smaller buoyancy, higher wind shear, and larger sensible heat flux forecast by the CMA-GFS model, which drive the boundary layer to develop to a significantly deeper height than the observations. This bias in the CMA-GFS model is mainly caused by the bias in the sensible heat flux and wind shear forecasts. In contrast, the CMA-GFS model underestimates the PBLH in the Tarim Basin. Our preliminary analysis shows that the boundary layer forecasted is unable to develop because the buoyancy effect of the model is too strong. Therefore, the bias of the predicted PBLH by the CMA-GFS model in China is mainly caused by inaccuracies in the sensible heat flux and wind shear forecasts.

With growing global demand for clean energy, offshore wind resource assessment is crucial. This study comprehensively utilizes multi-source Synthetic Aperture Radar (SAR) satellite data, including European Sentinel-1 and Chinese Gaofen-3 satellites, along with in-situ wind mast measurements, to systematically assess the wind energy resources in the offshore area of Zhuanghe, Liaoning Province, China, from 2017 to 2022. Sea surface wind fields were retrieved through data preprocessing, polarization conversion, geophysical model inversion, and multi-source data fusion. By coupling a vertical extrapolation model that accounts for sea state effects, the 10‑m wind speeds were extrapolated to turbine hub heights of 40 m, 80 m, and 100 m, enabling the calculation of key wind energy parameters such as wind speed and wind power density. Results indicate that at 100 m height, the annual average wind speed is 6.40 m·s−1 and the annual average wind power density is 312.8 W·m−2, corresponding to wind class 2. Wind speeds are higher in spring and winter and lower in summer and autumn, with prevailing northerly winds. Compared with ERA5 reanalysis data, SAR data offer advantages including high spatial resolution, rich historical records, and superior local accuracy, making them suitable for detailed offshore wind resource assessment. This research fills the gap in high‑precision wind energy evaluation under complex meteorological and data‑scarce conditions in the region, provides reliable data support for local offshore wind farm planning and development, and offers transferable experience for the application of SAR technology in similar coastal areas in China.

To address the nonlinear amplification of numerical weather prediction (NWP) errors and the difficult trade-off between coverage and sharpness in short-term offshore wind power forecasting, this paper proposes PRWind, a physics-guided framework for short-term probabilistic forecasting. Built upon a Transformer encoder, the framework integrates horizon-adaptively weighted power-curve, persistence, and wind-speed cubic-law baselines; propagates wind-speed uncertainty into a power distribution through a probabilistic wind-speed distribution with Monte Carlo integration; and employs spatial graph convolution to model inter-turbine correlations, together with online bias correction and segmented post hoc interval calibration. The framework is validated on two offshore wind farms in coastal China (8350 kW and 10,000 kW units), with all data analyzed at hourly resolution and forecasts issued for 23 hourly steps ahead. For point forecasting, the NMAE reaches 8.74% and 8.24%, with MAE reductions of 19.1% (95% CI: 18.1–19.9%) at Wind Farm A and 7.8% (95% CI: 7.3–8.4%) at Wind Farm B relative to the strongest baselines, both statistically significant under the Diebold–Mariano test (p < 0.001). For probabilistic forecasting, the PICP remains stable at 0.89, balancing coverage and sharpness. The error of PRWind accumulates markedly more slowly with the forecast horizon than that of the baselines, and ablation studies confirm the synergistic effect of the modules. Trained and evaluated independently on the two wind farms, PRWind performs consistently, demonstrating cross-wind-farm stability (rather than zero-shot generalization) conferred by physical guidance, and offering significant engineering value for intra-day and day-ahead offshore wind power dispatch.