Sea surface wind has been listed as an essential climate variable, playing crucial roles in regulating the global and regional weather and climate. Spaceborne synthetic aperture radar (SAR) has demonstrated the advantages in observing the wind field given its all-weather measurement capability. In this study, we present a convolutional neural network (CNN)-based framework for retrieving 10 m wind speed () from Sentinel-1 SAR wave mode (WV) imagery. The model is trained on SAR data acquired in 2017 using collocated ERA5 reanalysis wind vectors as the reference, with final performance evaluated against a temporally independent dataset from 2016 and in situ wind measurements. The CNN approach demonstrates improved retrieval accuracy compared to the conventional CMOD5.N-based result, achieving lower root mean square error (RMSE) and bias across both WV1 and WV2 incidence angle modes. Residual diagnostics show a systematic overestimation at low wind speeds and a slight underestimation at higher wind speeds. Spatial analyses of retrieval bias reveal regional variations, particularly in areas characterized by ocean swell or convective atmospheric activity, highlighting the importance of geophysical features in retrieval accuracy. These results support the viability of deep learning approaches for SAR-based ocean surface wind estimation and suggest a path forward for the development of more accurate, data-driven wind products suitable for both scientific research and operational marine forecasting.
This study focused on the measurement requirements of Digital Image Correlation (DIC) in high-temperature environments of aero-engines and systematically investigated the applicability and stability of high-temperature speckle materials. Five common coating materials (Ti, TiN, Ta, NiCr alloy, and SiC) were selected. Corresponding thin films were deposited on Al2O3 ceramic substrates using magnetron sputtering technology, and their surface color evolution from room temperature up to 1200 °C was examined. The film compositions were analyzed by Raman spectroscopy and X-ray photoelectron spectroscopy (XPS), revealing the mechanisms behind the color changes. The results indicate that Ti, TiN, Ta, and NiCr alloy exhibit significant color variations, which leads to insufficient color contrast for high-temperature speckle patterns. Further investigation shows that depositing an outer SiO2 coating can improve surface scattering and reflection, while also inhibiting oxygen penetration, thereby enhancing oxidation resistance and improving speckle contrast. The SiC/SiO2 composite structure demonstrates excellent thermal stability, making it an ideal speckle material for high-temperature DIC measurements.
High-temperature two-dimensional digital image correlation (2D-DIC) can produce spatially unstable strain fields and biased condition-averaged incremental strains, but these error signatures need not respond to the same processing. A smoother strain field is therefore not, by itself, evidence of improved scalar accuracy. We evaluated strain field visualization stability separately from extensometer-referenced, condition-level incremental strain accuracy. An MA956 specimen was tested at 600, 700 and 800 °C under 12 loading conditions spanning 500–2000 µε, with four manually triggered snapshots acquired per condition. The field path used affine registration, pixelwise four-snapshot median fusion and strain radius sensitivity analysis. The scalar path used only the original unfused images and compared four prespecified models under leave-one-temperature-out internal assessment. Increasing the strain radius from 10 to 15 grid points reduced the spatial standard deviation and P95-to-median span by 22.7% and 16.6%, respectively, with little change in scalar RMSE. Controlled simulations with known ground truth showed that local non-affine disturbances primarily increased field dispersion, whereas isotropic scale changes induced approximately common-mode axial and transverse strain drift. Across 20 mechanical strain–magnification conditions, a common-mode-suppressed axial–transverse feature reduced RMSE from 439.7 to 49.8 µε. Guided by this mechanistic evidence, combining the feature with temperature terms yielded the lowest pooled out-of-fold RMSE of 76.27 µε in the experimental data, 69.7% below the raw axial strain baseline. Within the conditions examined, these results demonstrate that improved strain field visualization is insufficient evidence of scalar accuracy and support separate validation of field stability and condition-level incremental strain error.