Abstract:In this paper, we presented a method for retrieving sea surface wind speed (SSWS) from Sentinel-1 synthetic aperture radar (SAR) horizontal-horizontal (HH) polarization data in extra-wide (EW) swath mode, which have been extensively acquired over the Arctic for polar monitoring. In contrast to the conventional algorithm, i.e., using a geophysical model function (GMF) to retrieve SSWS by spaceborne SAR, we introduced an alternative retrieval method based on a GMF-guided neural network. The SAR normalized radar cross section, incidence angle, and wind direction are used as the inputs of a back propagation (BP) neural network, and the output is the SSWS. The network is developed based on 11,431 HH-polarized EW images acquired in the marginal ice zone (MIZ) of the Arctic from 2015 to 2018 and their collocated scatterometer wind measurements. Verification of the neural network based on the testing dataset yields a bias of 0.23 m/s and a root mean square error (RMSE) of 1.25 m/s compared to the scatterometer wind data for wind speeds less than approximately 30 m/s. Further comparison of the SAR retrieved SSWS with independent buoy measurements shows a bias and RMSE of 0.12 m/s and 1.42 m/s, respectively. We also analyzed the uncertainty of the retrieval when reanalysis model wind direction data are used as inputs to the neural network. By combining the detected sea ice cover information based on SAR data, sea ice and marine-meteorological parameters can be derived simultaneously by spaceborne SAR at a high spatial resolution in the Arctic.

What problem does this paper attempt to address?

### Problems Addressed by the Paper This paper aims to extract Sea Surface Wind Speed (SSWS) from Sentinel-1 Synthetic Aperture Radar (SAR) data in the Marginal Ice Zone (MIZ) of the Arctic. Unlike traditional methods that use Geophysical Model Function (GMF) to extract SSWS from spaceborne SAR data, this paper proposes a new method based on a GMF-guided neural network. Specifically, the paper addresses the following key issues: 1. **Limitations of Traditional Methods**: - Traditional GMF methods have certain limitations when dealing with HH polarization SAR data, as HH polarization data is more sensitive to the sea surface but has a more complex dependency. - Traditional methods require extensive empirical or statistical adjustments, and the function forms are complex, containing multiple sub-functions and more than 30 coefficients. 2. **High Spatial Resolution SSWS Extraction**: - In the Arctic region, especially in the Marginal Ice Zone, high spatial resolution extraction of sea ice coverage and ocean dynamic parameters is of great significance for scientific research, resource utilization, and navigation safety. - The proposed method can combine sea ice information in SAR data to simultaneously extract SSWS and other ocean meteorological parameters. 3. **Application of Machine Learning**: - Machine learning algorithms show great potential in mining information from ocean remote sensing data, especially in fitting nonlinear curves. - By using a large amount of S1 data, researchers can explore methods to extract ocean meteorological parameters from spaceborne SAR data based on machine learning algorithms. ### Method Overview - **Dataset**: - Used 11,431 HH polarization S1 EW mode images and their corresponding scatterometer wind speed measurement data obtained in the Marginal Ice Zone from 2015 to 2018. - Also used wind speed measurement data from NDBC buoys and the Chinese icebreaker "Xuelong" in the Arctic for validation. - **Neural Network Design**: - Designed a backpropagation (BP) neural network with an input layer including Normalized Radar Cross Section (NRCS), incidence angle, and wind direction, and an output layer for SSWS. - The network structure includes an input layer, an output layer, and three hidden layers with 6, 10, and 8 nodes, respectively. - **Training and Validation**: - Validated the performance of the trained BP neural network by comparing it with ASCAT wind speed data. - Further validated the accuracy of the extracted SSWS by comparing it with in-situ measurement data from NDBC buoys and "Xuelong". ### Results - **Validation Results**: - Compared to ASCAT wind speed data, the bias of the SSWS extracted by the neural network is 0.23 m/s, and the Root Mean Square Error (RMSE) is 1.25 m/s. - Compared to independent buoy measurement data, the bias is 0.12 m/s, and the RMSE is 1.42 m/s. - **Uncertainty Analysis**: - Analyzed the uncertainty when using reanalysis model wind direction data as input to the neural network. ### Conclusion The GMF-guided neural network method proposed in this paper shows good performance in extracting SSWS in the Marginal Ice Zone of the Arctic, with high accuracy and reliability. This method provides a new approach for high spatial resolution extraction of ocean meteorological parameters, supporting scientific research and applications in the Arctic region.

Retrieval of Sea Surface Wind Speed from Spaceborne SAR over the Arctic Marginal Ice Zone with a Neural Network

Low-to-Moderate Wind Speed Retrieval from Sentenel-L Dual-Polarized Sar Images.

New Method for High Wind Speed Retrieval Using Altimeters Observations

A New Wind Speed Retrieval Method for an Ocean Surface Using the Waveform Width of a Laser Altimeter

A New Approach for Ocean Surface Wind Speed Retrieval Using Sentinel-1 Dual-Polarized Imagery

Validation of a SAR-Only Wind-Vector Retrieval Against Shipborne In Situ Wind Observations in the European Arctic

A Neural Network Method for Retrieving Sea Surface Wind Speed for C-Band SAR

Developing a Subswath-Based Wind Speed Retrieval Model for Sentinel-1 VH-Polarized SAR Data over the Ocean Surface

Preliminary analysis of wind and wave retrieval from Chinese 1mC-SAR image

Retrieving Ocean Surface Winds and Waves from Augmented Dual-Polarization Sentinel-1 SAR Data Using Deep Convolutional Residual Networks

Retrieval of Snow Depth over Arctic Sea Ice Using a Deep Neural Network

A Novel Method for the Estimation of Sea Surface Wind Speed from SAR Imagery

On the Use of Azimuth Cutoff for Sea Surface Wind Speed Retrieval From SAR

Toward Daily Snow Depth Estimation on Arctic Sea Ice During the Whole Winter Season From Passive Microwave Radiometer Data

Assessment of Sea-Surface Wind Retrieval from C-Band Miniaturized SAR Imagery

DNN‐Based Retrieval of Arctic Sea Ice Concentration From GNSS‐R and Its Effects on the Synoptic‐Scale Forecasting as Supplementary Observation Source

A Wind Field Retrieval Method Based on the Expected NRCS

Influence of mesoscale eddies on wind retrieval from C-band synthetic aperture radar images

Pan-Arctic sea ice concentration from SAR and passive microwave

Extraction of Sea Ice Cover by Sentinel-1 SAR Based on Support Vector Machine With Unsupervised Generation of Training Data

Analysis of coastal wind speed retrieval from CYGNSS mission using artificial neural network