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

Yan Wang, Yan Li, Yanshuang Xie, Guomei Wei, Zhigang He, Xupu Geng, Shaoping Shang2023 · DOI

TL;DR

This paper assesses the performance of two new miniaturized C-band SAR satellites, HiSea-1 and Chaohu-1, for retrieving sea-surface wind fields (SSWFs). The study demonstrates that these cost-effective satellites can provide accurate wind speed and direction data, comparable to larger, more traditional SAR systems, validating their design and potential for broader oceanographic applications.

What Problem Does This Paper Address?

Traditional Synthetic Aperture Radar (SAR) systems, while powerful for observing the Earth's surface, are often large, expensive to build and launch, and consequently, their data can be costly. This limits their widespread application in areas like ocean monitoring. There's a growing trend towards developing smaller, more affordable SAR systems. However, the performance of these miniaturized SARs for specific applications, such as retrieving sea-surface wind fields (SSWFs), has not been thoroughly evaluated. This paper aims to fill that gap by assessing the SSWF retrieval capabilities of China's first commercial miniaturized C-band SAR satellites, HiSea-1 and Chaohu-1.

Methods

The study employs a three-step variational method for SSWF retrieval, leveraging the CMOD5.N geophysical model function (GMF) and CCMP wind reanalysis data.

  1. Wind Cell Partitioning: SAR imagery is divided into 1 km x 1 km "wind cells." CCMP wind data is interpolated to match this resolution.
  2. Homogeneity Check: Each wind cell's normalized radar cross-section (NRCS) distribution is checked for homogeneity using normalized variance. Cells with excessive variance (A ≥ 1.8) are excluded to ensure reliable NRCS values.
  3. Variational Retrieval: A cost function is minimized to determine the optimal wind speed and direction. This function balances the difference between observed SAR NRCS and simulated NRCS from CMOD5.N, along with the difference between retrieved wind components and those from CCMP reanalysis data.

A quality control procedure, based on an empirical elliptical region derived from Sentinel-1 data, is also applied to filter out unreliable retrievals from HiSea-1 and Chaohu-1 images.

Datasets

  • HiSea-1 and Chaohu-1 SAR Images: 1053 images (378 from HiSea-1, 675 from Chaohu-1) acquired in stripmap mode (Level-2 ORG product) covering the Southeast China Sea. Each image has a spatial resolution of 3m and covers approximately 25 km x 25 km. Data was collected between March and December 2022.
  • Buoy Winds: Data from 16 stations were used for validation. A conversion formula was applied to standardize 3m wind speed measurements to 10m.
  • ERA5 Winds: Global reanalysis data with a spatial resolution of 0.25°x 0.25° and a temporal resolution of 1 hour, used for comparison.
  • CCMP Winds: Cross-Calibrated Multi-Platform (CCMP) wind vectors with a spatial resolution of 0.25°x 0.25° and a temporal resolution of 6 hours, used as input for the retrieval method.
  • ASCAT Winds: Data from the Advanced Scatterometer (ASCAT) were also used for assessment.
  • Sentinel-1 SAR Images: 720 Ground Range Detected (GRD) images were used to establish the quality control region.

Key Insights and Ideas

  • Miniaturization Success: HiSea-1 and Chaohu-1, despite their significantly reduced size and weight compared to traditional SARs like Gaofen-3, demonstrate comparable imaging performance. This highlights the feasibility of developing high-performance, cost-effective miniaturized SAR systems.
  • Variational Method Robustness: The variational method, by incorporating external wind data (CCMP) and a GMF, proves effective in retrieving SSWFs from the new miniaturized SAR data.
  • Quality Control is Crucial: The study emphasizes the importance of quality control, particularly for new sensors. The use of Sentinel-1 data to define an empirical quality-controlled region for HiSea-1 and Chaohu-1 data is a practical approach to ensure data reliability.
  • Wind Field Dynamics: Visualizations of the retrieved wind fields reveal clear patterns, including the presence of frontal surfaces and significant wind speed gradients, showcasing the ability of these SARs to capture mesoscale oceanographic features.

Key Results

After quality control, the retrieved SSWFs from HiSea-1 and Chaohu-1 were compared against reference datasets:

  • vs. ERA5:
    • Wind Speed: RMSE of 2.42 m/s, MBE of -0.44 m/s.
  • Wind Direction: RMSE of 11.5°, MBE of -1.3°.
  • vs. Buoy Winds:
    • Wind Speed: RMSE of 1.64 m/s, MBE of 1.08 m/s.
  • Wind Direction: RMSE of 36.8°, MBE of 2.4°.
  • vs. ASCAT Winds:
    • Wind Speed: RMSE of 3.29 m/s, MBE of -1.65 m/s.
  • Wind Direction: RMSE of 41.7°, MBE of -8.8°.

The results indicate that the errors in the retrieved wind vectors are acceptable, validating the potential and practicality of HiSea-1 and Chaohu-1 for SSWF retrieval.

Limitations and Future Work

  • Limited Buoy Data: The study notes a limitation due to the absence of buoy data for December 2022, which restricted the number of direct comparisons with buoy measurements.
  • Swath Limitations: The limited swath of HiSea-1 and Chaohu-1 SAR images can affect the spatial coverage and the number of coincident data points for validation, especially with buoy data.
  • Future Work: The paper implicitly suggests that further validation with more extensive datasets and potentially across different oceanographic conditions would be beneficial. Expanding the application of these miniaturized SARs to other ocean parameters beyond wind fields is also a natural progression.

Summary updated

BibTeX
@misc{wang2023assessment,
  author = {Yan Wang and Yan Li and Yanshuang Xie and Guomei Wei and Zhigang He and Xupu Geng and Shaoping Shang},
  title = {{Assessment of Sea-Surface Wind Retrieval from C-Band Miniaturized SAR Imagery}},
  year = {2023},
  doi = {10.3390/s23146313},
}

Loading related content...

Figure