Details

Autor(en) / Beteiligte

• Sun, Weifeng
• Jia, Chen
• Fan, Chenqing
• Li, Wen
• Dai, Yongshou
• Huang, Weimin

Titel

Maximum a Posteriori Based Ocean Surface Current Inversion for Doppler Scatterometer

Ist Teil von

• IEEE journal of selected topics in applied earth observations and remote sensing, 2024, Vol.17, p.2067-2076

Ort / Verlag

Piscataway: IEEE

Erscheinungsjahr

2024

Quelle

Alma/SFX Local Collection

Beschreibungen/Notizen

• Doppler Scatterometer (DopScat) is a new tool for sea surface wind and current fields remote sensing with rapid global coverage and wide observation swath. Existing current inversion methods for DopScat are based on maximum likelihood estimation (MLE), and its inversion accuracy cannot meet the requirements of many offshore operations. To improve the accuracy of ocean surface current measurement for DopScat, a method using prior probability distributions extracted from historical current fields is presented. First, according to the temporal correlation of ocean surface current, the minimum root mean square differences of current speed and direction are used to select the historical ocean current data correlated to those of the observation area. Next, a fitting method based on maximum likelihood is employed to fit the selected current speed and direction data to determine their prior probability distributions. Then, the obtained distributions are used to construct the cost functions of the proposed maximum a posteriori (MAP)-based current inversion method. Finally, taking the current result provided by the MLE-based current inversion method as initial guess, the cost functions of the proposed MAP-based method are optimized to obtain the final current field. Validation experiments were conducted using simulated DopScat data based on the current generated by the ocean surface current analyses real-time model, the results show that the biases of the estimated current speed and direction are better than 0.05 m/s and 15<inline-formula><tex-math notation="LaTeX">^{\circ }</tex-math></inline-formula>, respectively. Compared with that of the MLE-based method, the biases are reduced by 0.16 m/s and 9<inline-formula><tex-math notation="LaTeX">^{\circ }</tex-math></inline-formula>, respectively.