以迴歸模式計算南海海深：融合衛星測高及光學影像結果

Abstract

本研究使用1990年至2000年之Geosat/GM與ERS-1/GM測高資料，以及最新的Jason-1/GM與Cryosat-2測高資料，結合四種衛星測高資料計算研究區重力異常模型，並推估南海水深模型。利用次波形門檻值演算法改善測距精度不佳的衛星測高資料。以Inverse Vening Meinesz (IVM)與Least Squares Collocation (LSC)兩種方法計算重力異常模型，並分析其精度。計算南海水深模型方面，利用南海區域已知的重力模型與水深模型，以迴歸模式(Regression Model)估計南海水深模型，並比較重力地質密度法計算之水深模型精度。重力成果方面，船測重力作為檢核資料，利用次波形門檻值演算法可以提升30%的精度，加入Jason-1/GM與Cryosat-2資料可以提升4%的精度；水深模型成果方面，利用大陸礁層水深值作為檢核資料，將Sandwell V16.1水深模型與高精度重力異常模型作為先驗知識，利用迴歸模式計算得水深模型，比利用重力地質密度法計算之水深模型精度改善34 m，結合衛星光學影像方面，於研究區選定28個島礁，經過光學島礁零公尺海岸線修正，每個島礁海岸線誤差改善率平均為75%。

In this study, satellite altimeter data from missions Geosat/GM and ERS-1/GM in the 1990s and 2000s, and from the latest missions Jason-1/GM and Cryosat-2, are used to compute gravity anomaly models and then to construct bathymetry models in the South China Sea (SCS). Sub-waveform threshold retracking is used to improve altimeter range accuracy. The Inverse Vening Meinesz (IVM) and Least Squares Collocation (LSC) are employed to compute gravity anomalies from retracked altimeter data. The regression model, based on a priori knowledge of gravity and depth in the SCS, is used to estimate depths from altimeter-derived gravity, which are compared with that from the gravity-geological method (GGM). Comparisons of altimeter-derived gravity anomalies with shipborne gravity anomalies show that, the gravity precision is increased by 30% from altimeter data that are improved by sub-waveform retracking and increased by 4% from using Jason-1/GM and Cryosat-2 altimeter data. The regression model outperforms the GGM, based on assessments using shipborne depths. We fuse depths from altimetry and optical images over atolls. On average, the fusion with optical images improves the definitions of coastlines over atolls by compared to the altimetry-only depths.