顾及海底地形坡度的南海海底地形分区GGM反演

Considering the gradient of ocean topography to predict bathymetry with the gravity-geology method in the South China Sea

利用重力地质法(Gravity-Geologic Method,GGM)反演海底地形时,海水与海底洋壳的密度差异常数是影响反演精度的一项关键参数.由于海底地形复杂程度的影响,不同区域密度差异常数存在差异.针对于此,本文以南海局部海域(113°E—119°E,12°N—19°N)为实验区,依据区域内坡度及地形分布,将研究区域划分为6个子区域,分别求取每个子区域的最佳密度差异常数,以改善海底地形反演的精度.基于HY-2A测高数据获得的重力异常数据,反演了研究区域格网间隔1′×1′的SGGM海底地形模型.结果表明,将SGGM模型与检核点实测水深进行比较,其差值的标准差为101.46 m,相对精度为2.82%,优于GEBCO_2021模型、V19.1模型对比实测水深差值的标准差(131.50 m、129.81 m)与相对精度(3.64%、3.59%).结合各子区域不同的坡度及船测点分布,统计各子区域相对精度为2%~4%,并分析得出反演较差结果集中分布在地形突变区域,而坡度平缓区域的标准差可达到42.20 m.此外,同未进行区域划分而采用一个密度差异常数反演的结果相比,反演精度提高了9.68 m,说明了本文模型的构建精度得到了有效提升.

The density contrast between the seawater and the ocean bottom topographic mass is the key parameter that affects the accuracy when the Gravity-Geology Method(GGM)is used to predict bathymetry.Due to the complexity of geological structure,the density contrast in different regions is different.In response to this,the local area of the South China Sea(113°E—119°E,12°N—19°N)is selected as the experimental area,and the area is divided into six sub-areas according to the gradient and topographic distribution in this paper.The optimum density contrast of each sub-area is calculated separately to improve the accuracy of predicting bathymetry.Based on the ocean gravity anomaly data obtained from HY-2 A altimetry,the bathymetry model SGGM with 1′×1′is constructed.Comparing the SGGM model with the ship data at check points,the Standard Deviation(STD)of the difference is 101.46 m,and the relative accuracy is 2.82%,which is better than the GEBCO_2021 model(131.50 m,3.64%)and V19.1 model(129.81 m,3.59%).According to the different gradients and distribution of shipborne points in sub-areas,the relative accuracy of each sub-area is 2%~4%,and the analysis shows that the worse results are mainly distributed in the terrain mutation area and the STD can reach 42.20 m in the gentle gradient region.In addition,compared with the previous model without area division,the accuracy of the predicted model is improved by 9.68 m,the construction accuracy of the model has been effectively improved.