基于STO_IEU2020的海底地形数据尺度提升方法比较

Comparison of several scaling methods for submarine terrain grid data

栅格小于1′的全球海底地形数据通常以卫星测高海洋重力数据构建的1′海底地形数据作为背景场模型,采用数据格网化方法得到。基于此,以STO_IEU2020模型作为海底地形背景场数据,使用反距离加权法、改进Shepard拟合方法、径向基函数法、Kriging方法4种常规网格化方法以及“移去-恢复”方法,分别构建了试验区域栅格大小为30″和15″海底地形模型,并对试验结果开展了分析评估。试验结果表明,采用改进Shepard拟合方法构建的30″和15″栅格海底地形模型相较其他3种常规海底地形格网化方法检核精度高。进一步基于改进Shepard网格化方法,采用“移去-恢复”方法构建的30″和15″栅格海底地形模型结果显示,“移去-恢复”方法可进一步提升建模精度,据此建议制作多尺度海底地形数据,可重点考虑基于“移去-恢复”的改进Shepard拟合方法。

The global seafloor topography data with a grid size smaller than 1'is usually constructed using 1'seafloor topography derived by satellite altimetry ocean gravity data as the background field,and is obtained using the grid method.Combining STO_IEU2020 background field,four conventional grid methods(inverse distance weighting method,improved Shepard fitting method,radial basis function method,Kriging method)and"remove-restore"method were used to construct seafloor topography with grid sizes of 30"and 15".The experimental results showed that the 30"and 15"grid seafloor topography models constructed using the improved Shepard fitting method have higher verification accuracy compared to the other three conventional grid methods.Next,a 30"and 15"grid seafloor topography was constructed,by the"remove-restore"method(improved Shepard grid method).The results showed that the"remove-restore"method can further improve modeling accuracy.Further,it is recommended to focus on the improved Shepard fitting method based on"remove-restore"when producing multi-scale seafloor topography data.