Research On The Method Of Water Depth Inversion Treatment Around The Oceanic Reefs In The South China Sea

The sea area around the Oceanic Reef of the South China Sea is rich in natural resources,and the strategic position of the national defense military is very important.The water depth remote sensing technology can realize the macroscopic dynamic observation of shallow seawater in a cost-effective manner,and obtain the latest water depth information,which is important for some sea areas that are difficult to reach.Value provides the basic data source for resource development and utilization and the construction of a maritime power.The use of remote sensing technology to obtain seabed topography has the advantages of high precision and high efficiency.However,in the actual detection process,there are problems such as low inversion precision,poor stability of the inversion model,and excessive error of 0-5m shallow water depth.Construction of high-precision seabed topography.According to the research target problem,the water depth inversion model and the multi-core SVR inversion model based on depth adaptation are constructed respectively.The two new methods and water depth inversion results have important theoretical significance and practical application value for terrain detection near the island reef.The intended purpose.The main research contents and conclusions are as follows:1)The applicability of the inversion of the planet image is studied.The band blue and green,red and near-infrared bands are used to construct the band combination model and the band ratio model respectively,and the landsat8 medium resolution image is used as the contrast experiment.The inversion of the blue and green bands of the image is better.The error is always controlled within 3m during the experiment,and the inversion accuracy is high.It is determined that the planetary multispectral satellite image is suitable for the data source of remote sensing water depth inversion.2)Exploring a depth adaptive based inversion method to solve the problem of poor stability of the model.The measured water depth points are divided into small units according to different depths,and the basic model of each research sub-area unit is fitted.Finally,a unified depth-adaptive remote sensing inversion model of the research area is constructed,which overcomes the shortcomings of the same set of model parameters of the global model.The partition model is based on the spatial characteristics of local regions to establish a number of models of depth adaptability,and these sub-regions are independent of each other,so the established series of models will be less affected by the sediment and water quality.The depth-adaptive model has a root mean square error of 1.025m and an average absolute error of 0.993m,which is 0.936m and 0.884m higher than the traditional global inversion model,and the average relative error is 13.28%,which is more than the traditional global inversion.The model is improved by 6.3%,and the average absolute error is controlled within 1.2m in each water depth segment,which effectively solves the problem of poor stability of the traditional global model.3)The study compared four single-core SVR(Support Vector Regression)regression models and a multi-core SVR regression model.The multi-core SVR achieved good inversion results in the study area,with an average relative error of 8.14%,which was 5.28%higher than the traditional model.Compared with the four single-core SVR models LINEAR core,POLY core,RBF core and SIGMOID core,respectively.3.08%,3.03%,1.81%,4.48%,the average absolute error is 1.022m,which is 0.855m higher than the traditional model,which is 0.347m,0.339m,0.185m,.0519m higher than the four single-core SVR models.For the analysis results:whether the whole or the partition is inversion research,the multi-core SVR model can greatly improve the accuracy of water depth inversion,and can control the error within a certain range in the shallow water area(0-5m).The stability and robustness are very good.Compared with the traditional methods,the research results of this paper have significantly improved the accuracy of remote sensing water depth inversion,both in terms of overall accuracy and inter-partition precision,and realized rapid and large-area measurement of the research sea area,enriching the remote sensing sounding.Means have important theoretical and practical application values.