| Shallow sea bathymetry is one of the core contents of ocean basic surveying and mapping,and it is of great significance for offshore navigation,ocean research and ocean space management.In recent years,a range of sounding techniques have been used to measure and estimate water depth.Among them,shipborne sonar has the characteristics of high sounding accuracy,but there are safety risks such as hitting rocks and being stranded in shallow water areas,so it is inefficient and difficult to carry out measurement.In addition to high measurement accuracy,airborne lidar has the characteristics of flexibility and maneuverability,but it also faces the limitations of low efficiency and high cost,and is affected by factors such as weather,airspace and geographical environment,and the range that can be measured is still limited.Relatively speaking,satellite sensors have the ability to obtain periodic and extensive optical information in water areas,so many satellite remote sensing data have been widely used in shallow water bathymetry,such as hyperspectral and multispectral remote sensing images.After analyzing and processing these images,it is found that the use of hyperspectral remote sensing information can effectively improve the accuracy of bathymetry results.However,bathymetry technology based on hyperspectral and multispectral remote sensing images needs to rely on empirical formulas and bathymetric calibration points,and these empirical formulas have different coefficients in different waters,Moreover,it is impossible or difficult to obtain bathymetric calibration points in certain areas or periods of time,thereby limiting the accuracy of bathymetry.With the development of technology and the promotion of engineering requirements,photon counting lidar has been widely used in hydrology,meteorology and other fields,and has shown unique advantages in the field of bathymetry.However,since the sending and receiving signals of the photon counting lidar system are very weak,it is greatly affected by interference factors such as solar background noise,which brings great challenges to the processing and application research of photon counting lidar data.Moreover,with the increasing demand for modern shallow sea surveying and mapping,the requirements for data quality and accuracy are getting higher and higher.It is difficult for a single bathymetry technology and method to meet and solve various complex 3D terrain detection needs.By synergistically integrating active ATLAS bathymetry data and passive multispectral satellite imagery,their respective advantages can be fully utilized to achieve high-precision shallow sea depth detection in a wider range.Based on the above background,the thesis carried out the following aspects of work:(1)A photon data simulation model of spaceborne photon counting lidar in water body is established,and a variable-size elliptical spatial density filtering method with adaptive water depth is proposed.Based on the photon counting lidar bathymetry theory and the parameter configuration of the laser lidar equation coupling ICESat-2 satellite ATLAS payload,a detection model of the photon counting laser detection system is established to simulate the change of photon signal quantity in different environments.Aiming at the distribution of original photons acquired by photon-counting lidar,a variable-size elliptical spatial density filtering method adaptive to water depth is proposed,which can accurately extract signal photons from water surface and bottom.Based on the density characteristics and signal-to-noise ratio characteristics of photon signals,the reliability evaluation of photon signals in water bodies is established.The accuracy and effectiveness of the filtering method in this thesis are verified by comparison with other denoising algorithm results and confidence evaluation.(2)A photon refraction correction method for bottom signal based on JONSWAP ocean wave spectrum is proposed.The laser will be refracted when it passes through the sea level.Therefore,in order to obtain the real coordinates after refraction,it is necessary to correct the refraction and coordinates of the underwater photons.In this thesis,a highprecision JONSWAP wave spectrum modeling method is used to construct an instantaneous sea surface profile and an accurate underwater photon coordinate correction model,and realize a photon coordinate refraction correction model in different directions,thereby obtaining high-precision water depth profile data in the study area.By comparing the sounding results of this method with other refraction correction methods and the measured data of airborne lidar,the sounding accuracy of this method is verified.(3)A method for joint water depth retrieval of spaceborne photon bathymetry data and multispectral satellite images is proposed.Aiming at the characteristics of photon bathymetry data,a water depth inversion method for the fusion of photon bathymetry data and multispectral satellite images is proposed.The method first establishes the mapping relationship between the reflectance value of the optical image and the water depth value extracted from the photon data.Then two semi-empirical and semi-theoretical models(multiband linear regression and dual-band ratio regression models)and two machine learning methods(support vector regression and random forest)were used to perform water depth retrieval on multispectral satellite images.Finally,the inversion accuracy of each model is verified and analyzed by airborne lidar data.Aiming at the difficulties faced by spaceborne photon counting lidar shallow water photon point cloud data processing and multispectral satellite image water depth inversion,this thesis proposes more effective point cloud filtering methods,refraction correction methods and active and passive data fusion for water depth inversion method.The above-mentioned photon data processing and water depth inversion method well combine the advantages of active and passive remote sensing bathymetry,laying the foundation for high-precision and efficient island and reef water depth and terrain detection of all-satellite links. |
