Research On Shallow Water Depth Extraction Of The GF-Satellite Remote Sensing

Bathymetry is an important work for marine development,offshore engineering and military deployment.There are many navigational hazards,such as reefs and shoals in Shallow sea.So shallow sea bathymetry has a great significance to ensure the safety of ship navigation.The common bathymetric method is using the bathymetric equipment and positioning equipment that has already been installed on the surveying vessel to measure the water depth directly.It is necessary for the surveying vessel to sail on the surveying sea area according to the planned surveying route.For the more dangerous sea areas;such as areas that contain reefs or other hazards such as areas that are contested by international powers often make it difficult to complete the bathymetric survey.With the progress and development of ocean satellite technology in China,the ability of using remote sensing to observe ocean information has been enhanced.The extraction and quantitative inversion of ocean information by using remote sensing data has gradually become a new research topic.The development of depth inversion technology based on satellite remote sensing information platform and its application in the shallow water areas that there are dangers or disputes.It has become a new method of depth measurement,and this is a great significance for shipping safety,ocean development,military deployment and so on.In this paper,the Meiji reef area in the South China Sea is taken as the research subject.Based on the radiation calibration,atmospheric correction,geometric correction and land-water separation of remote sensing images,the shallow water depth inversion method based on GF-2 multispectral data corresponding to the area is studied.Firstly,analyze the spectral characteristics of shallow water,and the radiation energy is quantitatively described from the satellite sensor through the atmosphere to the sea surface.Then,the process of energy transfer from the sea surface to the sea floor is analyzed.The relationship between radiation energy and water depth is analyzed,and the feasibility of using radiation energy to invert water depth is explored.Secondly,the traditional depth inversion methods are introduced,including statistical correlation model and semi-theoretical semi-empirical models.The statistical correlation model is established by analyzing the correlation between the spectral band value of remote sensing imagery and the measured water depth value,and the calculation is relatively simple.However,the correlation between the spectral band and the measured water depth can not be guaranteed,so the inversion of the statistical correlation model is not accurate.The semi-theoretical and semi-empirical models are based on the attenuation characteristics of radiation energy in selected bodies of water and combines theoretical analytical modelling with empirical parameters to establish inversion depth modelling which can obtain higher inversion accuracy statistics.In this paper,single-band,multi-band and band-to-band ratio models are established respectively by using pre-processed GF-2 multispectral data and water depth samples.Finally,for quantitative depth inversion,the permeable band ratio and normalized principal component characteristic index are proposed as depth inversion factors which can effectively remove the effects of impurity suspended matter,chlorophyll and seabed geology.In this paper,a neural network water depth inversion method based on water feature parameters is proposed.This paper introduces the realization principle,model structure and training process of BP neural networking,with emphasis on the realization process of neural network water depth inversion methods.The simulation experiment of the neural network water depth inversion model is realized by using GF-2 multi-spectral data and water depth samples.The results show that the proposed method is more accurate than the simple neural network method.Compared with the traditional water depth measurement method,this method can realize the large-scale and rapid measurement of pre-designated shallow water areas,and has good practicability.It can solve the problem of whether there is water depth information in specific water areas,and enrich the methods of bathymetric survey.