| In September 2018,NASA launched the first spaceborne photon-counting lidar(i.e.the Advanced Topographic Laser Altimeter System,ATLAS).The ATLAS obtain the subaqueous signal photons in clear waters during the periods of operation,which indicates that the spaceborne photon-counting lidar has a great potential on conducting the satellite derived bathymetry.Compared with the airborne and shipborne bathymetric instruments,the spaceborne photon-counting lidar are not restricted by the working geographical locations especially for some unsuitable and inaccessible water areas.This new technique is very helpful to address the requirement of shallow water bathymetry for our country,and is of a great significance to the assure national maritime rights and national defense constructions.While spaceborne photon-counting lidars introduce promising prospects for bathymetry,many issues still need to be solved,e.g.,the bathymetry capability,bathymetry accuracy,and whether the seamless bathymetry can be achieved.The bathymetry capability of spaceborne photon-counting lidar depends not only on whether the water surface and bottom signal photons can be detected,but also on whether they can be extracted and identified.The photon density of the water surface and bottom is generally related to the system parameters and environmental parameters including inherent optical properties of water.It is necessary to establish a robust theoretical model to quantitatively analyze the signal and noise level of spaceborne photoncounting lidars.Different from the conventional linear lidars recording full-waveforms,photoncounting lidars can only detect the presence of photons,but cannot record the intensity,which means that it cannot directly distinguish the categories of the received photons.As a result,a specific extraction algorithm is needed to detect signal photons from the raw noisy data photons.The signal photon elevation difference between the water surface and water bottom is influenced by the undulation and refraction of the water surface.It is necessary to correct the bathymetry error to improve the bathymetric accuracy for spaceborne photon-counting lidars.At present,spaceborne photon-counting lidars already have the ability of multi-beam detection.However,it can only obtain the along-track water depth profile rather than the water depth of the whole shallow water area due to the large spacing between adjacent beams.Hence,to integrate the active and passive water depth inversion research(using bathymetric lidar points and multispectral imagery)is essential to overcome the drawback of the insufficient coverage of photon-counting lidars in the cross-track direction.The main content of this thesis includes the following aspects:1.Theoretical signal and noise model of spaceborne photon-counting lidars is proposed based on the basic lidar theory,the response model of single-photon detector,the distribution model water of surface profile and slope,and scattering and attenuation model of light in water column.The photon point simulation software of spaceborne photon counting lidars is developed,by which the influence of single-photon detector and environmental factors on the distribution of signal and noise photons is quantitatively analyzed.A maximum bathymetric depth model to simulate satellite photon-counting lidar performance is established based on the signal-to-noise ratio of received data.The estimated bathymetric depths from the model are verified by the actual maximum bathymetric depths of ATLAS at six typical regions,and deviations are less than 15%,which proves the correctness of the proposed model.Based on the maximum bathymetric depth model,the influence of environmental parameters and system parameters is analyzed in detail,which is beneficial to optimize the system parameters when designing a new photon-counting lidar.2.The subaqueous signal photons generally have a low density near coasts and islands,which suffer from the noise photons of the background radiance and scattering effect.An adaptive signal extraction algorithm is improved to process the low signal-to-noise data.First,through the noise point density analysis,the classification threshold can be automatically adjusted and the signal points are roughly extracted.Then,the signal points from the bottom and water surface are separately identified.The remaining scattering points near the bottom signal points are further filtered through the spline fitting.By adding noise points with different noise rates to the points obtained by the ATLAS,the extraction result of bottom signal points is quantitatively analyzed.The results indicate that the signal point extraction method proposed in this thesis performs better than the classical density-based spatial clustering.3.Based on the ocean wave spectrum and microfacet model,a simulation model of the sea surface signal for photon-counting lidars is established,and the random noise distribution on the sea surface is analyzed under different wind speed conditions.By fitting the slope curve of water surface signal point clouds,the refraction error of extracted water bottom points is corrected.The original laser points obtained by the ATLAS and airborne bathymetric lidar in shallow water areas of St.Thomas and St.Croix are used,and the geolocation accuracy of the corrected bottom signal points of ICESat-2 is evaluated.Near St.Thomas,the mean absolute error(MAE)of bathymetric signal photons is about 0.39 m,and the root mean square error(RMSE)is about 0.54 m.Near St.Croix area,the MAE of the photon signal is about 0.68 m,and the RMSE is about 1.01 m.4.To overcome the shortcoming of insufficient coverage in the cross-track direction for spaceborne photon counting lidars,a study on integrating the active and passive data was carried out using spaceborne bathymetric lidar points and multispectral images.During the water depth inversion process,the bottom signal points are matched with the pixels of multispectral images via the geographical location(i.e.the latitude and longitude).The matched laser reference points were further filtered by the cross-validation.Using the ATLAS lidar reference points and the Sentinel-2 multispectral images of a 10 m resolution at St.Croix and Riyue Bay,the shallow water depth inversion in two regions is proposed based on three classical water depth inversion models.According to the comparison between the inverted water depth data and the in-situ airborne bathymetry data,it is shown that through the selection of reference points,the water depth inversion accuracy of three water depth inversion models is improved,which solves the contradiction between the quality and quantity of reference water depths from spaceborne photon-counting lidar data. |
PDF Full Text Request