The Advanced Topographic Laser Altimeter System (ATLAS) on ICESat-2 (Ice, Cloud, and Land Elevation Satellite-2) uses a 532 nm band photon-counting LiDAR (Light Detection and Ranging), which has certain penetrability to water bodies, and the measured data show that the bathymetric ability reaches nearly one Secchi depth. ATLAS has a limited number of beams and a fixed ground track, and only collects section elevation along the track direction. The fusion of active laser point cloud and passive optical remote sensing satellite image can fill the gap of shallow water depth data in a large range. This paper takes ATLAS as the research object, with the aim of exploring the effective algorithm flow of spaceborne active laser and passive optical fusion processing, and systematically evaluating the bathymetric accuracy of the fusion algorithm. An adaptive Gauss filtering technology based on the density of point cloud was firstly improved to achieve accurate denoising under the condition of uneven surface/underwater density. Subsequently, the depth of underwater points was calculated automatically through steps of water surface modelling, refraction correction, etc. Finally, the control points and check points were randomly extracted to solve the parameters of multispectral inversion model and verify the internal accuracy of the model. In this paper, the accuracy of ATLAS bathymetry was verified by Airborne LiDAR Bathymetry (ALB) data in Oahu Island, Hawaii, and the results indicate that the vertical root mean square error (RMSE) ranges from 0.56 m to 1.11 m. In Yongle islands and Qilianyu area of the South China Sea, WorldView-2 (WV2) 4 bands multispectral images and ATLAS data were used to carry out the active-passive fusion bathymetry, and the ALB and sonar data were used to evaluate the accuracy. Experimental results show that the internal compliance accuracy of the fusion model is better than 1.25 m (RMSE), and the real bathymetry accuracy is better than 1.42 m (RMSE). The above results reveal the great potential of active-passive fusion bathymetry based on ICESat-2 and other high-resolution remote sensing satellites, which can provide strong technical support for filling the blank of shallow water depth information.

Two-media photogrammetry appears to be a useful technology for mapping shallow water depth. The objective of this paper is to explore the feasibility of shallow water bathymetry using satellite two-media photogrammetry. This paper presents a two-media photogrammetry technique for WorldView-2 stereo multispectral imagery. In this method, near-infrared band is used for sun glint elimination. A combination of Scale Invariant Feature Transform and Semi-Global Matching algorithm is adopted for dense stereo matching. Rational Function Model is applied for raw Digital Elevation Model generation and the water surface elevation is determined by interpolating water edge elevation. Finally, the approximate refraction correction model is adopted in the water region to correct the vertical coordinate offsets. Our experimental results of Ganquan Island and Zhaoshu Island show that a reasonable depth estimate with relative error less than 17% and 14% can be obtained for the water depth range between 5 and 20 m, and the root-mean-square error are 2.09 m and 1.76 m, respectively. However, less accurate estimate is found for the very shallow water depth range between 0 and 5 m. The overall accuracy of satellite two-media photogrammetry is at the same level with the traditional multispectral-based model for bottom depth retrieval.

The US ICESat-2 (Ice, Cloud, and Land Elevation Satellite-2) satellite adopts 532 nm single-photon lidar with shallow sea bathymetry capability. In order to realize high-precision and automated shallow sea bathymetric mapping based on ICESat-2 photon data, an adaptive underwater point denoising algorithm that considers the search direction and search size is proposed in this paper, and a detailed data processing process is discussed to further verify the technical feasibility. The accuracy of direct bathymetry and active-passive fusion bathymetry from ICESat-2 is systematically analyzed using airborne in-situ data. First, the underwater photon points are separated by surface position identification; then, the signal point cloud search strategy is improved, the search size increases with water depth, the search angle is rotated and the direction of the maximum number of point clouds is taken as the main direction, and the threshold is automatically determined by histogram Gaussian fitting of the point cloud density to achieve automatic underwater signal extraction; then the refraction correction is carried out based on the light geometry and the water depth is obtained; finally, the active-passive fusion bathymetry is performed by combining the optical remote sensing images WorldView-2 and Sentinel-2, and the accuracy is verified by using the airborne lidar bathymetric data provide by NOAA. The experimental results show that the proposed denoising algorithm can accurately discriminate the underwater signal/noise, and the overall accuracy is better than 86%; the root mean square error (RMSE) of ICESat-2 direct bathymetry is between 0.42-0.98 m; the RMSE of active-passive fusion bathymetry is between 0.84-1.88 m. Our workflow and experimental results demonstrate a means of using ICESat-2 to produce relatively accurate bathymetric maps in shallow, clear water environments.

Optical satellite remote sensing (RS) is a time- and cost-effective approach for shallow-water bathymetry over large areas. The photogrammetric and physics-based methods, which do not require in situ depth calibration data, are two implementations of RS-based bathymetry. This article compares these two bathymetric methods by using the same WorldView-2 imagery product of a study area in the Xisha Islands of the South China Sea. The focus is to investigate the difference between the resulting depths. The resulted accuracy of the photogrammetric method was slightly higher than that of the physics-based method based on the backward image, but both of them were significantly higher than that of the physics-based method based on the forward image. In the calm clear waters, the resulted depth of the photogrammetric method was closer to the ground truth than that of the physics-based method based on the backward image. In the shoal and adjacent waters where the water was turbid or larger waves were present, the photogrammetric method produced a highly noisy digital depth model (DDM), whereas the physics-based method derived a low-noise DDM from the backward image. In conclusion, the photogrammetric method outperforms the physics-based method in calm clear waters, but the physics-based method performs better than the photogrammetric method in shoal and adjacent waters that has turbid water or waves larger than the image spatial resolution. The photogrammetric DDM is noisier than the physics-based DDM. The accuracy of the physics-based method is also influenced by the non-optimal sun-target-sensor geometry of the images used.

The Ice, Cloud and Land Elevation Satellite (ICESat-2) offers an unprecedented high density of laser data on the Earth’s surface. Evaluation and calibration of geolocation errors are important for the scientific application of laser data. In this study, a crossover method is proposed to evaluate and calibrate errors. This method mainly relies on small-scale reference data and consistency conditions at the trajectory crossovers. By matching the profile measurements of the ascending and descending altimeter tracks to find the location where the crossover discrepancy is minimized, the geolocation errors of the wide-area laser data are estimated. Three years of ICESat-2 measurements over the Ridgecrest (CA19) in the southern United States and McMurdo Dry Valleys (MDV) in eastern Antarctica were used to test the performance of the method. It was shown that the method can be used to estimate the geolocation error of the ICESat-2 laser data, which ranged from 4.65 m to 5.62 m for both sites. After calibration, the mean absolute error of the crossover discrepancies was reduced by approximately 50%. In addition, with the reference digital elevation model (DEM) as the true value, the elevation accuracy (RMSE) of the laser data at the CA19 site improved from 0.38 m before calibration to 0.34 m after calibration, and the improvement at the MDV site was even more significant, about 0.22m. In conclusion, the proposed method is effective and feasible and can improve the consistency of long-period laser data quality in large areas.