The satellite altimetry missions could measure high-accuracy sea surface heights (SSHs) that can be used to recover the marine gravity field. Traditional methods for estimating the marine gravity field from SSHs all rely on approximate physical correlations between SSHs and gravity, which may neglect nature’s complex nonlinearity. This work presents a new deep network-based method to recover the gravity anomaly. This new method uses a multi-channel convolutional neural network (MCCNN) architecture to capture the nolinear features between ship-borne gravity and a group of input parameters including deflections of the vertical (DOVs), submarine topography and the geo-locations. To validate the gravity, ship-borne gravity anomalies on the two independent cruises were not used in the deep learning process. For comparison, we also estimated the gravity using the traditional inverse Vening Meinesz (IVM) method. Our results indicate that the MCCNN method can derive high-quality marine gravity anomalies. The assessments using 1 mGal-accuracy ship-borne gravity anomalies show that the average accuracy for gravity from MCCNN method is higher than 3 mGal and this method achieves 0.05-0.50 mGal improvement over benchmark methods IVM. Assessed by marine gravity anomaly models with the accuracy of 1-2 mGal, the MCCNN method has shown to improve the accuracy of gravity by at least 4%. Comparisons with the IVM results show that improvements of the MCCNN method were mainly in wavelengths between 8 km and 100 km due to the using of bathymetry. The results show that our deep learning method maintains good performance and is promising for gravity recovering.

The digital elevation model (DEM) is known as one kind of the most significant fundamental geographical data models. The theory, method and application of DEM are hot research issues in geography, especially in geomorphology, hydrology, soil and other related fields. In this paper, we improve the efficient sub-pixel convolutional neural networks (ESPCN) and propose recursive sub-pixel convolutional neural networks (RSPCN) to generate higher-resolution DEMs (HRDEMs) from low-resolution DEMs (LRDEMs). Firstly, the structure of RSPCN is described in detail based on recursion theory. This paper explores the effects of different training datasets, with the self-adaptive learning rate Adam algorithm optimizing the model. Furthermore, the adding-“zero” boundary method is introduced into the RSPCN algorithm as a data preprocessing method, which improves the RSPCN method’s accuracy and convergence. Extensive experiments are conducted to train the method till optimality. Finally, comparisons are made with other traditional interpolation methods, such as bicubic, nearest-neighbor and bilinear methods. The results show that our method has obvious improvements in both accuracy and robustness and further illustrate the feasibility of deep learning methods in the DEM data processing area.

High-resolution DEMs can provide accurate geographic information and can be widely used in hydrological analysis, path planning, and urban design. As the main complementary means of producing high-resolution DEMs, the DEM super-resolution (SR) method based on deep learning has reached a bottleneck. The reason for this phenomenon is that the DEM super-resolution method based on deep learning lacks a part of the global information it requires. Specifically, the multilevel aggregation process of deep learning has difficulty sufficiently capturing the low-level features with dependencies, which leads to a lack of global relationships with high-level information. To address this problem, we propose a global-information-constrained deep learning network for DEM SR (GISR). Specifically, our proposed GISR method consists of a global information supplement module and a local feature generation module. The former uses the Kriging method to supplement global information, considering the spatial autocorrelation rule. The latter includes a residual module and the PixelShuffle module, which is used to restore the detailed features of the terrain. Compared with the bicubic, Kriging, SRCNN, SRResNet, and TfaSR methods, the experimental results of our method show a better ability to retain terrain features, and the generation effect is more consistent with the ground truth DEM. Meanwhile, compared with the deep learning method, the RMSE of our results is improved by 20.5% to 68.8%.

High-resolution digital elevation model (HR DEM) plays an important role in hydrological analysis, cartographic generalization, and national security. As the main high-precision DEM data supplementary method, DEM super-resolution (DEM SR) based on deep learning has been widely studied. However, its accuracy has fallen into a bottleneck at present, which is more prominent in complex regions. The reason for this issue is that existing methods are difficult to capture enough local features from the low-resolution (LR) input data, and part of the global information (Contour information of long-distance features such as rivers and ridges) will also be lost in the network transmission process. To resolve this issue, a novel feature-enhanced deep learning network (FEN) is designed in this paper. The proposed FEN includes a global feature SR (GFSR) module and a local feature SR (LFSR) module. The former provides global information by using an interpolation method (Kriging) including geographical laws (spatial autocorrelation); The latter fully captures the features in the input data by integrating powerful feature extraction modules and then provides sufficient local features for DEM SR tasks. Thus, DEM SR tasks for complex regions can be realized by integrating the results of GFSR and LFSR modules. Extensive experiments show that FEN achieves state-of-the-art performance in DEM SR tasks facing complex regions. Specifically, compared with the existing DEM SR method (TfaSR, SRResNet, Bicubic, SRCNN, and Kriging), the result by FEN is closer to HR DEM and can retain more local DEM features. Meanwhile, the FEN is more than 20% ahead of other DEM SR methods based on deep learning in elevation accuracy.

The underwater acoustic communication technique for high-speed and highly reliable information transmission in the ocean has been one of the popular research focuses facing the fast-growing information technology sector and the accelerating development of ocean resources. In order to achieve a high information transmission rate with limited underwater acoustic channel bandwidth, researchers have paid much attention to the underwater acoustic communication technique based on orthogonal frequency division multiplexing (OFDM). A traditional OFDM-based frequency domain equalization algorithm relies on cyclic prefixes for the effective resistance to the multipath effect of an underwater acoustic channel. However, a redundant cyclic prefix may lead to a severe waste of energy and bandwidth in the underwater acoustic system if it is too long. The high utilization rate of OFDM signal channel will not be practically achieved in this case. Based on the limitations of the existing frequency domain equalization algorithm, this paper studied the influence of the multipath effect on the OFDM signal transmission. Subsequently, the principles of the OFDM-based frequency domain equalization were further explored for an improved structural model design of the communication system. On this basis, a novel frequency domain adaptive equalization algorithm was put forward. In addition, the proposed algorithm was optimized to address the problem of increased computation. The simulation results proved that the novel frequency domain equalization algorithm delivers a better symbol error ratio than the existing algorithm, and the compensation for the multipath effect through frequency selective fading. The proposed algorithm can realize the information transmission at a low symbol error ratio when fewer cyclic prefixes are used, so that it takes up a lower number of channels with cyclic prefixes in the OFDM communication system.

To address the data storage, management, analysis, and mining of ship targets, the object-oriented method was employed to design the overall structure and functional modules of a ship trajectory data management and analysis system (STDMAS). This paper elaborates the detailed design and technical information of the system’s logical structure, module composition, physical deployment, and main functional modules such as database management, trajectory analysis, trajectory mining, and situation analysis. A ship identification method based on the motion features was put forward. With the method, ship trajectory was first partitioned into sub-trajectories in various behavioral patterns, and effective motion features were then extracted. Machine learning algorithms were utilized for training and testing to identify many types of ships. STDMAS implements such functions as database management, trajectory analysis, historical situation review, and ship identification and outlier detection based on trajectory classification. STDMAS can satisfy the practical needs for the data management, analysis, and mining of maritime targets because it is easy to apply, maintain, and expand.