Short-term prediction of wave height is paramount in oceanic operation-related activities. Statistical models have advantages in short-term wave prediction as complex physical process is substantially simplified. However, previous statistical models have no consideration in selection of predictive variables and dealing with prediction uncertainty. This paper develops a machine learning model by combining the dynamic Bayesian network (DBN) with the information flow (IF) designated as DBN-IF. IF is focused on selecting the best predictive variables for DBN by causal analysis instead of correlation analysis. DBN for probabilistic prediction is constructed by structure learning and parameter learning with data mining. Based on causal theory, graph theory, and probability theory, the proposed DBN-IF model could deal with the uncertainty and shows great performance in significant wave height prediction compared with the artificial neural network (ANN), random forest (RF) and support vector machine (SVM) for all lead times. The interpretable DBN-IF is proven as a promising tool for nonlinear and uncertain wave height prediction.

The reconstruction and monitoring of visibility over marine environments is critically important because of a lack of observations. To travel safely in marine environments, a high quality of visibility data is needed to evaluate navigation risk. Currently, although visibility is available through numerical weather prediction models as well as ground and spaceborne remote sensing platforms and ship measurements, issues still exist over the remote marine environments and northern latitudes. To improve visibility prediction and reduce navigational risks, gridded visibility data based on artificial neural network analysis can be used over marine environments, and the problem can be regarded as an air quality prediction problem based on machine learning algorithms. This new method based on artificial intelligence techniques developed here is tested over the Indian Ocean. The mean error of the inferred visibility from the artificial neural network analysis is found to be less than 8.0%. The results suggested that satellite-based optical thickness and numerical model-based reanalysis data can be used to infer gridded visibility values based on artificial neural network analysis, and that could help us reconstruct and monitor surface gridded visibility values over marine and remote environments.

At present, some shortcomings of the research on coupling modeling of the oceanic front–sound field may need attention: (1) Most of the acoustic propagation simulation is based on ideal front models, but the application of investigated marine data is lacking. (2) Most studies focus on the acoustic field characteristics, with the influence of fronts on acoustic propagation, but few studies aim at the direct quantitative analysis of the performance of underwater acoustic detection in oceanic fronts. To deal with the above problems, based on the measured data in the northwest Pacific Ocean, here, we first design different sound source layout schemes and calculate sound field characteristics in the sub-Arctic front using the ray theory. Then, the cumulative detection probability model is built based on the active sonar equation to evaluate the efficiency of underwater detection. Finally, the detection probability is calculated and expressed by regionalization, and the influence of the sub-Arctic front on underwater detection is quantitatively analyzed. The results show that the sub-Arctic front can significantly affect the underwater acoustic detection. The sound source located in the front in the cold-water mass has a better detection performance, especially detecting towards cold water (horizontal detection range > 60 km). In contrast, the sound source located in the warm-water mass has a poor detection performance (horizontal detection range in shallow sea < 10 km).

The effective prediction of storm track (ST) is greatly beneficial for analyzing the development and anomalies of mid-latitude weather systems. For the non-stationarity, nonlinearity, and uncertainty of ST intensity index (STII), a new probabilistic prediction model was proposed based on dynamic Bayesian network (DBN) and wavelet analysis (WA). We introduced probability theory and graph theory for the first time to quantitatively describe the nonlinear relationship and uncertain interaction of the ST system. Then a casual prediction network (i.e., DBN) was constructed through wavelet decomposition, structural learning, parameter learning, and probabilistic inference, which was used for expression of relation among predictors and probabilistic prediction of STII. The intensity prediction of the North Pacific ST with data from 1961–2010 showed that the new model was able to give more comprehensive prediction information and higher prediction accuracy and had strong generalization ability and good stability.

Feature subset selection is the most important and difficult task in the field of fatigue fracture image identification. In this paper, a new method which is hybrid of linear prediction, called LP-Based Multi-Objective Genetic Algorithms (LP-MOGA) is proposed for fatigue fracture feature subset selection. In LP-MOGA, predicted new solutions with elite solutions by liner prediction to improve the local search ability. For fatigue fracture identification, texture character and fractal dimension feature are extracted for original features; and then, feature subset selection is performed by LP-MOGA, in which, the objective functions minimize error identification rate, undetected identification rate and selected featured number; at last, the identification is executed by quadratic distance classifier. Compared with other methods, the experiment results of actual data demonstrate the presented algorithm is effective.