Basic principles of the strapdown inertial navigation system (SINS) using the outputs of strapdown gyros and accelerometers are explained, and the main equations are described. A mathematical model of SINS is established, and its Matlab implementation is developed. The theory is illustrated by six examples which are static status, straight line movement, circle movement, s-shape movement, and two sets of real static data.

Underwater target radiated noise feature extraction and classification are important issues in underwater acoustic applications. In this paper., feature extraction is processed based on Gammatone filter and the target classification is processed using machine learning (ML). From the processed result of the real underwater target data, it showed that Gammatone filter is an efficient way to do feature extraction and it also has better classification accuracy compared with some other feature extracting methods. It also showed that machine learning is an efficient tool when applied in underwater target radiated noise classification where the assignment is a label to given input value.

Sea surface salinity (SSS) is one of the Essential Climate Variables (ECVs) as defined by the Global Climate Observing System (GCOS). Acquiring high-quality SSS datasets with high spatial-temporal resolution is crucial for research on the hydrological cycle and the earth climate. This study assessed the quality of SSS data provided by five high-resolution ocean reanalysis products, including the Hybrid Coordinate Ocean Model (HYCOM) 1/12° global reanalysis, the Copernicus Global 1/12° Oceanic and Sea Ice GLORYS12 Reanalysis, the Simple Ocean Data Assimilation (SODA) reanalysis, the ECMWF Oceanic Reanalysis System 5 (ORAS5) product and the Estimating the Circulation and Climate of the Ocean Phase II (ECCO2) reanalysis. Regional comparison in the Mediterranean Sea shows that reanalysis largely depicts the accurate spatial SSS structure away from river mouths and coastal areas but slightly underestimates the mean SSS values. Better SSS reanalysis performance is found in the Levantine Sea while larger SSS uncertainties are found in the Adriatic Sea and the Aegean Sea. The global comparison with CMEMS level-4 (L4) SSS shows generally consistent large-scale structures. The mean ΔSSS between monthly gridded reanalysis data and in situ analyzed data is −0.1 PSU in the open seas between 40° S and 40° N with the mean Root Mean Square Deviation (RMSD) generally smaller than 0.3 PSU and the majority of correlation coefficients higher than 0.5. A comparison with collocated buoy salinity shows that reanalysis products well capture the SSS variations at the locations of tropical moored buoy arrays at weekly scale. Among all of the five products, the data quality of HYCOM reanalysis SSS is highest in marginal sea, GLORYS12 has the best performance in the global ocean especially in tropical regions. Comparatively, ECCO2 has the overall worst performance to reproduce SSS states and variations by showing the largest discrepancies with CMEMS L4 SSS.

In shallow water, passive sonar usually has great difficulty in discriminating a surface acoustic source from an underwater one. To solve this problem, a supervised machine learning method using only one hydrophone is implemented in this paper. Firstly, simulated training data are generated by a normal mode model KRAKEN with the same environment setup as that in SACLANT 1993 experiment. Secondly, the k-nearest neighbor (kNN) classifiers are trained and evaluated using the scores of precision, recall, F1 and accuracy. Thirdly, the random subspace kNN classifiers are finely trained on three hyperparameters (the number of nearest neighbors, the number of predictors selected at random and the number of learners in the ensemble) to obtain the best model. Fourthly, a deep learning method called ResNet-18 is also applied, and it reaches the best balance between precision and recall, while the accuracies of both simulation and experimental data are all 1.0. Further, data from all 48 hydrophones of the vertical linear array (VLA) are analyzed using the three kinds of machine learning methods (kNN, random subspace kNN and ResNet-18) separately, and the results are compared. It is concluded that the performance of random subspace kNN is the best. Both the simulation and experimental results suggest the feasibility of machine learning as a surface and underwater acoustic source discrimination method even with only a single hydrophone.

Aiming at the problems in the bearing-only tracking (BOT) of underwater weak target in the presence of interferers, such as discontinuity of tracking trajectory and target loss, a novel joint algorithm based on interference suppression (IS) and Kalman filter (ISKF) is proposed in this paper. In the proposed algorithm, in order to smooth the corrupted trajectory of weak target and improve the precision of tracking, a feedback structure which connects IS module and Kalman filter (KF) module is built. Furthermore, the feedback structure renders the measured results of bearing trajectory shareable mutually between IS module and KF module in real-time. In ISKF joint algorithm, IS module is employed as a preprocessing module to eliminate the power of interferers, and KF module is employed as tracker to obtain the bearing trajectory of weak target. In the initial step, the coarse target bearing and interference bearing are estimated using the conventional beamforming (CBF) method. In the updated step, the precise BOT trajectory of weak target is jointly updated using IS and KF. Moreover, when the target intersects with the interference, the ISKF joint algorithm can still track the weak target effectively with the aid of prediction model of KF. Finally, the sea trial results show that the proposed algorithm can achieve more accurate and robust real-time tracking than KF only, which benefits from the feedback between the IS module and the KF module.