Detection of Climate Change-Induced Underwater Sound Speed Variations in the Global Ocean Using Satellite Data and Machine Learning

Rising ocean temperature due to global warming caused several environmental alterations, such as rising sea levels, ocean acidification, and changes in underwater sound propagation. Investigation of sound speed variations due to global warming is necessary since sound propagation changes heavily impact underwater acoustic applications and threaten many marine lives. However, the global-scale study of sound speed variation is impossible in the traditional way, where sound speed is calculated from in-situ profiling measurements of temperature, salinity, and pressure, which is expensive and, therefore, limited in locations and time. Based on satellite observed sea surface data, we used a machine learning approach to predict sound speed for any given time and location in the global ocean. The results show that our model prediction had a root-mean-square error of 0.26 m/s and a coefficient of determination of 0.99. About 99% of the estimates lie within ±0.4 m/s of the sound speeds obtained from in situ temperature and salinity profiles. Since our model correctly identified the sound speed variations due to latitudes and seasons, it can be used to detect climate change-induced underwater sound speed variations in the global ocean.