Abstract
Changes in ocean temperature over time have important implications for marine ecosystems and global climate change. Marine temperature changes with time and has the features of closeness, period, and trend. This paper analyzes the temporal dependence of marine temperature variation at multiple depths and proposes a new ocean-temperature time-series prediction method based on the temporal dependence parameter matrix fusion of historical observation data. The Temporal Dependence-Based Long Short-Term Memory (LSTM) Networks for Marine Temperature Prediction (TD-LSTM) proves better than other methods while predicting sea-surface temperature (SST) by using Argo data. The performances were good at various depths and different regions.
Highlights
Seawater temperature is an important indicator that measures water heat and is, one of the most important physical factors of the marine environment
The performance of TD-Long Short-Term Memory (LSTM) is better than other variants of Support Vector Regression (SVR) and MLP
The LSTM method is better than the variants of SVR and Multilayer Perceptron Regressor (MLPR)
Summary
Seawater temperature is an important indicator that measures water heat and is, one of the most important physical factors of the marine environment. Ocean-temperature prediction makes humans better understand global climate change and marine ecosystems. Plan (ARGO), the Integrated Ocean Observation System (IOOS), the "NEPTUNE" seafloor observatory network planning (NEPTUNE), and European programs such as ESONET and DONET These observed data have been used for numerical results comparison and to increase knowledge on global ocean. From traditional empirical statistics methods to artificial intelligence approaches, there are many data-driven techniques for predicting ocean temperatures [8]. We put forward a TD-LSTM network to predict ocean-temperature changes. In TD-LSTM, we first analyze the temperature history of marine observation values, refer to their closeness, period, and trend, using the ocean-temperature fusion method based on time dependence to reconstruct the data sequence. According to the reconstructed LSTM model, we can predict the time series of ocean temperatures.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
