Predicting Future Global Sea Level Rise From Climate Change Variables Using Deep Learnin

Abstract

Rapid climate change accelerates global temperature rise, causing thermal expansion of seawater and melting of ice-based lands, such as ice sheets and glaciers; these anomalies eventually result in global sea level rise. Since the beginning of satellite records, the sea level has risen significantly faster in recent decades than in prior decades, affecting people living in coastal areas directly as well as indirectly causing many environmental abnormalities. It is now possible to continuously monitor the level of seawater using current technology, but to battle this problem, it is necessary to understand the current scenario as well as predict the future scenario of sea level so that people may prepare and researchers can develop a viable solution, which is the main objective of this study. Here, 29 years of data on variables that are closely related to climate change, such as global temperature anomaly, ocean heat content change, carbon dioxide level in the atmosphere, and mass variation in Antarctica and Greenland, was gathered to build a multivariant prediction model using advance deep learning algorithms such as Recurrent Neural Network (RNN), Long Short Term Memory (LSTM), Gated Recurrent Unit (GRU), WaveNet (a type of Deep Convolutional Neural Network), and Deep Hybrid Network to predict the future scenario of global sea level rise. The results indicate that each method performs up to a certain level, but the deep hybrid model performed best in terms of accurately detecting the pattern of the dataset where MAE is 5.77 and RMSE is 7.67. Deep learning algorithms are admirable at identifying patterns in time series datasets, and with the necessary optimization, they can also assist in uncovering future data.