Analysis and forecast of sea level change in the China Seas and neighboring oceans from 1993 to 2020

Estimates and projections of sea level change are critical for coastal areas. In this work, we utilize satellite altimetry (SA) and tide gauge (TG) technologies to estimate variations in sea level, and we also evaluate the consistency of sea level changes obtained using TG and SA from 1993 to 2020. Additionally, we use deep learning models (artificial neural network (ANN), gated recurrent unit (GRU), and long short-term memory (LSTM)) to forecast sea level changes with SA time series. Our results reveal that the average absolute sea level (ASL) rate in the China Seas and the neighboring ocean based on SA is 3.55 mm/yr, which is higher than the global rate of 3.30 mm/yr. Specifically, the ASL rates of East China Sea and South China Sea are 3.21 mm/yr and 4.24 mm/yr, respectively. The sea level change in the South China Sea is significantly greater than that in the East China Sea. Secondly, the relative sea level (RSL) rate based on TGs is 3.88 mm/yr. We perform VLM correction on TGs with co-located GNSS following the method of Zhou et al. (2022) and obtain a TG-based ASL result of 3.77 mm/yr. Our results show that there is good consistency between coastal sea level changes estimated using tide gauges and satellite radar altimetry. Finally, we use the ANN, GRU, and LSTM models to predict sea level change with SA. The results show that LSTM’s prediction accuracy is better than that of the other models, with average RMSE, MAE, and R² values of 48.92 mm, 35.99 mm, and 0.85, respectively.