Global monthly sea surface temperature forecasting using the SARIMA, LSTM, and GRU models

BİLGİLİ M., PINAR E., Durhasan T.

Earth Science Informatics, cilt.18, sa.1, 2025 (SCI-Expanded) identifier

  • Yayın Türü: Makale / Tam Makale
  • Cilt numarası: 18 Sayı: 1
  • Basım Tarihi: 2025
  • Doi Numarası: 10.1007/s12145-024-01585-z
  • Dergi Adı: Earth Science Informatics
  • Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Academic Search Premier, CAB Abstracts, Geobase, INSPEC
  • Anahtar Kelimeler: Climate change, Climate dynamics, GRU model, LSTM model, SARIMA model, Sea surface temperature, Time series analysis
  • Çukurova Üniversitesi Adresli: Evet

Özet

Global warming has become one of the world’s most pressing problems in recent years, accompanied by rising sea surface temperature (SST). The SST time series data are an essential component in balancing the energy at the planet’s surface. It is of the utmost importance to forecast future SSTs to assist us in better comprehending the climate dynamics and identifying catastrophic circumstances in advance based on historical observations received from earth observation systems. In this sense, monitoring and forecasting SST has become vital for better understanding future climate trends. In this regard, this study proposes a gated recurrent units (GRUs) model, a long short-term memory (LSTM) neural network technique, and a seasonal auto-regressive integrated moving average (SARIMA) statistical model to predict global monthly SST data. According to the findings from the testing procedure, the MAPE values were 0.1377% for the SARIMA model, 0.1374% for the LSTM model, and 0.1390% for the GRU model. All models were found to have MAE, RMSE, and R values within the ranges of 0.0250–0.0253 oC, 0.032–0.0323 oC, and 0.9772–0.9775, respectively. The results of the proposed SARIMA, LSTM, and GRU models showed that they could accurately and satisfactorily predict the global monthly SST time series.