Back to Search
Start Over
Combining Neural Networks and CMIP6 Simulations to Learn Windows of Opportunity for Skillful Prediction of Multiyear Sea Surface Temperature Variability.
- Source :
- Geophysical Research Letters; 6/16/2024, Vol. 51 Issue 11, p1-11, 11p
- Publication Year :
- 2024
-
Abstract
- We use neural networks and large climate model ensembles to explore predictability of internal variability in sea surface temperature (SST) anomalies on interannual (1–3 years) and decadal (1–5 and 3–7 years) timescales. We find that neural networks can skillfully predict SST anomalies at these lead times, especially in the North Atlantic, North Pacific, Tropical Pacific, Tropical Atlantic and Southern Ocean. The spatial patterns of SST predictability vary across the nine climate models studied. The neural networks identify "windows of opportunity" where future SST anomalies can be predicted with more certainty. Neural networks trained on climate models also make skillful SST predictions in reconstructed observations, although the skill varies depending on which climate model the network was trained. Our results highlight that neural networks can identify predictable internal variability within existing climate data sets and show important differences in how well patterns of SST predictability in climate models translate to the real world. Plain Language Summary: We train neural networks (a machine learning model) to predict sea surface temperature (SST) between 3 and 7 years in the future. The neural networks are trained using data from existing climate model simulations. The regions where neural networks make the most accurate predictions depend on which climate model is used for training. The neural networks also make accurate predictions when given a data set of reconstructed SST observations, which means some of the patterns learned from the climate models also apply to the real climate system. However, there are unique differences between prediction accuracy in climate models and the reconstructed observations, which suggests directions for future research. Key Points: Neural networks can learn predictable signals of internal sea surface temperature (SST) variability at 1–3, 1–5, and 3–7 years lead timesNeural networks trained on climate model output can skillfully predict SST variability in reconstructed observationsNeural network skill in predicting observed SST variability depends on the climate model used for training [ABSTRACT FROM AUTHOR]
- Subjects :
- OCEAN temperature
ATMOSPHERIC models
FORECASTING
Subjects
Details
- Language :
- English
- ISSN :
- 00948276
- Volume :
- 51
- Issue :
- 11
- Database :
- Complementary Index
- Journal :
- Geophysical Research Letters
- Publication Type :
- Academic Journal
- Accession number :
- 177798696
- Full Text :
- https://doi.org/10.1029/2023GL108099