An official website of the United States government
Abstract We show that explainable neural networks can identify regions of oceanic variability that contribute predictability on decadal timescales in a fully coupled Earth‐system model. The neural networks learn to use sea‐surface temperature anomalies to predict future continental surface temperature anomalies. We then use a neural‐network explainability method called layerwise relevance propagation to infer which oceanic patterns lead to accurate predictions made by the neural networks. In particular, regions within the North Atlantic Ocean and North Pacific Ocean lend the most predictability for surface temperature across continental North America. We apply the proposed methodology to decadal variability, although the concept is generalizable to other timescales of predictability. Furthermore, while our approach focuses on predictable patterns of internal variability within climate models, it should be generalizable to observational data as well. Our study contributes to the growing evidence that explainable neural networks are important tools for advancing geoscientific knowledge.
more »
« less
Combining Neural Networks and CMIP6 Simulations to Learn Windows of Opportunity for Skillful Prediction of Multiyear Sea Surface Temperature Variability
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.
more »
« less
- Award ID(s):
- 1749261
- PAR ID:
- 10512173
- Publisher / Repository:
- DOI PREFIX: 10.1029
- Date Published:
- Journal Name:
- Geophysical Research Letters
- Volume:
- 51
- Issue:
- 11
- ISSN:
- 0094-8276
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Abstract An open question in the study of climate prediction is whether internal variability will continue to contribute to prediction skill in the coming decades, or whether predictable signals will be overwhelmed by rising temperatures driven by anthropogenic forcing. We design a neural network that is interpretable such that its predictions can be decomposed to examine the relative contributions of external forcing and internal variability to future regional sea surface temperature (SST) trend predictions in the near-term climate (2020–2050). We show that there is additional prediction skill to be garnered from internal variability in the Community Earth System Model version 2 Large Ensemble, even in a relatively high forcing future scenario. This predictability is especially apparent in the North Atlantic, North Pacific and Tropical Pacific Oceans as well as in the Southern Ocean. We further investigate how prediction skill covaries across the ocean and find three regions with distinct coherent prediction skill driven by internal variability. SST trend predictability is found to be associated with consistent patterns of decadal variability for the grid points within each region.more » « less
-
Abstract Interannual variability of the winter AR activities over the Northern hemisphere is investigated. The leading modes of AR variability over the North Pacific and North Atlantic are first identified and characterized. Over the Pacific, the first mode is characterized by a dipole structure with enhanced AR frequency along the AR peak region at about 30° N and reduced AR frequency further north. The second mode exhibits a tri-pole structure with a narrow band of positive AR anomalies at about 30° N and sandwiched by negative anomalies. Over the Atlantic, the first mode exhibits an equatorward shift of the ARs with positive anomalies and negative anomalies located on the equatorward and poleward side of the AR peak region at about 40° N , respectively. The second mode is associated with the strengthening and eastward extension of the AR peak region which is sandwiched by negative anomalies. A large ensemble of atmospheric global climate models from the Coupled Model Intercomparison Project phase 6 (CMIP6), which shows high skills in simulating these modes, is then used to quantify the roles of sea surface temperature (SST) forcing versus internal atmospheric variability in driving the formation of these modes. Results show that SST forcing explains about half of the variance for the Pacific leading modes, while that number drops to about a quarter for the Atlantic leading modes, suggesting higher predictability for the Pacific AR variability. Additional ensemble driven only by observed tropical SST is further utilized to demonstrate the more important role that tropical SST plays in controlling the Pacific AR variability while both tropical and extratropical SST exert comparable influences on the Atlantic AR variability.more » « less
-
Abstract The influence of El Niño–Southern Oscillation (ENSO) in the Asian monsoon region can persist through the post-ENSO summer, after the sea surface temperature (SST) anomalies in the tropical Pacific have dissipated. The long persistence of coherent post-ENSO anomalies is caused by a positive feedback due to interbasin ocean–atmospheric coupling, known as the Indo-western Pacific Ocean capacitor (IPOC) effect, although the feedback mechanism itself does not necessarily rely on the antecedence of ENSO events, suggesting the potential for substantial internal variability independent of ENSO. To investigate the respective role of ENSO forcing and non-ENSO internal variability, we conduct ensemble “forecast” experiments with a full-physics, globally coupled atmosphere–ocean model initialized from a multidecadal tropical Pacific pacemaker simulation. The leading mode of internal variability as represented by the forecast-ensemble spread resembles the post-ENSO IPOC, despite the absence of antecedent ENSO forcing by design. The persistent atmospheric and oceanic anomalies in the leading mode highlight the positive feedback mechanism in the internal variability. The large sample size afforded by the ensemble spread allows us to identify robust non-ENSO precursors of summer IPOC variability, including a cool SST patch over the tropical northwestern Pacific, a warming patch in the tropical North Atlantic, and downwelling oceanic Rossby waves in the tropical Indian Ocean south of the equator. The pathways by which the precursors develop into the summer IPOC mode and the implications for improved predictability are discussed.more » « less
-
Abstract In the U.S. Great Plains (GP), diagnosing precipitation variability is key in developing an understanding of the present and future availability of water in the region. Building on previous work investigating U.S. GP pluvial years, this study uses ERA twentieth century (ERA-20C) reanalysis data to investigate key circulation anomalies driving GP precipitation anomalies during a subset of GP pluvial years (called in this paper Pattern pluvial years). With previous research showing links between tropical Pacific sea surface temperature (SST) anomalies and GP climate variability, this study diagnoses the key circulation anomalies through an analysis of SSTs and their influence on the atmosphere. Results show that during Pattern southern Great Plains (SGP) pluvial years, central tropical Pacific SST anomalies are coincident with key atmospheric anomalies across the Pacific basin and North America. During northern Great Plains (NGP) Pattern pluvial years, no specific pattern of oceanic anomalies emerges that forces the circulation anomaly feature inherent in specific NGP pluvial years. Utilizing the results for SGP pluvial years, a conceptual model is developed detailing the identified pathway for the occurrence of circulation patterns that are favorable for pluvial years over the SGP. Overall, results from this study show the importance of the identified SGP atmospheric anomaly signal and the potential for predictability of such events.more » « less
