Deep learning for multi-year ENSO forecasts.

Variations in the El Niño/Southern Oscillation (ENSO) are associated with a wide array of regional climate extremes and ecosystem impacts1. Robust, long-lead forecasts would therefore be valuable for managing policy responses. But despite decades of effort, forecasting ENSO events at lead times of more than one year remains problematic2. Here we show that a statistical forecast model employing a deep-learning approach produces skilful ENSO forecasts for lead times of up to one and a half years. To circumvent the limited amount of observation data, we use transfer learning to train a convolutional neural network (CNN) first on historical simulations3 and subsequently on reanalysis from 1871 to 1973. During the validation period from 1984 to 2017, the all-season correlation skill of the Nino3.4 index of the CNN model is much higher than those of current state-of-the-art dynamical forecast systems. The CNN model is also better at predicting the detailed zonal distribution of sea surface temperatures, overcoming a weakness of dynamical forecast models. A heat map analysis indicates that the CNN model predicts ENSO events using physically reasonable precursors. The CNN model is thus a powerful tool for both the prediction of ENSO events and for the analysis of their associated complex mechanisms.

Author Correction: El Niño-Southern Oscillation complexity.

In this Review, the middle initial of author Kim M. Cobb was omitted. The original Review has been corrected online.

El Niño-Southern Oscillation complexity.

El Niño events are characterized by surface warming of the tropical Pacific Ocean and weakening of equatorial trade winds that occur every few years. Such conditions are accompanied by changes in atmospheric and oceanic circulation, affecting global climate, marine and terrestrial ecosystems, fisheries and human activities. The alternation of warm El Niño and cold La Niña conditions, referred to as the El Niño-Southern Oscillation (ENSO), represents the strongest year-to-year fluctuation of the global climate system. Here we provide a synopsis of our current understanding of the spatio-temporal complexity of this important climate mode and its influence on the Earth system.

Seasonal Predictions of Summer Precipitation in the Middle-lower Reaches of the Yangtze River with Global and Regional Models Based on NUIST-CFS1.0.

Accurate prediction of the summer precipitation over the middle and lower reaches of the Yangtze River (MLYR) is of urgent demand for the local economic and societal development. This study assesses the seasonal forecast skill in predicting summer precipitation over the MLYR region based on the global Climate Forecast System of Nanjing University of Information Science and Technology (NUIST-CFS1.0, previously SINTEX-F). The results show that the model can provide moderate skill in predicting the interannual variations of the MLYR rainbands, initialized from 1 March. In addition, the nine-member ensemble mean can realistically reproduce the links between the MLYR precipitation and tropical sea surface temperature (SST) anomalies, but the individual members show great discrepancies, indicating large uncertainty in the forecasts. Furthermore, the NUIST-CFS1.0 can predict five of the seven extreme summer precipitation anomalies over the MLYR during 1982-2020, albeit with underestimated magnitudes. The Weather Forecast and Research (WRF) downscaling hindcast experiments with a finer resolution of 30 km, which are forced by the large-scale information of the NUIST-CFS1.0 predictions with a spectral nudging method, display improved predictions of the extreme summer precipitation anomalies to some extent. However, the performance of the downscaling predictions is highly dependent on the global model forecast skill, suggesting that further improvements on both the global and regional climate models are needed.

Indian Ocean warming modulates Pacific climate change.

It has been widely believed that the tropical Pacific trade winds weakened in the last century and would further decrease under a warmer climate in the 21st century. Recent high-quality observations, however, suggest that the tropical Pacific winds have actually strengthened in the past two decades. Precise causes of the recent Pacific climate shift are uncertain. Here we explore how the enhanced tropical Indian Ocean warming in recent decades favors stronger trade winds in the western Pacific via the atmosphere and hence is likely to have contributed to the La Niña-like state (with enhanced east-west Walker circulation) through the Pacific ocean-atmosphere interactions. Further analysis, based on 163 climate model simulations with centennial historical and projected external radiative forcing, suggests that the Indian Ocean warming relative to the Pacific's could play an important role in modulating the Pacific climate changes in the 20th and 21st centuries.

Multi-task machine learning improves multi-seasonal prediction of the Indian Ocean Dipole.

As one of the most predominant interannual variabilities, the Indian Ocean Dipole (IOD) exerts great socio-economic impacts globally, especially on Asia, Africa, and Australia. While enormous efforts have been made since its discovery to improve both climate models and statistical methods for better prediction, current skills in IOD predictions are mostly limited up to three months ahead. Here, we challenge this long-standing problem using a multi-task deep learning model that we name MTL-NET. Hindcasts of the IOD events during the past four decades indicate that the MTL-NET can predict the IOD well up to 7-month ahead, outperforming most of world-class dynamical models used for comparison in this study. Moreover, the MTL-NET can help assess the importance of different predictors and correctly capture the nonlinear relationships between the IOD and predictors. Given its merits, the MTL-NET is demonstrated to be an efficient model for improved IOD prediction.

Increasing dominance of Indian Ocean variability impacts Australian wheat yields.

The relationships between crop productivity and climate variability drivers are often assumed to be stationary over time. However, this may not be true in a warming climate. Here we use a crop model and a machine learning algorithm to demonstrate the changing impacts of climate drivers on wheat productivity in Australia. We find that, from the end of the nineteenth century to the 1980s, wheat productivity was mainly subject to the impacts of the El Niño Southern Oscillation. Since the 1990s, the impacts from the El Niño Southern Oscillation have been decreasing, but those from the Indian Ocean Dipole have been increasing. The warming climate has brought more occurrences of positive Indian Ocean Dipole events, resulting in severe yield reductions in recent decades. Our findings highlight the need to adapt seasonal forecasting to the changing impacts of climate variability to inform the management of climate-induced yield losses.

Multi-year El Niño events tied to the North Pacific Oscillation.

Multi-year El Niño events induce severe and persistent floods and droughts worldwide, with significant socioeconomic impacts, but the causes of their long-lasting behaviors are still not fully understood. Here we present a two-way feedback mechanism between the tropics and extratropics to argue that extratropical atmospheric variability associated with the North Pacific Oscillation (NPO) is a key source of multi-year El Niño events. The NPO during boreal winter can trigger a Central Pacific El Niño during the subsequent winter, which excites atmospheric teleconnections to the extratropics that re-energize the NPO variability, then re-triggers another El Niño event in the following winter, finally resulting in persistent El Niño-like states. Model experiments, with the NPO forcing assimilated to constrain atmospheric circulation, reproduce the observed connection between NPO forcing and the occurrence of multi-year El Niño events. Future projections of Coupled Model Intercomparison Project phases 5 and 6 models demonstrate that with enhanced NPO variability under future anthropogenic forcing, more frequent multi-year El Niño events should be expected. We conclude that properly accounting for the effects of the NPO on the evolution of El Niño events may improve multi-year El Niño prediction and projection.

Extreme fire weather is the major driver of severe bushfires in southeast Australia.

In Australia, the proportion of forest area that burns in a typical fire season is less than for other vegetation types. However, the 2019-2020 austral spring-summer was an exception, with over four times the previous maximum area burnt in southeast Australian temperate forests. Temperate forest fires have extensive socio-economic, human health, greenhouse gas emissions, and biodiversity impacts due to high fire intensities. A robust model that identifies driving factors of forest fires and relates impact thresholds to fire activity at regional scales would help land managers and fire-fighting agencies prepare for potentially hazardous fire in Australia. Here, we developed a machine-learning diagnostic model to quantify nonlinear relationships between monthly burnt area and biophysical factors in southeast Australian forests for 2001-2020 on a 0.25° grid based on several biophysical parameters, notably fire weather and vegetation productivity. Our model explained over 80% of the variation in the burnt area. We identified that burnt area dynamics in southeast Australian forest were primarily controlled by extreme fire weather, which mainly linked to fluctuations in the Southern Annular Mode (SAM) and Indian Ocean Dipole (IOD), with a relatively smaller contribution from the central Pacific El Niño Southern Oscillation (ENSO). Our fire diagnostic model and the non-linear relationships between burnt area and environmental covariates can provide useful guidance to decision-makers who manage preparations for an upcoming fire season, and model developers working on improved early warning systems for forest fires.

Over-projected Pacific warming and extreme El Niño frequency due to CMIP5 common biases.

Extreme El Niño events severely disrupt the global climate, causing pronounced socio-economic losses. A prevailing view is that extreme El Niño events, defined by total precipitation or convection in the Niño3 area, will increase 2-fold in the future. However, this projected change was drawn without removing the potential impacts of Coupled Model Intercomparison Project phase 5 (CMIP5) models' common biases. Here, we find that the models' systematic biases in simulating tropical climate change over the past century can reduce the reliability of the projected change in the Pacific sea surface temperature (SST) and its related extreme El Niño frequency. The projected Pacific SST change, after removing the impacts of 13 common biases, displays a 'La Niña-like' rather than 'El Niño-like' change. Consequently, the extreme El Niño frequency, which is highly linked to the zonal distribution of the Pacific SST change, would remain mostly unchanged under CMIP5 warming scenarios. This finding increases confidence in coping with climate risks associated with global warming.

Decadal climate variability in the tropical Pacific: Characteristics, causes, predictability, and prospects.

Climate variability in the tropical Pacific affects global climate on a wide range of time scales. On interannual time scales, the tropical Pacific is home to the El Niño­Southern Oscillation (ENSO). Decadal variations and changes in the tropical Pacific, referred to here collectively as tropical Pacific decadal variability (TPDV), also profoundly affect the climate system. Here, we use TPDV to refer to any form of decadal climate variability or change that occurs in the atmosphere, the ocean, and over land within the tropical Pacific. "Decadal," which we use in a broad sense to encompass multiyear through multidecadal time scales, includes variability about the mean state on decadal time scales, externally forced mean-state changes that unfold on decadal time scales, and decadal variations in the behavior of higher-frequency modes like ENSO.

Using large-scale climate drivers to forecast meteorological drought condition in growing season across the Australian wheatbelt.

Recurring drought has caused large crop yield losses in Australia during past decades. Long-term drought forecasting is of great importance for the development of risk management strategies. Recently, large-scale climate drivers (e.g. El Niño-Southern Oscillation) have been demonstrated as useful in the application of drought forecasting. Machine learning-based models that use climate drivers as input are commonly adopted to provide drought forecasts as these models are easy to develop and require less information compared to physical-based models. However, few machine learning-based models have been developed to forecast drought conditions during growing season across all Australian cropping areas. In this study, we developed a growing season (Apr.-Nov.) meteorological drought forecasting model for each climate gauging location across the Australian wheatbelt based on multiple lagged (past) large-scale climate indices and the Random Forest (RF) algorithm. The Standardized Precipitation Index (SPI) was used as the response variable to measure the degree of meteorological drought. Results showed that the RF model could provide satisfactory drought forecasts in the eastern areas of the wheatbelt with Pearson's correlation coefficient r > 0.5 and normalized Root Mean Square Error (nRMSE) < 23%. Forecasted drought maps matched well with observed drought maps for three representative periods. We identified NINO3.4 sea surface temperature and Multivariate ENSO Index as the most influential indices dominating growing season drought conditions across the wheatbelt. In addition, lagged impacts of large-scale climate drivers on growing season drought conditions were long-lasting and the indices in previous year could also potentially affect drought conditions during current year. As large-scale climate indices are readily available and can be rapidly used to feed data driven models, we believe the proposed meteorological drought forecasting models can be easily extended to other regions to provide drought outlooks which can help mitigate adverse drought impacts.

Pantropical climate interactions.

The El Niño-Southern Oscillation (ENSO), which originates in the Pacific, is the strongest and most well-known mode of tropical climate variability. Its reach is global, and it can force climate variations of the tropical Atlantic and Indian Oceans by perturbing the global atmospheric circulation. Less appreciated is how the tropical Atlantic and Indian Oceans affect the Pacific. Especially noteworthy is the multidecadal Atlantic warming that began in the late 1990s, because recent research suggests that it has influenced Indo-Pacific climate, the character of the ENSO cycle, and the hiatus in global surface warming. Discovery of these pantropical interactions provides a pathway forward for improving predictions of climate variability in the current climate and for refining projections of future climate under different anthropogenic forcing scenarios.

Inter-basin sources for two-year predictability of the multi-year La Niña event in 2010-2012.

Multi-year La Niña events often induce persistent cool and wet climate over global lands, altering and in some case mitigating regional climate warming impacts. The latest event lingered from mid-2010 to early 2012 and brought about intensive precipitation over many land regions of the world, particularly Australia. This resulted in a significant drop in global mean sea level despite the background upwards trend. This La Niña event is surprisingly predicted out to two years ahead in a few coupled models, even though the predictability of El Niño-Southern Oscillation during 2002-2014 has declined owing to weakened ocean-atmosphere interactions. However, the underlying mechanism for high predictability of this multi-year La Niña episode is still unclear. Experiments based on a climate model that demonstrates a successful two-year forecast of the La Niña support the hypothesis that warm sea surface temperature (SST) anomalies in the Atlantic and Indian Oceans act to intensify the easterly winds in the central equatorial Pacific and largely contribute to the occurrence and two-year predictability of the 2010-2012 La Niña. The results highlight the importance of increased Atlantic-Indian Ocean SSTs for the multi-year La Niña's predictability under global warming.

Internal Variability-Generated Uncertainty in East Asian Climate Projections Estimated with 40 CCSM3 Ensembles.

Regional climate projections are challenging because of large uncertainty particularly stemming from unpredictable, internal variability of the climate system. Here, we examine the internal variability-induced uncertainty in precipitation and surface air temperature (SAT) trends during 2005-2055 over East Asia based on 40 member ensemble projections of the Community Climate System Model Version 3 (CCSM3). The model ensembles are generated from a suite of different atmospheric initial conditions using the same SRES A1B greenhouse gas scenario. We find that projected precipitation trends are subject to considerably larger internal uncertainty and hence have lower confidence, compared to the projected SAT trends in both the boreal winter and summer. Projected SAT trends in winter have relatively higher uncertainty than those in summer. Besides, the lower-level atmospheric circulation has larger uncertainty than that in the mid-level. Based on k-means cluster analysis, we demonstrate that a substantial portion of internally-induced precipitation and SAT trends arises from internal large-scale atmospheric circulation variability. These results highlight the importance of internal climate variability in affecting regional climate projections on multi-decadal timescales.

Robust contribution of decadal anomalies to the frequency of central-Pacific El Niño.

During year-to-year El Niño events in recent decades, major sea surface warming has occurred frequently in the central Pacific. This is distinct from the eastern Pacific warming pattern during canonical El Niño events. Accordingly, the central-Pacific El Niño exerts distinct impacts on ecosystems, climate and hurricanes worldwide. The increased frequency of the new type of El Niño presents a challenge not only for the understanding of El Niño dynamics and its change but also for the prediction of El Niño and its global impacts at present and future climate. Previous studies have proposed different indices to represent the two types of El Niño for better understanding, prediction and impact assessment. Here, we find that all popularly used indices for the central-Pacific El Niño show a dominant spectral peak at a decadal period with comparatively weak variance at interannual timescales. Our results suggest that decadal anomalies have an important contribution to the occurrence of the central-Pacific El Niño over past decades. Removing the decadal component leads to a significant reduction in the frequency of the central-Pacific El Niño in observations and in Coupled Model Intercomparison Project Phase 5 simulations of preindustrial, historical and future climate.

Skilful multi-year predictions of tropical trans-basin climate variability.

Tropical Pacific sea surface temperature anomalies influence the atmospheric circulation, impacting climate far beyond the tropics. The predictability of the corresponding atmospheric signals is typically limited to less than 1 year lead time. Here we present observational and modelling evidence for multi-year predictability of coherent trans-basin climate variations that are characterized by a zonal seesaw in tropical sea surface temperature and sea-level pressure between the Pacific and the other two ocean basins. State-of-the-art climate model forecasts initialized from a realistic ocean state show that the low-frequency trans-basin climate variability, which explains part of the El Niño Southern Oscillation flavours, can be predicted up to 3 years ahead, thus exceeding the predictive skill of current tropical climate forecasts for natural variability. This low-frequency variability emerges from the synchronization of ocean anomalies in all basins via global reorganizations of the atmospheric Walker Circulation.

Impacts of El Niño Southern Oscillation on the global yields of major crops.

The monitoring and prediction of climate-induced variations in crop yields, production and export prices in major food-producing regions have become important to enable national governments in import-dependent countries to ensure supplies of affordable food for consumers. Although the El Niño/Southern Oscillation (ENSO) often affects seasonal temperature and precipitation, and thus crop yields in many regions, the overall impacts of ENSO on global yields are uncertain. Here we present a global map of the impacts of ENSO on the yields of major crops and quantify its impacts on their global-mean yield anomalies. Results show that El Niño likely improves the global-mean soybean yield by 2.1-5.4% but appears to change the yields of maize, rice and wheat by -4.3 to +0.8%. The global-mean yields of all four crops during La Niña years tend to be below normal (-4.5 to 0.0%). Our findings highlight the importance of ENSO to global crop production.