Seasonal Prediction Research Articles

Ensemble design for seasonal climate predictions: studying extreme Arctic sea ice lows with a rare event algorithm

Abstract. Initialized ensemble simulations can help identify the physical drivers and assess the probabilities of weather and climate extremes based on a given initial state. However, the significant computational burden of complex climate models makes it challenging to quantitatively investigate extreme events with probabilities below a few percent. A possible solution to overcome this problem is to use rare event algorithms, i.e. computational techniques originally developed in statistical physics that increase the sampling efficiency of rare events in numerical simulations. Here, we apply a rare event algorithm to ensemble simulations with the intermediate-complexity coupled climate model PlaSim-LSG to study extremes of pan-Arctic sea ice area reduction under pre-industrial greenhouse gas conditions. We construct four pairs of control and rare event algorithm ensemble simulations, each starting from four different initial winter sea ice states. The rare event simulations produce sea ice lows with probabilities of 2 orders of magnitude smaller than feasible with the control ensembles and drastically increase the number of extremes compared to direct sampling. We find that for a given probability level, the amplitude of negative late-summer sea ice area anomalies strongly depends on the baseline winter sea ice thickness but hardly on the baseline winter sea ice area. Finally, we investigate the physical processes in two trajectories leading to sea ice lows with conditional probabilities of less than 0.001 %. In both cases, negative late-summer pan-Arctic sea ice area anomalies are preceded by negative spring sea ice thickness anomalies. These are related to enhanced surface downward longwave radiative and sensible heat fluxes in an anomalously moist, cloudy and warm atmosphere. During summer, extreme sea ice area reduction is favoured by enhanced open-water-formation efficiency, anomalously strong downward solar radiation and the sea ice–albedo feedback. This work highlights that the most extreme summer sea ice conditions result from the combined effects of preconditioning and weather variability, emphasizing the need for thoughtful ensemble design when turning to real applications.

Seasonal Forecasts of Tropical Cyclones Using GFDL SPEAR and HiFLOR-S

Abstract The seasonal prediction skill of tropical cyclone (TC) activity is evaluated using the Seamless System for Prediction and Earth System Research (SPEAR), a modeling system developed at the Geophysical Fluid Dynamics Laboratory (GFDL) for experimental real-time seasonal forecasts. Compared with previous GFDL seasonal prediction models, SPEAR demonstrates improved skill in predicting TC activity for the western North Pacific, while exhibiting comparable or slightly degraded skill for the eastern North Pacific and North Atlantic. These changes in prediction skill do not always align with changes in prediction skill in large-scale variables, particularly over the North Atlantic. This study highlights that changes in the model’s response of TCs to large-scale variables, as well as the changes in the amplitude of interannual variations in TC genesis frequency, are crucial for the changes in TC prediction skill. Using the predicted sea surface temperatures from SPEAR as lower boundary conditions, the High-Resolution Forecast-Oriented Low Ocean Resolution (HiFLOR-S) model was employed to predict intense TCs, demonstrating skillful predictions of major hurricanes that are comparable to the previous HiFLOR coupled model predictions. Significance Statement This study reveals the prediction skill in the seasonal forecasting of tropical cyclones using a new experimental real-time seasonal prediction system developed at the Geophysical Fluid Dynamics Laboratory. The new system demonstrates skillful prediction of tropical cyclones in the western North Pacific, eastern North Pacific, and North Atlantic a few months before the hurricane season, with notable differences in the skill compared to the previous prediction system. The findings suggest that higher prediction skill in large-scale variables, such as vertical wind shear and sea surface temperatures, does not necessarily lead to higher prediction skill for tropical cyclones. This underscores that even when a model accurately predicts large-scale variables, its predictions of tropical cyclones could still be inaccurate. Our findings emphasize the need to refine the model’s response of tropical cyclones to specific large-scale environments, rather than focusing only on improving large-scale environment predictions, to enhance the accuracy of dynamical seasonal predictions for tropical cyclones.

Seasonal Prediction of Atmospheric Rivers in the Western North Pacific Using a Seasonal Prediction Model

ABSTRACTSeasonal prediction of atmospheric rivers (ARs) in the western north Pacific (WNP) is examined using a seasonal prediction model with and without atmospheric initialisation. A 20‐year seasonal prediction was conducted to evaluate the model's prediction skill, particularly focusing over the Japan area. The prediction skill of the present model indicated that the seasonal AR frequency is predictable with a lead time of up to 7–10 months, and the atmospheric initialisation further improved the skill. An additional investigation was conducted to identify the source of predictability for seasonal ARs. One significant source is the predictability of the Pacific‐Japan (PJ) pattern, which is influenced by the model's skill in predicting tropical sea surface temperature (SST) variability. The anticyclonic circulation southeast of Japan is well predicted when the tropical SST variability and PJ pattern are accurately predicted. Another source of predictability difference originated from the subsurface sea temperature (SBT) beneath the subtropical high in the North Pacific. When the SBT prediction is improved with atmospheric initialisation, it enhances the air‐sea interactions over the subtropical high in the WNP and southeast of Japan, leading to better predictability of seasonal ARs.

Evaluation of Seasonal Precipitation in South Asian Monsoon Using FGOALS-f2 Seamless Prediction System

Abstract Timely and effective seasonal climate prediction is crucial to reduce the losses from hydrometeorological disasters by providing early warning. Evaluating the model’s performance is essential for its accurate interpretation and implementation. Here, we examined the precipitation forecast of the South Asian monsoon using the second-generation finite-volume version of the Flexible Global Ocean–Atmosphere–Land System Model (FGOALS-f2), a seamless prediction system. The hindcast skill was assessed using multiple metrics for lead times of 0–2 months. The result demonstrates that the prediction skill is positive across the majority of regions up to a 1-month lead. However, the skill varies on lead times and geographical areas, with notably better skills up to a 1-month lead in central (particularly surrounding Nepal) and western (southern Pakistan and Afghanistan) regions. Furthermore, we proposed a physical mechanism to explain the seasonal prediction skill of precipitation and the tropical sea surface temperature (SST) signal, along with the Indian Ocean dipole (IOD) relationship. Typically, the model shows better forecasting accuracy in regions where it accurately represents the strong connection with SST indices. Additionally, for a 0-month lead forecast, the model performs well in predicting El Niño–Southern Oscillation (ENSO) and IOD indices, with correlation values of 0.71 and 0.52, respectively. We also explored how tropical cyclones could affect predicting precipitation in South Asia. Finally, we presented and compared real-time forecasts from 2020 to 2023. The prediction results produced by FGOALS-f2 are conclusively valuable for most of South Asia.

The WWRP/WCRP S2S Project and Its Achievements

Abstract The World Weather Research Programme (WWRP)/World Climate Research Programme (WCRP) Subseasonal to Seasonal (S2S) Prediction project was launched in 2013 with the primary goals of improving forecast skill and understanding sources of predictability on the subseasonal time scale (from 2 weeks to a season) around the globe. Particular emphasis was placed on high-impact weather events, on developing coordination among operational centers, and on promoting the use of subseasonal forecasts by the application communities. This 10-yr project ended in December 2023. A key accomplishment was the establishment of a database of subseasonal forecasts, called the S2S database. This database enhanced collaboration between the research and operational communities, enabled studies on a wide range of topics, and contributed to significant advances toward a better understanding of subseasonal predictability and windows of opportunity that contributed to improvements in forecast skill. It was used to train machine learning methods and test their performance in the S2S artificial intelligence/machine learning (AI/ML) prize challenge. The S2S project coorganized several coordinated research experiments to advance understanding of subseasonal predictability and the Real-Time Pilot Initiative that provided real-time access to subseasonal data for 15 application projects. A sequence of training courses sustained over 10 years enhanced the capacity of national meteorological services in the Global South to make subseasonal forecasts. A major legacy of the S2S project was the establishment and designation of the World Meteorological Organization (WMO) Global Producing Centres and Lead Centre for Subseasonal Prediction Multi-Model Ensemble, which will provide real-time subseasonal multimodel ensemble (MME) products to national and regional meteorological services.

Predictable equatorial Atlantic variability from atmospheric convection-ocean coupling

The Atlantic Niño exerts profound impacts on regional and global atmospheric circulation and climate, and on equatorial Atlantic biogeochemistry and ecosystems. However, the mode’s prediction remains a challenge which has been partly attributed to weak atmosphere-ocean coupling in the region. Here we introduce a framework that enhances the detection of the coupling between meridional migrations of atmospheric deep convection and zonal thermocline feedback. This approach reveals high predictive skill in a 196-member seasonal prediction ensemble, demonstrating robust predictability at 1–5-month forecast initialization lead times. The coupled mode is strongly correlated with land-precipitation variability across the tropics. The predictive skill largely originates in the Atlantic Ocean and is uncorrelated with El Niño Southern Oscillation in the Pacific, the leading mode of interannual climate variability globally. These skillful predictions raise hopes for enabled action in advance to avoid the most severe societal impacts in the affected countries.

A Regimes‐Based Approach to Identifying Seasonal State‐Dependent Prediction Skill

AbstractSubseasonal‐to‐decadal atmospheric prediction skill attained from initial conditions is typically limited by the chaotic nature of the atmosphere. However, for some atmospheric phenomena, prediction skill on subseasonal‐to‐decadal timescales is increased when the initial conditions are in a particular state. In this study, we employ machine learning to identify sea surface temperature (SST) regimes that enhance prediction skill of North Atlantic atmospheric circulation. An ensemble of artificial neural networks is trained to predict anomalous, low‐pass filtered 500 mb height at 7–8 weeks lead using SST. We then use self‐organizing maps (SOMs) constructed from 9 regions within the SST domain to detect state‐dependent prediction skill. SOMs are built using the entire SST time series, and we assess which SOM units feature confident neural network predictions. Four regimes are identified that provide skillful seasonal predictions of 500 mb height. Our findings demonstrate the importance of extratropical decadal SST variability in modulating downstream ENSO teleconnections to the North Atlantic. The methodology presented could aid future forecasting on subseasonal‐to‐decadal timescales.

Initialized seasonal prediction with the NCAR models in the North American Multi-Model Ensemble (NMME)

Initialized seasonal prediction with the NCAR models in the North American Multi-Model Ensemble (NMME)

A Skillful Prediction of Monsoon Intraseasonal Oscillation Using Deep Learning

AbstractThe northward‐propagating 30–60 days mode of monsoon rainfall anomalies over India, commonly referred to as the monsoon intraseasonal oscillation (MISO), plays a critical role in driving the active and break spells over the monsoon zone of the country. These oscillations are essential to understanding and predicting the variability of the Indian summer monsoon, which has significant implications for agriculture and water management. This study uses daily precipitation data from the TRMM/GPM satellite to derive MISO indices (MISO1 and MISO2). These indices were obtained through an extended empirical orthogonal function analysis conducted on 25 years of daily rainfall anomalies over the Indian region. The long time series of MISO1 and MISO2 indices generated from this analysis were then used to forecast future values using a transformer‐based deep learning model. The deep learning model demonstrated skilful predictions of the MISO indices for 2018–2022, with forecast lead times extending to 18 days. Notably, the model outperformed conventional operational numerical weather prediction models in predicting the MISO indices. These results indicate the potential for more reliable sub‐seasonal to seasonal (S2S) predictions of the Indian monsoon. The findings from this work highlight the effectiveness of using advanced deep learning techniques, such as Transformer architectures, in enhancing the predictability of complex atmospheric phenomena like MISO, thereby improving the outlook for monsoon forecasting.

Distinct Impacts of the Central and Eastern Atlantic Niño on West Antarctic Sea Ice

The sea ice variabilities in West Antarctica, crucial for both local and global climate systems, are profoundly affected by the sea surface temperature anomalies over the tropical Atlantic. Analyses based on observational data and numerical model experiments demonstrate that the two recently identified Atlantic Niño types, central and eastern Atlantic Niño (CAN and EAN), have distinct impacts on the sea ice concentration in West Antarctica. The CAN stimulates two atmospheric Rossby wave trains in the Southern Hemisphere through both direct and indirect pathways, collectively strengthening the Amundsen Sea Low. In contrast, the EAN only excites one atmospheric wave train over the South Pacific through an indirect pathway, due to its associated weaker local Hadley circulation, which fails to establish a significant Rossby wave source in the subtropical South Atlantic. Consequently, compared to the EAN, the atmospheric circulation and the associated sea ice concentration anomalies in West Antarctica during the CAN are stronger and more extensive. Therefore, distinguishing between the two Atlantic Niño types could potentially enhance the seasonal prediction capabilities for sea ice concentration in West Antarctica.

Seasonal Prediction of Ozone Pollution in Central-East China Using Machine Learning

Seasonal Prediction of Ozone Pollution in Central-East China Using Machine Learning

Seasonal prediction of Indian summer monsoon extreme rainfall frequency

Skillful forewarning of daily extreme rainfall activity (ERA) is imperative for adaptation against disastrous threats of socio-economic loss from Indian monsoon extreme rainfall events (ERE). Yet, unlike tropical cyclone (TC) activity forecasting, no attempt has been made for seasonal prediction of Indian monsoon ERE frequency and ERA. Here, we establish that the seasonal prediction of ERE frequency during Indian monsoon is associated with the global El Niño-Southern Oscillation (G-ENSO) in a manner similar to the Indian Summer Monsoon Rainfall (ISMR). We develop a deep learning model trained on the physical relationship between seasonal frequency of ERE and G-ENSO from an ensemble of Atmosphere-Ocean General Circulation Models (AOGCMs) for skillful seasonal forecast of ERE frequency at one-month lead. Integrating such seasonal forecasts of ERE frequency with ISMR seasonal forecast system is likely to be critical in disaster preparedness and loss minimization against increasing threat of ERE frequency damages in coming decades.

Links between hail hazard and climate modes of variability across Australia

AbstractHailstorms are destructive and dangerous phenomena that can cause large losses, motivating better understanding of their occurrence. As climate modes of variability influence temperature and moisture and hence convective instability, they offer predictive skill for hail conditions. Here, we examine relationships between hail‐prone days across Australia and the El Niño–Southern Oscillation (ENSO), Indian Ocean Dipole (IOD), and Southern Annular Mode (SAM). Hail‐prone days were identified using a hail proxy applied to European Centre for Medium‐Range Weather Forecasts Reanalysis v5 (ERA5) data from 1979 to 2022. Hail‐prone day anomalies were correlated with strength‐of‐mode indices. Broad areas of the country's interior show increased hail‐prone days during La Niña, negative IOD, and positive SAM in spring. The relationship with IOD and SAM is significant in winter for Brisbane and to some extent for Sydney, reversing sign in summer. Anomalies increase over Western Australia's south during El Niño and positive IOD in spring. Our work highlights potential connections between climate modes and hail‐prone conditions, investigates meteorological factors behind the observed correlations, and helps us understand annual variability to improve seasonal prediction.

A merged machine learning model for seasonal climate prediction in China

A merged machine learning model for seasonal climate prediction in China

A Positive Indian Ocean Dipole Leads to an Indian Ocean Basin Mode that Favors the Development of La Niña the Following Year

AbstractInteractions among the El Niño‐Southern Oscillation, Indian Ocean Basin mode (IOB), and Indian Ocean Dipole (IOD) significantly impact global climate variability and seasonal predictions. Traditionally, positive IOD (pIOD) and IOB warming events are associated with El Niño, driven by its influence on the tropical Indian Ocean through Walker Circulation anomalies. Our findings enrich this framework, revealing that a pIOD without El Niño can independently trigger IOB warming, and both types of pIODs can induce La Niña events. While El Niño primarily forces IOB warming and subsequent La Niña development via the atmospheric bridge across the Maritime Continent, pIODs independent of El Niño influence IOB warming through oceanic dynamics, which further favors La Niña development in the following year. The NMEFC‐CESM model sensitivity experiments underscore the critical role of thermocline processes in this mechanism, dependent on the pIOD's temperature amplitude, offering vital insights for forecasting post‐IOD, IOB, and La Niña events.

Predictable Patterns of Seasonal Atmospheric River Variability Over North America During Winter

AbstractAtmospheric rivers (ARs) are elongated areas of pronounced atmospheric water vapor transport that play an important role in the hydrological cycle over North America during winter. We investigate the sources of winter seasonal AR predictability over North America using average predictability time (APT) analysis. The skill of seasonal AR frequency predictions, in dynamical model forecasts provided by the Seamless System for Prediction and Earth System Research, is nearly entirely attributable to three physically interpretable APT modes that together represent about 19% of the total seasonal AR frequency variance. These three modes represent the AR response to the El Niño‐Southern Oscillation, anthropogenic forcing and equatorial heating over the eastern flank of the western Pacific warm pool, respectively. We further show that these three modes, calculated from AR frequency, explain nearly all winter seasonal precipitation forecast skill over North America.

Seasonal predictability of tropical cyclone frequency over the western North Pacific by a large-ensemble climate model

We assessed the seasonal prediction skill of tropical cyclone (TC) frequency over the western North Pacific by the large-ensemble SINTEX-F dynamical system. Although the prediction skills were limited, the correlation skill for the June–August prediction issued in early May was statistically significant around Okinawa and Taiwan. Particularly, the high TC activity in summer 2018 was well predicted. We found that the 2018 positive Indian Ocean Dipole (IOD) contributed to the predictability by the dynamical prediction system: suppressed convection in the eastern tropical Indian Ocean enhanced divergent wind from the eastern tropical Indian Ocean to the Okinawa and Taiwan areas. This helped to generate low pressure in the target area, which was favorable to the TC activity. The IOD contributions to the predictability were also seen in the correlation analyses in 1982–2022 and some case studies in 1994 and 1998. This could be useful for actionable early warnings.

Effect of Rocky Mountains and Tibetan Plateau 1998 Spring Land Temperature on N. American and East Asian Summer Precipitation Anomalies

AbstractThis work follows up on the GEWEX/LS4P Phase I (LS4P‐I) experiments, a community effort highlighting the spring land surface temperature anomalies in the Tibetan Plateau (TP) as a useful source for subseasonal to seasonal (S2S) prediction of summer precipitation in global hot spot regions, particularly in East Asia and North America. This paper extends the investigation to both the US Rocky Mountain (RM) region and the TP, considering the 1998 summer drought/flood event in North America/East Asia, respectively, as a case study. A previously developed initialization method for land surface temperature/subsurface temperature (LST/SUBT) is used in the NCEP Global Forecast System, coupled with a land model, SSiB2 (GFS/SSiB2), to produce observed RM cold May temperature anomaly. Forward simulation yields June precipitation anomalies at five remote locations. Likewise, the TP warm May temperature anomaly also produces June precipitation anomalies at these five locations. The effects of RM (cold) and TP (warm) temperature anomalies are consistent in the US South Coastal regions and the south Yangtze River Basin, yielding 49% (42%) of observed drought and 34% (44%) of observed flood, respectively. These LST/SUBT effects in RM and TP induce a global large‐scale wave train linking North America with the TP, affecting the subtropical westerly jet and thereby modulating summer precipitation. Global SST effect is examined for comparison but does not yield statistically significant June precipitation anomalies in GFS/SSiB2. This study adds to evidence that high‐mountain LST effects in the RM and TP are first‐order sources of S2S precipitation predictability in summer months.

Role of evolving sea surface temperature modes of variability in improving seasonal precipitation forecasts

The value of improving longer-lead precipitation forecasting in the water-stressed, semi-arid western United States cannot be overstated, especially considering the severity and frequency of droughts that have plagued the region for much of the 21st century. Seasonal prediction skill of current operational forecast systems, however, remain insufficient for decision-making purposes across a variety of applications. To address this capability gap, we develop a seasonal forecasting system that leverages the long-term memory of leading global and basin-scale modes of sea surface temperature variability. This approach focuses on characterizing and capitalizing on the spatiotemporal evolution of predictor modes over multiple antecedent seasons, instead of the customary use of predictive information from just the current season. Another distinctive methodological feature is the incorporation of sources of predictability spanning multiple timescales, from interannual to decadal-multidecadal. An evaluation of the forecast system’s performance from cross-validation analyses demonstrates skill over core winter precipitation regions—California, Pacific Northwest, and the Upper Colorado River basin. The developed model exhibits superior skill compared to dynamical and statistical benchmarks in predicting winter precipitation. Experimental seasonal precipitation forecasts from the model have the potential to provide critical situational awareness guidance to stakeholders in the water resources, agriculture, and disaster preparedness communities.

Seasonal prediction and potential predictability of the northern tropical Atlantic SST anomalies in current coupled climate models

Seasonal prediction and potential predictability of the northern tropical Atlantic SST anomalies in current coupled climate models