Pub Date : 2026-01-19DOI: 10.1038/s41612-026-01325-8
Leonardo A. Clarke, Jhordanne J. Jones, Michael A. Taylor, Matthew St. Michael Williams, Tajay Edwards, Tannecia S. Stephenson
Sea surface temperatures (SSTs) in the tropical North Atlantic have historically served as reliable predictors of early-season rainfall across the Caribbean. In particular, rainfall onset has been linked to SSTs exceeding the convective threshold necessary to support deep convection. However, recent warming trends appear to have altered this relationship. Here, we show that although SSTs routinely exceed the convective threshold earlier in the season, early rainfall has not increased. This decoupling reflects a shift in the atmospheric state, with enhanced stability, evidenced by reduced convective available potential energy and increased convective inhibition, increasingly suppressing convection. Reduced rainfall results in a more persistent Caribbean Low-Level Jet (CLLJ), further inhibiting rainfall by promoting subsidence and dry air advection. Correlations indicate that dynamic atmospheric variables now explain a larger share of rainfall variability than absolute SSTs. These findings signal a regime shift in Caribbean rainfall dynamics and raise concerns about the declining utility of SST-based predictors under continued climate warming. These results have significant implications for seasonal forecasting and adaptation planning across Caribbean Small Island Developing States.
{"title":"Breaking the link: warming disrupts early-season rainfall predictability in the Caribbean","authors":"Leonardo A. Clarke, Jhordanne J. Jones, Michael A. Taylor, Matthew St. Michael Williams, Tajay Edwards, Tannecia S. Stephenson","doi":"10.1038/s41612-026-01325-8","DOIUrl":"https://doi.org/10.1038/s41612-026-01325-8","url":null,"abstract":"Sea surface temperatures (SSTs) in the tropical North Atlantic have historically served as reliable predictors of early-season rainfall across the Caribbean. In particular, rainfall onset has been linked to SSTs exceeding the convective threshold necessary to support deep convection. However, recent warming trends appear to have altered this relationship. Here, we show that although SSTs routinely exceed the convective threshold earlier in the season, early rainfall has not increased. This decoupling reflects a shift in the atmospheric state, with enhanced stability, evidenced by reduced convective available potential energy and increased convective inhibition, increasingly suppressing convection. Reduced rainfall results in a more persistent Caribbean Low-Level Jet (CLLJ), further inhibiting rainfall by promoting subsidence and dry air advection. Correlations indicate that dynamic atmospheric variables now explain a larger share of rainfall variability than absolute SSTs. These findings signal a regime shift in Caribbean rainfall dynamics and raise concerns about the declining utility of SST-based predictors under continued climate warming. These results have significant implications for seasonal forecasting and adaptation planning across Caribbean Small Island Developing States.","PeriodicalId":19438,"journal":{"name":"npj Climate and Atmospheric Science","volume":"53 1","pages":""},"PeriodicalIF":9.0,"publicationDate":"2026-01-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146006138","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-19DOI: 10.1038/s41612-026-01331-w
S. Abhik, Fabio A. Capitanio, B. N. Goswami, Alexander Farnsworth, Peter D. Clift, Dietmar Dommenget
{"title":"Author Correction: A brief history of Asian summer monsoon evolution in the Cenozoic era","authors":"S. Abhik, Fabio A. Capitanio, B. N. Goswami, Alexander Farnsworth, Peter D. Clift, Dietmar Dommenget","doi":"10.1038/s41612-026-01331-w","DOIUrl":"https://doi.org/10.1038/s41612-026-01331-w","url":null,"abstract":"","PeriodicalId":19438,"journal":{"name":"npj Climate and Atmospheric Science","volume":"1 1","pages":""},"PeriodicalIF":9.0,"publicationDate":"2026-01-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146033642","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-17DOI: 10.1038/s41612-026-01326-7
Bo Christiansen, Shuting Yang
The North Atlantic Oscillation (NAO) is a dominant mode of variability in the Northern Hemisphere with strong impacts on temperature, precipitation, and storminess. The predictive skill of the NAO on annual to decadal scales is therefore an important topic, which is often studied using (initialized) climate models. The temporal structure of a time series is closely related to its predictability, and on inter-annual time scales the observed NAO is frequently described to have power at 2–10 years and sometimes with a distinct peak around 8 years. However, the observational record is brief, and such estimates have high uncertainty. Here, we present a thorough study to address the following questions: (1) Is the winter mean NAO distinguishable from white noise? (2) Does the temporal structure of the NAO differ between observations and historical experiments with contemporary climate models (CMIP6)? To this end, we use a range of statistical tools in both the temporal and spectral domains: Power-spectra, wavelet-spectra, autoregressive models, and various well-known time series test statistics. Overall, for both the observed and modelled NAO we find little evidence to reject the null-hypothesis of white noise. For observations, the peak in the power spectrum near 8 years is, taken alone, significant in the period after 1950 but not before. However, considering the complete spectrum, significant peaks will often occur at some frequencies, even for white noise. The large CMIP6 multi-model ensemble is statistically very similar to an ensemble of similar size of white noise. These results suggest limited decadal predictability of the NAO.
{"title":"The winter mean NAO: white noise and predictability","authors":"Bo Christiansen, Shuting Yang","doi":"10.1038/s41612-026-01326-7","DOIUrl":"https://doi.org/10.1038/s41612-026-01326-7","url":null,"abstract":"The North Atlantic Oscillation (NAO) is a dominant mode of variability in the Northern Hemisphere with strong impacts on temperature, precipitation, and storminess. The predictive skill of the NAO on annual to decadal scales is therefore an important topic, which is often studied using (initialized) climate models. The temporal structure of a time series is closely related to its predictability, and on inter-annual time scales the observed NAO is frequently described to have power at 2–10 years and sometimes with a distinct peak around 8 years. However, the observational record is brief, and such estimates have high uncertainty. Here, we present a thorough study to address the following questions: (1) Is the winter mean NAO distinguishable from white noise? (2) Does the temporal structure of the NAO differ between observations and historical experiments with contemporary climate models (CMIP6)? To this end, we use a range of statistical tools in both the temporal and spectral domains: Power-spectra, wavelet-spectra, autoregressive models, and various well-known time series test statistics. Overall, for both the observed and modelled NAO we find little evidence to reject the null-hypothesis of white noise. For observations, the peak in the power spectrum near 8 years is, taken alone, significant in the period after 1950 but not before. However, considering the complete spectrum, significant peaks will often occur at some frequencies, even for white noise. The large CMIP6 multi-model ensemble is statistically very similar to an ensemble of similar size of white noise. These results suggest limited decadal predictability of the NAO.","PeriodicalId":19438,"journal":{"name":"npj Climate and Atmospheric Science","volume":"4 1","pages":""},"PeriodicalIF":9.0,"publicationDate":"2026-01-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145993498","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-16DOI: 10.1038/s41612-025-01269-5
Xin Wang, Jianda Chen, Juntao Yang, Jeff Adie, Simon See, Kalli Furtado, Chen Chen, Troy Arcomano, Romit Maulik, Wei Xue, Gianmarco Mengaldo
Accurate and efficient climate simulations are crucial for understanding Earth’s evolving climate. However, current general circulation models (GCMs) face challenges in capturing unresolved physical processes, such as cloud and convection. A common solution is to adopt cloud-resolving models, which provide more accurate results than the standard subgrid parametrization schemes typically used in GCMs. However, cloud-resolving models, also referred to as super parametrizations, remain computationally prohibitive. Hybrid modeling, which integrates deep learning with equation-based GCMs, offers a promising alternative but often struggles with long-term stability and accuracy issues. In this work, we find that water vapor oversaturation during condensation is a key factor compromising the stability of hybrid models. To address this, we introduce CondensNet, a novel neural network architecture that embeds a self-adaptive physical constraint to correct unphysical condensation processes. CondensNet effectively mitigates water vapor oversaturation, enhancing simulation stability while maintaining accuracy and improving computational efficiency compared to super-parametrization schemes. We integrate CondensNet into a GCM to form PCNN-GCM (Physics-Constrained Neural Network GCM), a hybrid deep learning framework designed for long-term stable climate simulations in real-world conditions, including ocean and land. PCNN-GCM represents a significant milestone in hybrid climate modeling, as it shows a novel way to incorporate physical constraints adaptively, paving the way for accurate, lightweight, and stable long-term climate simulations.
{"title":"CondensNet: enabling stable long-term climate simulations via hybrid deep learning models with adaptive physical constraints","authors":"Xin Wang, Jianda Chen, Juntao Yang, Jeff Adie, Simon See, Kalli Furtado, Chen Chen, Troy Arcomano, Romit Maulik, Wei Xue, Gianmarco Mengaldo","doi":"10.1038/s41612-025-01269-5","DOIUrl":"https://doi.org/10.1038/s41612-025-01269-5","url":null,"abstract":"Accurate and efficient climate simulations are crucial for understanding Earth’s evolving climate. However, current general circulation models (GCMs) face challenges in capturing unresolved physical processes, such as cloud and convection. A common solution is to adopt cloud-resolving models, which provide more accurate results than the standard subgrid parametrization schemes typically used in GCMs. However, cloud-resolving models, also referred to as super parametrizations, remain computationally prohibitive. Hybrid modeling, which integrates deep learning with equation-based GCMs, offers a promising alternative but often struggles with long-term stability and accuracy issues. In this work, we find that water vapor oversaturation during condensation is a key factor compromising the stability of hybrid models. To address this, we introduce CondensNet, a novel neural network architecture that embeds a self-adaptive physical constraint to correct unphysical condensation processes. CondensNet effectively mitigates water vapor oversaturation, enhancing simulation stability while maintaining accuracy and improving computational efficiency compared to super-parametrization schemes. We integrate CondensNet into a GCM to form PCNN-GCM (Physics-Constrained Neural Network GCM), a hybrid deep learning framework designed for long-term stable climate simulations in real-world conditions, including ocean and land. PCNN-GCM represents a significant milestone in hybrid climate modeling, as it shows a novel way to incorporate physical constraints adaptively, paving the way for accurate, lightweight, and stable long-term climate simulations.","PeriodicalId":19438,"journal":{"name":"npj Climate and Atmospheric Science","volume":"85 1","pages":""},"PeriodicalIF":9.0,"publicationDate":"2026-01-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145993494","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-16DOI: 10.1038/s41612-025-01316-1
Ajin Cho, Hajoon Song, Il-Ju Moon, Hyodae Seo, Rui Sun, Matthew R. Mazloff, Aneesh C. Subramanian, Bruce D. Cornuelle, Arthur J. Miller
Current–wind interaction modulates air–sea momentum and turbulent heat fluxes, which are critical in the energy cycle of tropical cyclones (TCs). However, the effects of the surface currents on air–sea exchange under TCs have remained unclear. Here, using an atmosphere–ocean coupled model, we investigate the role of current–wind interaction in determining TC intensity. Surface currents generally align with surface winds. Accounting for the current–wind interaction, the alignment reduces both the air–sea turbulent heat flux and momentum flux (average 1.0% and 2.5%), which serve as the energy source and sink of TCs, respectively. The reduction in the energy source (sink) decreases (increases) the TC growth −1.9% (+1.3%) on average and up to −13.7% (+11.1%). For simulations extending beyond the seasonal scale, the accumulated impacts of current–wind interaction alter TC genesis, affecting surface wind speed and sea surface temperature during the TC season. These findings reveal an important feedback mechanism associated with TCs driven by the current–wind interaction.
{"title":"Modulation of tropical cyclone intensity by current–wind interaction","authors":"Ajin Cho, Hajoon Song, Il-Ju Moon, Hyodae Seo, Rui Sun, Matthew R. Mazloff, Aneesh C. Subramanian, Bruce D. Cornuelle, Arthur J. Miller","doi":"10.1038/s41612-025-01316-1","DOIUrl":"https://doi.org/10.1038/s41612-025-01316-1","url":null,"abstract":"Current–wind interaction modulates air–sea momentum and turbulent heat fluxes, which are critical in the energy cycle of tropical cyclones (TCs). However, the effects of the surface currents on air–sea exchange under TCs have remained unclear. Here, using an atmosphere–ocean coupled model, we investigate the role of current–wind interaction in determining TC intensity. Surface currents generally align with surface winds. Accounting for the current–wind interaction, the alignment reduces both the air–sea turbulent heat flux and momentum flux (average 1.0% and 2.5%), which serve as the energy source and sink of TCs, respectively. The reduction in the energy source (sink) decreases (increases) the TC growth −1.9% (+1.3%) on average and up to −13.7% (+11.1%). For simulations extending beyond the seasonal scale, the accumulated impacts of current–wind interaction alter TC genesis, affecting surface wind speed and sea surface temperature during the TC season. These findings reveal an important feedback mechanism associated with TCs driven by the current–wind interaction.","PeriodicalId":19438,"journal":{"name":"npj Climate and Atmospheric Science","volume":"180 1","pages":""},"PeriodicalIF":9.0,"publicationDate":"2026-01-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145968805","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-16DOI: 10.1038/s41612-026-01322-x
Yan Yan, Jia Wu, Qingchen Chao, Ying Sun
In this study, the variations in near-surface wind speed (SWS) across China were analysed using 55 years of observational data from 2044 meteorological stations spanning 1970–2024. The results indicate that the SWS in China experienced a persistent decline from 1970 to 2004, then remained constant from 2005 to 2014 and shifted to recovery after 2015. The seasonal and spatial analyses confirmed that the slowdown period occurred nearly nationwide, whereas the apparent recovery displayed a strong spatial heterogeneity. On the basis of long-term station records and metadata from China, station-by-station analysis revealed that SWS records are strongly influenced by widespread relocations; 1237 stations (60.5% of the total stations) with SWS discontinuities were related to relocations, leading to an overestimation of the so-called “recovery phase” at national and regional scales. This emphasized the importance of homogenization to ensure the reliability of the SWS dataset. Afterwards, the SWS breakpoints were detected and adjusted. After homogenization, the SWS displayed slight but significantly negative trends (–0.06 m s⁻¹ per decade) after 2005. Representative non-relocated stations were used to validate the homogenized results, and revealed that the homogenized series accurately captured more robust signal from non-relocated stations, which confirmed the consistent declining trends across China and further implied the effect of station relocation. Notably, compared with the original records, the reduction in the homogenized SWS ranged from 3% to 11% among the subregions. This could substantially impact on the assessment of wind energy resources and requires careful attention.
本文利用1970-2024年中国2044个气象站55年的观测资料，分析了中国近地面风速的变化。结果表明：1970 - 2004年中国SWS持续下降，2005 - 2014年保持不变，2015年后开始回升；季节和空间分析证实，减缓期几乎发生在全国范围内，而明显的恢复表现出强烈的空间异质性。基于长期台站记录和中国的元数据，逐站分析表明，SWS记录受到大范围迁移的强烈影响；1237个台站（占总台站的60.5%）的SWS不连续与迁移有关，导致在国家和区域尺度上高估了所谓的“恢复阶段”。这强调了均匀化对确保SWS数据集可靠性的重要性。之后，检测并调整SWS断点。均一化后，SWS在2005年后呈现轻微但显著的负趋势（-0.06 m s / 10年）。利用代表性的非搬迁台站对均质化结果进行验证，发现均质化序列更准确地捕获了来自非搬迁台站的更强信号，这证实了中国各地的持续下降趋势，并进一步暗示了搬迁台站的影响。值得注意的是，与原始记录相比，各次区域均一化SWS的减少幅度在3%至11%之间。这可能对风能资源的评估产生重大影响，需要仔细注意。
{"title":"Overestimation of the recent observed near-surface wind speed recovery in China","authors":"Yan Yan, Jia Wu, Qingchen Chao, Ying Sun","doi":"10.1038/s41612-026-01322-x","DOIUrl":"https://doi.org/10.1038/s41612-026-01322-x","url":null,"abstract":"In this study, the variations in near-surface wind speed (SWS) across China were analysed using 55 years of observational data from 2044 meteorological stations spanning 1970–2024. The results indicate that the SWS in China experienced a persistent decline from 1970 to 2004, then remained constant from 2005 to 2014 and shifted to recovery after 2015. The seasonal and spatial analyses confirmed that the slowdown period occurred nearly nationwide, whereas the apparent recovery displayed a strong spatial heterogeneity. On the basis of long-term station records and metadata from China, station-by-station analysis revealed that SWS records are strongly influenced by widespread relocations; 1237 stations (60.5% of the total stations) with SWS discontinuities were related to relocations, leading to an overestimation of the so-called “recovery phase” at national and regional scales. This emphasized the importance of homogenization to ensure the reliability of the SWS dataset. Afterwards, the SWS breakpoints were detected and adjusted. After homogenization, the SWS displayed slight but significantly negative trends (–0.06 m s⁻¹ per decade) after 2005. Representative non-relocated stations were used to validate the homogenized results, and revealed that the homogenized series accurately captured more robust signal from non-relocated stations, which confirmed the consistent declining trends across China and further implied the effect of station relocation. Notably, compared with the original records, the reduction in the homogenized SWS ranged from 3% to 11% among the subregions. This could substantially impact on the assessment of wind energy resources and requires careful attention.","PeriodicalId":19438,"journal":{"name":"npj Climate and Atmospheric Science","volume":"57 1","pages":""},"PeriodicalIF":9.0,"publicationDate":"2026-01-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145993496","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Over the past four decades, global tropical cyclogenesis (TCG) frequency has exhibited no trend related to strong interannual variability and uneven regional TCG trends, despite pronounced global sea surface temperature (SST) increases. This study offers new evidence explaining the disparity between the global TCG trend and global warming. Here we show that just three global tropics-wide SST patterns can account for TCG interannual variability in most global ocean areas where TCG is climatologically active. We further find that the warming mode of tropical SSTs explains 77% of global ocean areas observed with significant TCG trends, with the strongest influence in the tropics with robust observed TCG trends. These warming-related TCG trends are associated with a reorganization of the Pacific and Atlantic Walker circulations, which modulates environmental conditions governing regional TCG. These findings reveal how global warming influences regional TCG in an uneven manner, consistent with the mechanisms underpinning interannual variability.
{"title":"Tropical SST modes provide a unified explanation for global tropical cyclogenesis changes","authors":"Jiangyu Li, Xiangbo Feng, Shaoqing Zhang, Ralf Toumi, Jing Xu","doi":"10.1038/s41612-026-01323-w","DOIUrl":"https://doi.org/10.1038/s41612-026-01323-w","url":null,"abstract":"Over the past four decades, global tropical cyclogenesis (TCG) frequency has exhibited no trend related to strong interannual variability and uneven regional TCG trends, despite pronounced global sea surface temperature (SST) increases. This study offers new evidence explaining the disparity between the global TCG trend and global warming. Here we show that just three global tropics-wide SST patterns can account for TCG interannual variability in most global ocean areas where TCG is climatologically active. We further find that the warming mode of tropical SSTs explains 77% of global ocean areas observed with significant TCG trends, with the strongest influence in the tropics with robust observed TCG trends. These warming-related TCG trends are associated with a reorganization of the Pacific and Atlantic Walker circulations, which modulates environmental conditions governing regional TCG. These findings reveal how global warming influences regional TCG in an uneven manner, consistent with the mechanisms underpinning interannual variability.","PeriodicalId":19438,"journal":{"name":"npj Climate and Atmospheric Science","volume":"42 1","pages":""},"PeriodicalIF":9.0,"publicationDate":"2026-01-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145993485","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-15DOI: 10.1038/s41612-025-01318-z
Min Sothearith, Daeryong Park, Kuk-Hyun Ahn
Extreme precipitation (EP) is a major climate risk, yet its projections remain uncertain due to the combined influence of thermodynamic (TH) and dynamic (DY) processes. Using multi-model simulations under three emission scenarios, we separate TH and DY contributions to the annual maximum 1-day precipitation (Rx1Day) and quantify their uncertainties. TH consistently intensifies extremes with warming, while DY strongly modulates their magnitude and direction. DY processes dominate Rx1Day uncertainty, with internal variability within DY emerging as the leading contributor. Signal-to-noise ratio (SNR) analysis shows that the forced signal emerges more clearly for TH than DY, where chaotic variability fundamentally limits predictability. The strongest intensification occurs in equatorial regions, raising equity concerns for vulnerable populations. These results demonstrate that DY internal variability is the primary driver of EP uncertainty, highlighting limits to long-term predictability and the importance of properly representing natural dynamical fluctuations in future projections.
{"title":"Dynamic internal variability dominates uncertainty in modeling future extreme precipitation","authors":"Min Sothearith, Daeryong Park, Kuk-Hyun Ahn","doi":"10.1038/s41612-025-01318-z","DOIUrl":"https://doi.org/10.1038/s41612-025-01318-z","url":null,"abstract":"Extreme precipitation (EP) is a major climate risk, yet its projections remain uncertain due to the combined influence of thermodynamic (TH) and dynamic (DY) processes. Using multi-model simulations under three emission scenarios, we separate TH and DY contributions to the annual maximum 1-day precipitation (Rx1Day) and quantify their uncertainties. TH consistently intensifies extremes with warming, while DY strongly modulates their magnitude and direction. DY processes dominate Rx1Day uncertainty, with internal variability within DY emerging as the leading contributor. Signal-to-noise ratio (SNR) analysis shows that the forced signal emerges more clearly for TH than DY, where chaotic variability fundamentally limits predictability. The strongest intensification occurs in equatorial regions, raising equity concerns for vulnerable populations. These results demonstrate that DY internal variability is the primary driver of EP uncertainty, highlighting limits to long-term predictability and the importance of properly representing natural dynamical fluctuations in future projections.","PeriodicalId":19438,"journal":{"name":"npj Climate and Atmospheric Science","volume":"49 1","pages":""},"PeriodicalIF":9.0,"publicationDate":"2026-01-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145968806","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-15DOI: 10.1038/s41612-026-01320-z
Lluís Palma, Alejandro Peraza, David Civantos-Prieto, Amanda Duarte, Stefano Materia, Ángel G. Muñoz, Jesús Peña-Izquierdo, Laia Romero, Albert Soret, Markus G. Donat
Most operational climate services providers base their seasonal predictions on initialised general circulation models (GCMs) or empirical statistical techniques. GCMs are widely used but require substantial computational resources, limiting their capacity. In contrast, statistical methods often lack robustness due to the short historical records available. Recent works propose machine learning methods trained on climate model output, leveraging larger sample sizes. Yet, many of these studies focus on prediction tasks that may be restricted in spatial or temporal extent, thereby creating a gap with existing operational predictions. Others fail to disentangle the sources of skill in the context of climate change, where strong trends provide spurious estimates. This study combines variational inference with transformers to predict global and regional seasonal anomalies of temperature and rainfall. The model is trained on output from CMIP6 and tested using ERA5 reanalysis data. Temperature predictions demonstrate skill beyond the climatology and climate-change trend and even outperform the numerical state-of-the-art system SEAS5 in some ocean and land areas. Precipitation forecasts show more limited skill, with fewer regions outperforming climatology and fewer surpassing SEAS5. Furthermore, the consistency found in both teleconnections and skill spatial patterns against SEAS5 suggests that both systems build on similar sources of predictability.
{"title":"Data-driven seasonal climate predictions via variational inference and transformers","authors":"Lluís Palma, Alejandro Peraza, David Civantos-Prieto, Amanda Duarte, Stefano Materia, Ángel G. Muñoz, Jesús Peña-Izquierdo, Laia Romero, Albert Soret, Markus G. Donat","doi":"10.1038/s41612-026-01320-z","DOIUrl":"https://doi.org/10.1038/s41612-026-01320-z","url":null,"abstract":"Most operational climate services providers base their seasonal predictions on initialised general circulation models (GCMs) or empirical statistical techniques. GCMs are widely used but require substantial computational resources, limiting their capacity. In contrast, statistical methods often lack robustness due to the short historical records available. Recent works propose machine learning methods trained on climate model output, leveraging larger sample sizes. Yet, many of these studies focus on prediction tasks that may be restricted in spatial or temporal extent, thereby creating a gap with existing operational predictions. Others fail to disentangle the sources of skill in the context of climate change, where strong trends provide spurious estimates. This study combines variational inference with transformers to predict global and regional seasonal anomalies of temperature and rainfall. The model is trained on output from CMIP6 and tested using ERA5 reanalysis data. Temperature predictions demonstrate skill beyond the climatology and climate-change trend and even outperform the numerical state-of-the-art system SEAS5 in some ocean and land areas. Precipitation forecasts show more limited skill, with fewer regions outperforming climatology and fewer surpassing SEAS5. Furthermore, the consistency found in both teleconnections and skill spatial patterns against SEAS5 suggests that both systems build on similar sources of predictability.","PeriodicalId":19438,"journal":{"name":"npj Climate and Atmospheric Science","volume":"1 1","pages":""},"PeriodicalIF":9.0,"publicationDate":"2026-01-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145968807","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-14DOI: 10.1038/s41612-025-01280-w
Fei Ji, Moutassem El Rafei, Giovanni Di Virgilio, Jason P. Evans, Jatin Kala, Stephen White, Julia Andrys, Dipayan Choudhury, Eugene Tam, Yue Li, Rishav Goyal, Carlos Vieira Rocha, Matthew L. Riley
Regional climate simulations provide essential high-resolution information for climate services. This study evaluates future changes in mean climate and 10 extremes using three generations of the NARCliM (NSW and Australian Regional Climate Modelling) project, which downscale CMIP3, CMIP5, and CMIP6 models. Projections show statistically significant increases in maximum and minimum temperatures across all NARCliM generations, with consistent spatial patterns. The magnitude of warming is primarily influenced by driving GCMs and emissions scenarios. In contrast, precipitation projections exhibit greater variability between generations, reflecting model and scenario differences and underscoring the challenge of projecting future precipitation. Extreme heat indices are projected to increase across Australia, with consistent spatial patterns and stronger changes under higher emissions, indicating more frequent and severe extreme heat events. Precipitation extremes display more variability across regions, model generations, and scenarios, although certain trends are robust. The intensity of very extreme rainfall (above the 99th percentile) is projected to increase, as is the maximum length of dry spells. Conversely, the maximum length of wet spells and the number of heavy rain days are expected to decrease. NARCliM2.0 specifically suggests shorter wet periods and fewer heavy rain days, but more intense extreme rainfall. These findings demonstrate the relative robustness of temperature and its extremes compared to precipitation and emphasize the value of broader GCM ensembles in future downscaling efforts to improve confidence in regional projections.
{"title":"Three generations of NARCliM: future projections of mean and extreme climate over the CORDEX Australasia domain","authors":"Fei Ji, Moutassem El Rafei, Giovanni Di Virgilio, Jason P. Evans, Jatin Kala, Stephen White, Julia Andrys, Dipayan Choudhury, Eugene Tam, Yue Li, Rishav Goyal, Carlos Vieira Rocha, Matthew L. Riley","doi":"10.1038/s41612-025-01280-w","DOIUrl":"https://doi.org/10.1038/s41612-025-01280-w","url":null,"abstract":"Regional climate simulations provide essential high-resolution information for climate services. This study evaluates future changes in mean climate and 10 extremes using three generations of the NARCliM (NSW and Australian Regional Climate Modelling) project, which downscale CMIP3, CMIP5, and CMIP6 models. Projections show statistically significant increases in maximum and minimum temperatures across all NARCliM generations, with consistent spatial patterns. The magnitude of warming is primarily influenced by driving GCMs and emissions scenarios. In contrast, precipitation projections exhibit greater variability between generations, reflecting model and scenario differences and underscoring the challenge of projecting future precipitation. Extreme heat indices are projected to increase across Australia, with consistent spatial patterns and stronger changes under higher emissions, indicating more frequent and severe extreme heat events. Precipitation extremes display more variability across regions, model generations, and scenarios, although certain trends are robust. The intensity of very extreme rainfall (above the 99th percentile) is projected to increase, as is the maximum length of dry spells. Conversely, the maximum length of wet spells and the number of heavy rain days are expected to decrease. NARCliM2.0 specifically suggests shorter wet periods and fewer heavy rain days, but more intense extreme rainfall. These findings demonstrate the relative robustness of temperature and its extremes compared to precipitation and emphasize the value of broader GCM ensembles in future downscaling efforts to improve confidence in regional projections.","PeriodicalId":19438,"journal":{"name":"npj Climate and Atmospheric Science","volume":"24 1","pages":""},"PeriodicalIF":9.0,"publicationDate":"2026-01-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145968809","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：