Atmospheric emissions pose significant challenges to achieving the Sustainable Development Goals (SDGs). As key sources of emissions, cities are also critical actors in driving sustainable transitions. Their performance on emission-related SDG indicators is shaped not only by local factors but also by inter-city interactions. However, the spatial and temporal heterogeneity of these interactions remain poorly understood. Here, we investigate inter-city SDG interactions across 4,116 cities worldwide from 2010 to 2020, and simulate SDG performance under five interaction scenarios, including isolated and collaborative conditions. Results reveal that a 10% improvement in SDG 11.6 (particulate matter), SDG 12.4 (gaseous pollutants), and SDG 13.2 (greenhouse gases) in other cities can boost the focal city's performance by 0.450%, 0.390%, and 0.200%, respectively. Most cities (93.51%) are influenced by other cities. Inter-city coordination can improve SDG performance by 10.81% compared to isolated efforts. These findings highlight the importance of coordinated strategies to enhance urban sustainability.

Subseasonal-to-seasonal (S2S) forecasts play an essential role in providing a decision-critical weeks-to-months planning window for climate resilience and sustainability, yet a growing bottleneck is the last-mile gap: translating scientific forecasts into trusted, actionable climate services, requiring reliable multimodal understanding and decision-facing reasoning under uncertainty. Meanwhile, multimodal large language models (MLLMs) and corresponding agentic paradigms have made rapid progress in supporting various workflows, but it remains unclear whether they can reliably generate decision-making deliverables from operational service products (e.g., actionable signal comprehension, decision-making handoff, and decision analysis planning) under uncertainty. We introduce S2SServiceBench, a multimodal benchmark for last-mile S2S climate services curated from an operational climate-service system to evaluate this capability. S2SServiceBenchcovers 10 service products with about 150+ expert-selected cases in total, spanning six application domains - Agriculture, Disasters, Energy, Finance, Health, and Shipping. Each case is instantiated at three service levels, yielding around 500 tasks and 1,000+ evaluation items across climate resilience and sustainability applications. Using S2SServiceBench, we benchmark state-of-the-art MLLMs and agents, and analyze performance across products and service levels, revealing persistent challenges in S2S service plot understanding and reasoning - namely, actionable signal comprehension, operationalizing uncertainty into executable handoffs, and stable, evidence-grounded analysis and planning for dynamic hazards-while offering actionable guidance for building future climate-service agents.

The Tibetan Plateau (TP), also known as the "Water Tower of Asia," profoundly impacts regional and global climates. Existing climate models show substantial biases over the TP, primarily due to low spatial resolution, deficient driving data, and inadequate model domains. Leveraging meteorological station data, the CN05.1 meteorological dataset, and reanalysis data-sets, we comprehensively evaluate the performance of the Weather Research and Forecasting Model at gray-zone resolution (9 km) (WRF9km) in simulating TP air temperature and precipitation during 1980-2019 and identify bias causes. WRF9km effectively captures observed spatial air temperature patterns but shows a cold bias. This is mainly due to overestimated surface albedo (accounting for 64%), along with underestimated downward radiation and ground heat fluxes. WRF9km also simulates observed spatial precipitation patterns well, with significant seasonal correlations; however, precipitation biases exhibit pronounced spatial heterogeneity. Compared to CN05.1, precipitation is overestimated over the southern and eastern slopes of the TP, while it is underestimated over the interior TP, especially in the western part. These biases primarily arise from inadequate characterization of wind field dynamics and moisture transport within the model framework. Meanwhile, WRF9km captures annual precipitation variation with minor deviations, although it does not fully reproduce the temporal trends over the TP. Overall, compared to driving ERA5 data, WRF9km yields only marginal improvement for the cold bias but reduces the regional mean precipitation wet bias by 79%, particularly over the southern and eastern slopes. This evaluation provides critical insights for dynamic downscaling in complex terrain, highlighting the need for improved surface albedo parameterizations and higher-quality input data.

Abstract Atmospheric rivers (ARs) constitute a global, interconnected highway network rather than isolated regional events. In boreal summer, cross‐Pacific ARs originate over Southeast Asia, are fueled by subtropical outflows from the Asian monsoon plume, transport warm, moist air across the North Pacific, and make landfall in North America (NA). Our results show that diabatic heating anomalies over the Indian summer monsoon region and the Philippine Sea–Western North Pacific area jointly modulate AR pathways and landfalls. Numerical experiments verify that distinct heating archetypes generate diverse downstream triple‐pressure circulation structures, steering ARs toward different landfall locations. Cross‐Pacific AR activity is also modulated by climate oscillations; different phases preferentially induce distinct Indo‐Pacific heating patterns and thus redirect AR pathways. Therefore, tropical heating anomalies in Indo‐Western Pacific are valuable predictors of boreal‐summer AR activity. The interconnected “AR highways” linking Asian climate to NA and extend the predictability beyond Asia to the broader Pacific Rim.

Significant winter [December-February (DJF)] precipitation over southern China (SC) is one of the key features of the East Asian winter monsoon, accounting for nearly 20% of annual precipitation in the area. While oceanic drivers of its interannual variability are extensively studied, the influence of atmospheric rivers (ARs), contributing approximately 30%-40% of the climatological wintertime precipitation in SC, remains unclear. Additionally, how seasonal forecast models capture the impact of tropical sea surface temperature (SST) variations on winter precipitation through ARs requires further investigation using objective metrics. This study identifies a tropical SST pattern involving El Ni & ntilde;o-Southern Oscillation (ENSO), the Indian Ocean dipole, and the SST anomalies over the western North Pacific (WNP), whose coevolving structure rapidly develops from the preceding summer to winter. This anomalous SST configuration generates a hemispheric-scale circulation pattern from the tropics to the subtropics, which enhances vertical wind shear and meridional moisture transport over SC, favoring increased AR intrusion into the region. Consequently, significant precipitation anomalies occur particularly near SC along 20 degrees-30 degrees N, explaining over 50% of the interannual DJF precipitation variability. These ENSO-driven precipitation changes, mediated by AR activity, are reasonably predicted by two operational seasonal forecast systems, suggesting that ENSO and its interaction with WNP SST anomalies serve as the primary sources of forecast skill for winter ARs and SC precipitation. Furthermore, a screening scheme based on the observed SST and circulation states during October and November preceding the target winter is developed to determine the years in which the dynamical model forecast skill for SC DJF precipitation is higher than in other winters.

The linkage between atmospheric rivers (ARs) and humid heat (HH) remains a largely unexplored research domain. This study detected 834 humid heat days (HHDs) in eastern China from 1979 to 2018, with above half of the HHDs coinciding with ARs during the boreal summer. Compared to HH not associated with ARs, those related to ARs exhibit a broader impact area and greater severity. The western North Pacific subtropical high (WNPSH)-affected and low pressure system (LPS)-affected HH types are identified to illustrate the mechanisms, and the North rainfall-South HH Dipole exists in both types. The location of WNPSH-affected HH synchronizes with the shift of WNPSH, while the LPS-affected HH is situated near the low-level LPS. Regardless of AR presence, the HH are governed by the 500 hPa geopotential height (Z500) high-pressure system, with solar radiation acting as the dominant driver of amplified surface warming. AR-related HH is further marked by a stronger 500 hPa height anomaly and reduced cloud cover, allowing more solar radiation to reach the surface, while its higher humidity is likely associated with stronger moisture transport and enhanced local evaporation. Recognizing this AR-HH co-occurrence is essential for monitoring, predicting and adapting to the most impactful HH events over eastern China.

What: Climate scientists, forecasters, and stakeholders from governmental and private sectors from 12 countries gathered at this annual conference to address climate change, extreme weather, and water challenges. This year's forum featured the themes of seamless prediction on the subseasonal to seasonal time scale, artificial intelligence (AI) in Earth science, and atmospheric rivers, advancing global efforts for a climate-resilient future. Where: The Hong Kong University of Science and Technology (HKUST), Hong Kong, China

Atmospheric rivers (ARs) and marine heatwaves (MHWs) are two major extreme events of the climate system that strongly influence the ocean-atmosphere interface, yet their mutual interactions remain poorly understood. Here we use long-term oceanic and atmospheric reanalysis datasets (OISST and ERA5) from 1982 to 2023 to quantify the interactions and feedbacks between ARs and MHWs over the North Pacific. Longer and more intense events exhibit a higher probability of overlapping, resulting in nearly 85% of ARs and 57% of MHWs being linked to the other system. Pronounced hotspots of co-occurrence emerge in the mid-latitudes, where both systems frequently develop. ARs promote ocean surface warming and exacerbate MHW intensity by enhancing surface heat fluxes dominated by latent heat, together with increased downward longwave and sensible heat fluxes over the North Pacific north of 40 degrees N. Conversely, MHWs slightly suppress local AR intensity by weakening horizontal winds, while mesoscale convection and cyclonic disturbances disrupt integrated vapor transport. This effect is partially offset by enhanced moisture associated with stronger convection. These findings reveal a bidirectional coupling between oceanic and atmospheric extremes and highlight the need to examine their relationship across other ocean basins and under future warming scenarios to better anticipate compound climate risks.

Subseasonal predictions from 2 weeks to 2 months have made significant advancements over the past decade, driven by progress in physical understanding, climate modeling, computational capabilities, and artificial intelligence (AI). These predictions are increasingly in demand due to their potential to provide stakeholders with adequate lead time for effective disaster adaptation, mitigation, and resource management. However, there remain critical gaps in the engagement between prediction providers and service users. Providers often lack insight into the specific needs of users and do not have transferrable strategies to build trust through tailored evaluations and clear confidence levels, which often results in repeatedly devising approaches for each provider–user interaction. Further, users frequently struggle to interpret predictions and are hesitant to make decisions based on these uncertain outcomes. This paper attempts to make “last-mile efforts” by reviewing relevant literature, operational systems, and the most informative communications and engagement strategies with key sectors. It proposes a preliminary framework to standardize the approach for provider–user interaction in the context of subseasonal prediction and services, with potential applicability and extension to seamless prediction systems in the future. Lastly, we underscore future directions for subseasonal predictions, emphasizing the integration of dynamic climate modeling and AI-driven enhancements with large ensemble techniques to improve both reliability and confidence. This review is part of the United Nations Educational Scientific and Cultural Organization (UNESCO) International Decade of Sciences for Sustainable Development (2024–33) and contributes to the Seamless Prediction and Services for Sustainable Natural and Built Environment (SEPRESS) Program (2025–32), an initiative endorsed under this global framework.

The Madden-Julian Oscillation (MJO) is a key source of predictability for subseasonal-to-seasonal (S2S) forecasts, with important implications for early warning of high-impact weather and related disaster-risk reduction. However, in most S2S models, including IAP-CAS v1.3, the ensemble spread remains insufficient to capture atmospheric uncertainty, thereby limiting MJO forecast skill. An enhanced version of the IAP-CAS model has been developed to improve MJO forecasts by incorporating the Second-Order Exact Sampling (SOES) method into the initialization process, using large historical samples to extract the leading modes of uncertainty and generate physically consistent perturbations along the primary error-growth pathways with minimal computational cost. Based on selected MJO events during the winters of 2019–2023, a series of sensitivity experiments were designed to optimize the ensemble generation strategy. As a result, the upgraded model achieved an improvement in MJO forecast skill of up to 6 days. This improvement in MJO forecast skill is primarily attributed to a more realistic representation of the MJO moisture mode and background temperature stratification, leading to more accurate simulations of MJO-related convection and a demonstrable impact on precipitation forecasts over China. On one hand, the improved MJO forecasts enhance the forecasts of the position of the Western Pacific Subtropical High (WPSH), thereby increasing the accuracy of precipitation forecasts over Southern China. On the other hand, Rossby wave signals triggered by the MJO propagate into the mid- and high-latitudes, contributing to improved precipitation forecasts over central and northern China. This study underscores the importance of a well-designed ensemble generation strategy tailored to the model for S2S forecasts and reinforces the necessity of improving MJO forecast skill to support reliable forecasts of downstream climate systems at S2S timescales. However, the current SOES implementation remains an initial step, as the temporally static perturbations during nudging limit its ability to represent flow-dependent error growth, indicating the need for dynamically evolving perturbations to better regulate ensemble dispersion.

Cold regions are vital to the Earth system, influencing water storage, energy balance, and ecological stability. China's diverse terrain includes extensive cold regions that are shrinking due to global warming, with profound implications for climate resilience. Despite their importance, comprehensive assessments of these regions' past trends and future projections are lacking. This study employs an array of data, including gridded daily dataset (CN05.1), China Meteorological Forcing Dataset (CMFD), European Centre for Medium-Range Weather Forecasts (ECMWF) Reanalysis v5-Land (ERA5-Land), and original and bias-corrected outputs from Coupled Model Intercomparison Project Phase 6 (CMIP6) global climate models (GCMs), to analyze the historical (1979-2016) and projected (2015-2100) dynamics of China's cold regions under different emission scenarios. Our findings indicate that from 1979 to 2016, China's cold regions covered an average of 3.87 million km2, or 40.4 % of China's land area, with the most substantial areas located in the Tibetan Plateau, Tianshan and Pamirs mountain ranges, and the northeastern region. These regions experienced a significant decline, at a rate of 154,000 km2 per decade. Future projections based on bias-corrected CMIP6 data suggest an accelerating contraction, with estimates ranging from a decrease of 54,000 km2 per decade under the SSP126 scenario to 210,000 km2 per decade under the SSP585 scenario. By the end of the 21st century, the cold regions' extent could diminish by more than 50 % under the high-emission SSP585 scenario compared to the historical baseline. The analysis shows that nearly 95 % of variations in the extent of China's cold regions can be attributed to shifts in areas with an annual average temperature at or below 5 degrees C. These results underscore the need for urgent climate adaptation strategies, particularly as China's cold regions continue to shrink in response to climate warming driven by human-induced emissions.

Purely data-driven deep learning (DL) models show superior performance in runoff simulation but suffer from a critical training-projection gap, producing physically implausible results under significantly altered climate conditions. This study develops a physics informed Long Short-Term Memory (LSTM) framework that structurally embeds dynamic water balance and evapotranspiration modules to enforce physical consistency. Using the Pearl River Basin (PRB)-China's second largest by discharge-and three additional subtropical basins as testbeds, we systematically assess the prevalence of this gap and demonstrate the framework's ability to overcome it. Results show that standard LSTM models, which are purely data-driven, yield anomaly ratios of 61-100% under warming scenarios, often misinterpreting temperature rise as a runoff-generating signal. In contrast, the hybrid framework eliminates anomalous projections while maintaining or improving accuracy relative to process-based and purely data-driven models. The hybrid framework also corrects systematic biases in the PRB under CMIP6 scenarios: standard LSTMs erroneously amplify flood risks (overestimating up to 78 events per decade) and suppress drought signals (underestimating up to 9 events per decade) relative to the physically consistent benchmark provided by the hybrid DL. The model further highlights wet-season intensification in the PRB, with up to 30% increases in annual runoff but sharper seasonal contrasts. These basin-and sub-basin-level insights provide a more reliable scientific basis for climate adaptation strategies. Overall, the proposed framework offers a flexible, physics-encoded architecture that bridges the training-projection gap and advances DL toward more trustworthy hydrological projections under climate change.

Reliable description of the inflow boundary conditions of the fully developed turbulence in atmospheric boundary layers for large-eddy simulation (LES) has been a subject of significant research interest due to its considerable impact on simulation results. In existing synthetic random Fourier methods for generating inhomogeneous inflow turbulence, the inhomogeneous field can be obtained by superimposing multiple homogeneous turbulence fields, each weighted by an assigned weighting function. However, inappropriate mapping strategies may distort the spatial correlations and continuity of turbulence structures from the targeted values, which could further deteriorate the simulation results. In this study, the impacts of the weighting functions on the total-energy profiles, spatial correlations, and the continuity condition of turbulence are clarified, based on which an adaptive energypreserving mapping (AEPM) strategy is developed. The results from a priori numerical test demonstrate that inflow turbulence fields generated by the established synthetic random Fourier framework in combination with the AEPM accurately preserve the target turbulence energy profiles while maintaining good agreement with spatial correlation properties. Furthermore, the robustness and effectiveness of the AEPM are demonstrated through three LES case studies. Case 1 and Case 2 simulate atmospheric boundary layers around an isolated rectangular building and are validated against experimental data under neutral and unstable thermal stratification, respectively. In these two cases, wall-modeled LES (WMLES) results are compared with wall-resolved LES to investigate the performance of WMLES in urban environment simulations. In Case 3, the building in Case 1 is removed to form an empty domain, in which downstream turbulence statistics are evaluated to further assess the AEPM-generated inflow.

Responding to the urgent need for precise, one-month-ahead crop growth predictions in Central Southwestern Asia (CSWA), this study introduces a fully operational convolutional neural network (CNN)-climate dynamical hybrid model designed for real-time agricultural planning and management. It is engineered to accurately forecast the Normalized Difference Vegetation Index (NDVI), a vital indicator of crop health, with a one-month lead time. The model integrates multi-temporal data, including soil moisture and temperature from the preceding months, and historical NDVI, enhancing its predictive accuracy with 500hPa geopotential heights and 2-meter surface temperatures refined through a U-Net-based CNN. These meteorological inputs are sourced from the Flexible Global Ocean-Atmosphere-Land System Model finite volume version 2 (FGOALS-f2), an advanced global dynamical prediction system. Empirical validation across CSWA demonstrates the model's robust performance, with pattern correlation coefficients of 0.60, 0.70, and 0.58, root mean squared errors of 0.036, 0.029, and 0.022, and sign consistency rates of 74.8 %, 77.1 %, and 73.3 % for April, May, and June, respectively. Seamlessly integrated into the operational framework of FGOALS-f2, this model enables real-time, one-month advance predictions of NDVI. This pioneering approach not only enhances the accuracy of subseasonal crop growth forecasts in CSWA but also sets a new standard for subseasonal climate services.

Subseasonal precipitation whiplashes, marked by sudden shifts between dry and wet extremes, can disrupt ecosystems and human well-being. Predicting these events two to six weeks in advance is crucial for disaster management. Here, we show that the propagation diversity of the Madden-Julian Oscillation (MJO)-a key source of subseasonal predictability-will alter under anthropogenic warming. This is evidenced by a 40% increase in fast-propagating events by the late 21st century. Fast-propagating MJOs may rise in a period as early as 2028-2063, increasing the global risk of precipitation whiplashes through teleconnections. We propose a heuristic framework diagnosing that MJO's acceleration is primarily driven by enhanced atmospheric stabilization and El Niño-like sea surface warming. The expected rise in fast-propagating MJOs could improve the predictability of subseasonal weather whiplashes, offering critical lead time for disaster preparedness. Understanding these impending shifts is essential for enhancing subseasonal prediction capabilities.

Global atmospheric rivers are major conduits for moisture and energy transport, playing a critical role in the global hydrological cycle and energy redistribution as climate changes. Here, we assess their evolving roles using observations and climate simulations. By 2100, under a high-emission scenario, about 70% of mid-latitude atmospheric rivers are projected to carry more moisture than the Amazon River, with 11% of this intensification driven by future warming. This trend is expected to increase flood risks in densely populated basins such as the Yangtze, Loire, and Sacramento. Moist static energy provides insight into atmospheric rivers' behavior on subseasonal scales. Atmospheric rivers redistribute meridional energy transport during active seasons, a process projected to intensify and shift poleward. Their diverse heat archetypes have varying impacts on precipitation and temperature. These findings highlight the growing influence of atmospheric rivers as dynamic agents of freshwater delivery and heat transport in a warming climate.

Climate science demands automated workflows to transform comprehensive questions into data-driven statements across massive, heterogeneous datasets. However, generic LLM agents and static scripting pipelines lack climate-specific context and flexibility, thus, perform poorly in practice. We present ClimateAgent, an autonomous multi-agent framework that orchestrates end-to-end climate data analytic workflows. ClimateAgent decomposes user questions into executable sub-tasks coordinated by an Orchestrate-Agent and a Plan-Agent; acquires data via specialized Data-Agents that dynamically introspect APIs to synthesize robust download scripts; and completes analysis and reporting with a Coding-Agent that generates Python code, visualizations, and a final report with a built-in self-correction loop. To enable systematic evaluation, we introduce Climate-Agent-Bench-85, a benchmark of 85 real-world tasks spanning atmospheric rivers, drought, extreme precipitation, heat waves, sea surface temperature, and tropical cyclones. On Climate-Agent-Bench-85, ClimateAgent achieves 100

This study examines the remote influence of the Tibetan Plateau (TP) diabatic heating on atmospheric river (AR) activity in the North Pacific. First, we identify a sensitive heating region on the southern TP establishing a positive correlation between its heating and AR frequency. This correlation is attributed to latent heat release supported by a substantial moisture supply. Further analysis reveals that the dynamic effect of the eastward-propagating Rossby waves, originating from the Atlantic Ocean and modulated by the TP, facilitates upward moisture lifting. Using the Water Accounting Model-2Layers, we demonstrate that this anomalous heating is primarily due to increased moisture from the Indian Ocean, the Arabian Sea, and the Bay of Bengal, amplified by the westward extension of the western North Pacific subtropical high (WNPSH). Additional moisture contributions are observed from Eurasia. Moreover, Rossby wave activity over the TP propagates to the east of Japan, enhancing westerlies with upper-level divergence field which develops a cyclonic and anticyclonic vortex structure. This structure attracts abundant moisture to the North Pacific, thereby increasing AR activity. This study also highlights a positive feedback mechanism involving the southern TP heating and the eastward-propagating upper-level anticyclone, which enhance the western extension of the WNPSH. These findings underscore the global climatic impacts of the TP, emphasizing its role as a critical factor in AR dynamics across the North Pacific. SIGNIFICANCE STATEMENT: This study explores the remote impact of the Tibetan Plateau (TP) heating on the atmospheric river (AR) in the North Pacific. Our findings indicate that southern TP heating resulting from remote moisture contribution, quantitatively assessed using Water Accounting Model-2Layers, coupled with the western extension of the western North Pacific subtropical high (WNPSH) and sea surface temperature warming anomaly as well as the eastward-propagating Rossby waves derived from the Atlantic Ocean, plays a crucial role in AR activity in the North Pacific. TP heating affects Rossby waves propagating east of Japan enhancing the westerlies and the easterlies. Moreover, the coupled divergence and cyclonic vortex structures attract more moisture toward the North Pacific fostering AR activity. Additionally, the interplay between southern TP heating and WNPSH builds a positive feedback mechanism.