Accurate subseasonal-to-seasonal predictions over North China pose significant challenges, as the region is increasingly affected by extreme summer precipitation. However, the modulation of summer subseasonal precipitation by land surface processes, especially soil moisture, remains poorly understood, limiting model improvements. This study assesses the predictive skill of the China Meteorological Administration-Climate Prediction System version 3 (CMA-CPSv3) and the high-resolution Climate–Weather Research and Forecasting (CWRF) model for summer subseasonal precipitation over North China, leveraging CWRF to investigate how soil moisture influence July precipitation. Results show that CWRF, with higher resolution and improved physical parameterizations, partially corrects CMA-CPSv3's overestimation of precipitation in central and northern North China, improving predictions of spatial distribution and intensity. However, constrained by biases in both the CMA-CPSv3 initial conditions and the CWRF physics schemes, the downscaling results still exhibit limited accuracy improvements. Moisture budget analysis identifies the southern boundary of North China as the primary moisture influx pathway, with correlations to the net moisture budget of 0.58, 0.65, and 0.43 in the 1st, 3rd, and 5th pentads of July. Soil moisture prediction bias explains 27%–28% of the 925 hPa temperature variance. A soil moisture gradient across the Taihang Mountains creates a “west warm and east cold” thermal contrast through evaporative cooling (wetter east side) and sensible heating (drier west side). Together with Loess Plateau topography, this contrast enhances southerly wind and moisture convergence on the eastern mountain slope, favoring precipitation. The upward propagation of the soil moisture-induced thermal signal is effective when the soil moisture bias gradient aligns with the prevailing geopotential height gradient. This study highlights soil moisture as a key source of predictability for subseasonal precipitation over North China, offering insights for subseasonal forecast improvements.

Accurate wind speed forecasting is essential for renewable energy integration. However, the relative contributions of global numerical weather prediction (NWP) selection, regional downscaling, and artificial intelligence (AI) post-processing within operational forecasting chains remain poorly quantified. This study systematically evaluates ECMWF HRES and GFS within a unified framework integrating Weather Research and Forecasting (WRF) downscaling and Pyraformer-based AI correction, using three years of observations from wind farms in northwestern China's Gobi region. Three principal findings emerge. First, ECMWF HRES consistently outperforms GFS, with average advantages of 3-4 % RMSE reduction that increase systematically with forecast lead time. Second, AI post-processing contributes approximately 20 % RMSE reduction compared to 3-4 % from NWP switching. AI effectiveness diminishes at extended lead times while NWP quality differences become more pronounced. Third, forecast activity metrics expose a critical limitation invisible to traditional error measures. AI corrections systematically suppress forecast variability by 20-30 %, degrading extreme event representation essential for operational decisions. These findings demonstrate that AI post-processing delivers five-to six-fold greater improvements than premium NWP subscriptions, reshaping resource allocation priorities. The results provide practical guidance for forecasting system design and highlight the necessity of evaluation frameworks balancing error reduction with variability preservation.

Abstract Atmosphere‐ocean‐land coupled forecasting systems, despite their comprehensiveness, face substantial challenges in the “predictability desert” at subseasonal to seasonal (S2S) timescales, particularly for precipitation—a variable crucial for socioeconomic activities yet of stunning spatiotemporal variance. Post‐processing methods developed for numerical weather prediction and climate projections are not directly applicable to S2S forecasting, as they cannot distinctively address initialization errors, chaos‐induced state uncertainty growth, and model systematic biases. Additionally, regression‐based deep learning corrections introduce smoothing artifacts and ensemble under‐dispersion, limiting their ability to capture key processes and extreme events. We propose an integrated framework using Generative Adversarial Networks (GANs) for ensemble post‐processing. The approach exploits the ability of deep generative models to represent high‐dimensional distributions, combining trajectory constraints from short‐term forecasts with distributional constraints from long‐term climatology. In a case study using ECMWF hindcasts over Southern China, our model calibrates ensemble forecasts while enhancing both ensemble size and spatial resolution. The post‐processed forecasts maintain deterministic skill (anomaly correlation coefficient) while showing improved probabilistic forecast metrics, such as Continuous Ranked Probability Score (CRPS) and Brier score, extending the skillful probabilistic forecast horizon to week 3. The predicted fields demonstrate improved spatial distribution matching and maintain linear covariance across variables. The framework demonstrates strong spread‐error correlations for effective advance error estimation, and helps disentangle forecast uncertainties into propagated dynamical and post‐processing components, each with distinct lead‐time dependencies. This unified framework demonstrates the potential to advance seamless forecast capabilities while addressing the growing demand for high‐resolution, physically consistent S2S products.

Climate events, particularly El Nino and La Nina, exert a significant influence on surface ozone (O3) concentration anomalies across China through atmospheric circulation and regional synoptic patterns. This study investigates the disparities in summer O3 concentrations between four El Nino and four La Nina events and elucidates the underlying mechanisms driving these variations. Leveraging the GEOS-Chem chemical transport model and meteorological reanalysis datasets, we examined the spatiotemporal O3 anomalies during the El Nino (2015) and La Nina (2010) event. When considering only the impact of natural factors (with anthropogenic emission levels held constant), the O3 concentration in the North China Plain (NCP) during El Nino summer case (June-July-August, JJA) is 1.93 ppbv lower than during La Nina. In contrast, concentrations in the Yangtze River Delta (YRD), Pearl River Delta (PRD), Sichuan Basin (SCB), and Guanzhong Plain (GZP) are 1.68, 3.85, 3.35, and 2.02 ppbv higher, respectively, during El Nino summers. The combined fields of multiple El Ninos and La Ninas have similar anomalous patterns. However, when anthropogenic influences are incorporated, the differences induced by natural factors are amplified. This indicates that meteorological conditions, in conjunction with enhanced anthropogenic emissions, collectively exacerbate O3 pollution. Another key finding is that during late spring and early autumn, O3 concentrations in El Nino years are significantly higher than in La Nina years. The study indicates that emission reduction measures should be dynamic. In meteorological environments or climate events that are conducive to O3 production, such as high temperatures and droughts, emission reduction standards should be more stringent; conversely, emission constraints be relaxed.

Abstract In the second half of this two-part study, we apply a mesoscale convective system (MCS) tracking method to two satellite datasets, together covering nearly four decades—the longest observational MCS record to date over High Mountain Asia (HMA). Using this dataset, we examine past trends of warm-season MCS characteristics and their environmental drivers. A consistent decreasing trend in MCS frequency is found during 1985–2023. Analysis of dynamic and thermodynamic variables indicates increasing convective inhibition (CIN), which represents the energy barrier to convective initiation, as the primary driver of this trend. Enhanced CIN modulates the convective population by suppressing weaker convection, which is common over HMA, while favoring fewer but stronger convective systems. We further leverage simulations from a high-resolution global climate model and a global storm-resolving model to assess MCS responses to global warming. Both models consistently project continued reductions in MCS frequency under warmer conditions, suggesting a sustained downward trend in future climates. Changes in MCS-related precipitation are more nuanced, reflecting a balance between reduced MCS frequency and increased precipitation intensity per event. These findings provide a multidecadal perspective on the evolution and projected response of MCSs over HMA, highlighting the critical role of CIN in regulating convective organization and the potential for rarer but more intense storms to amplify hydrological and societal risks in a warming climate. Significance Statement Mesoscale convective systems (MCSs)—large, organized thunderstorms—play a crucial role in providing warm-season rainfall over High Mountain Asia (HMA). Using nearly four decades of satellite observations, this study finds that MCSs have become less frequent but more intense, meaning fewer storms now deliver heavier rainfall. This shift appears linked to stronger convective inhibition, a warming-related barrier that makes it harder for weaker storms to form. High-resolution GCM (∼25 km) and global storm-resolving model (∼3 km) simulations project a continued reduction in MCS activity under warming scenarios. The response of MCS-related precipitation is complex. Fewer events are offset by stronger precipitation per MCS. These changes could increase the risk of floods and landslides in this fragile mountain region where water resources are vital for millions of people.

Global wave forecasts play a crucial role in international shipping, global trade, and various economic and social activities. However, current numerical wave models, such as WAVEWATCH III (WW3), still exhibit substantial forecast errors, particularly in wave-active regions. Therefore, we propose an Intelligent Error Correction Method (IECM) that integrates multiple U-Nets with a boundary fusion algorithm to correct global Significant Wave Height (SWH) forecasts from WW3. This approach effectively considers different wave characteristics in different ocean regions and alleviates performance degradation at forecast boundaries. Following error correction, the mean Root Mean Squared Errors (RMSEs) of WW3 global SWH forecasts at lead times of 24, 48, and 72 h are reduced from 0.44 m, 0.46 m, and 0.49 m to 0.22 m, 0.25 m, and 0.31 m, corresponding to percentage reductions of 50

Lake sediments are valuable archives for reconstructing regional histories of trace metal contamination and evaluating the environmental impacts of human activities. Although studies on lacustrine sediments in southwestern China have improved our understanding of modern contamination trends, trace metal accumulation and its environmental effects during historical periods-particularly the Ming and Qing dynasties-remain poorly constrained. This limitation may lead to underestimation of anthropogenic inputs when pre-industrial background values are used as the baseline and may obscure the long-term environmental legacy of early mining and smelting activities. Here, we used a well-dated sediment core from Dalongtan Crater Lake to reconstruct high-resolution records of Pb, Sn, Sb, Bi, Cu, and Ag over the past millennium to apportion natural and anthropogenic sources. The results indicate that 1250-1640 CE corresponds to relatively low enrichment levels, suggesting limited anthropogenic influence. From 1640 to 1950 CE, a pronounced mining-smelting contamination signature emerged, marked by synchronous enrichment of Pb, Sn, Sb, and Bi that closely aligned with historically intensified regional metallurgical activity. Since 1950 CE, Pb and Sn have declined, whereas Sb has increased exponentially, consistent with a transition from metallurgy-dominated inputs to combustion-related atmospheric emissions. These findings demonstrate that using pre-industrial background values as the sole baseline can substantially underestimate the extent of modern contamination and mask the long-term legacy of historical metallurgical activity. Our results provide a basis for more accurate assessments of human-environment interactions and inform evidence-based contamination mitigation strategies in southwestern China.

Holocene vegetation and climate variability in Northeast China have attracted increasing attention in recent years. However, precise quantitative reconstructions remain limited, constraining a comprehensive understanding of regional environmental evolution and the underlying climatic mechanisms. In this study, we present a well-dated Holocene pollen record from Lake Daerbinluo, northern Northeast China. Vegetation dynamics were inferred using the biomization method, while mean annual temperature (Tann) and annual precipitation (Pann) were quantitatively reconstructed using a random forest approach. Our results reveal that a mosaic of forest and steppe dominated the region between 12,600 and 6200 Cal yr BP, transitioning to a temperate deciduous broadleaf forest as the dominant vegetation type after 6200 Cal yr BP. Climatic reconstructions show distinct trajectories for Tann and Pann. Tann exhibited a warming trend during the early Holocene, remained relatively high throughout the mid-Holocene, and experienced a slight decline during the late Holocene, broadly consistent with trends observed across the Northern Hemisphere. In contrast, Pann remained low but gradually increased prior to 5500 Cal yr BP, followed by an abrupt rise and sustained high levels during the mid to late Holocene, despite a minor decreasing tendency. These findings provide new insights into the climatic evolution and potential forcing mechanisms in Northeast China. The increased humidity during the mid-to late Holocene in our study area was likely driven by a strengthened westerly influence, despite the EASM remaining the principal moisture source, possibly via more frequent frontal cyclones and enhanced westerly moisture transport.

Heavy metals have accumulated continuously during industrialization and agricultural expansion, posing serious threats to ecosystems and human health. Accurately identifying natural baselines and quantifying anthropogenic forcing are essential for effective pollution control. However, most previous studies relied on static baselines, lacked quantitative characterization of threshold exceedance, and were limited in temporal scope. This study focuses on Sihailongwan Maar Lake (SHML) in northeastern China, reconstructing a 1,600-year high-resolution history of heavy metal accumulation. Results show a critical shift around 1930 CE from natural to anthropogenic control. Before 1930 CE, concentrations remained at natural levels; afterward, rapid industrialization and agricultural expansion became the dominant drivers, sharply increasing pollution. Using a Dynamic Natural Baseline (DNB) and Baseline Deviation Multiple (BDM), Cd and Pb exhibited the strongest anthropogenic deviations, while Zn, As, Hg, Sb, and Cr showed moderate levels. A Threshold Exceedance Index (TEI) further distinguished disturbance types: Cd, Pb, Sb, and Zn displayed sustained forcing, remaining at high levels, whereas As, Hg, and Cr showed transient forcing with reversible, stage-dependent exceedance. By integrating DNB, BDM, and TEI, this study proposes an analytical approach to quantitatively identify threshold-exceedance processes, providing a robust scientific basis for understanding the evolution of heavy metal pollution under anthropogenic forcing and for developing targeted pollution control strategies.

Seasonal precipitation anomalies are largely regulated by atmospheric circulation, which dynamical models predict with greater reliability than precipitation itself. Here, we employ a deep learning model that translates dynamical circulation predictions into precipitation estimates. Predictions initialized from March to May consistently indicate a dry anomaly over central China in summer 2026. Retrospective evaluations revealed higher predictive skill in the analogue years, which also tended to feature central equatorial Pacific warming persisting from the preceding winter into summer. This warming favors an anomalous cyclonic circulation over the western North Pacific-South China Sea-South China region, which induces northerly winds and moisture divergence that jointly suppress rainfall over central China. Supporting this mechanism, layer-wise relevance propagation (LRP) independently identifies these northerly winds as the dominant driver of the prediction among all model inputs. Perturbation tests supported this attribution: removing LRP-identified features effectively eliminates the dry anomaly. Our framework thus provides physically interpretable explanations for AI-derived regional climate projections, facilitating evidence-based assessment before observational data become available.

Accurate short-term (0-72 h) forecasting of wind speed and power is essential for wind energy integration, yet remains hindered by persistent biases in Numerical Weather Prediction (NWP) models. This study presents the first systematic evaluation of Transformer-based time series forecasting (TSF) models for post-processing Weather Research and Forecasting (WRF) simulations. Using three years (2020-2022) of data from four operational wind farms in northwestern China, we assessed Pyraformer, Autoformer, Reformer, and other TSF variants. Multi-resolution architectures, particularly Pyraformer, demonstrate clear superiority. Compared to raw WRF hub-height wind speeds, Pyraformer reduces root mean squared error (RMSE) by up to 27 %, and relative to other Transformers, improves power forecasting accuracy by 3-6 % and qualification rate by 3-9 %. Mechanistic analysis reveals that decomposition-based Transformers fail by spuriously fitting temporal trends in stationary WRF biases, while multi-resolution architectures effectively correct these regime-dependent errors through scaleaware attention. A unified evaluation framework shows that nonlinear wind-to-power conversion amplifies meteorological uncertainties, resulting in consistently higher normalized errors in power forecasts than in wind speed forecasts across lead times and wind regimes. All methods are implemented within an integrated Global Forecast System (GFS)-WRF-AI pipeline. These findings offer mechanistic insights for next-generation hybrid forecasting model development and demonstrate the potential of advanced TSF architectures for operational wind power forecasting.

Paleofire studies have provided critical long-term data on past fire dynamics. However, previous research has predominantly focused on the influence of climatic factors on paleofires, paying little attention to the role of climatic stability. Here, we use a precisely established varve chronology to reconstruct a similar to 1400-year high-resolution fire history from charcoal and soot records in the sediments of Sihailongwan Maar Lake, a representative forested region of northeastern China. We further investigate the role of climatic stability in controlling fire frequency and size. We show that during periods of climatic instability (ca. 900-1100 CE, ca. 1250-1500 CE, ca. 1600-1800 CE), fire frequency markedly increased, while the individual fires were generally small. In contrast, during periods of climatic stability (ca.710-900 CE, ca. 1100-1250 CE, ca. 1500-1600 CE), fires occurred less frequently but were larger in size. Climatic stability regulated fire dynamics by affecting fuel accumulation and moisture content through climatic whiplash (extreme wet-dry oscillations). Against the recent backdrop of rapid global warming and regional anthropogenic fire suppression, the potential risk of fires may increase dramatically. Our result supports a policy of shifting fire suppression policies toward an "healthy forest" management to balance ecological conservation and fire management needs.

Photovoltaic (PV) panel surface temperature (PST) is critical for assessing PV conversion efficiency, optimizing maintenance scheduling, and managing thermal risks in utility-scale PV power plants particularly in warm season (May-October). However, PST retrieval from moderate-resolution thermal infrared (TIR) imagery poses unique challenges due to mixed pixels, varied satellite viewing geometry, and PV-specific low and directional emissivity. In this study, we develop a 1-km PST retrieval that combines thermal component decomposition, directional PV areal fraction estimation, and PV-specific directional emissivity parameterization based on threedimensional array-sensor geometry, with PST solved using the inverse Planck function. Validation against in-situ PST measurements in an arid site (Wujiaqu) and a more humid site (Ganzi) shows substantial accuracy gains over a land-surface-emissivity baseline: root-mean-square error (RMSE) decreases from 11.3 degrees C (15.2 degrees C) to 3.7 degrees C (6.4 degrees C) at 10:30 and from 10.8 degrees C (18.9 degrees C) to 5.5 degrees C (8.6 degrees C) at 13:30 in warm season at the arid site (the humid site). These improvements are primarily attributed to the lower emissivity of PV panels and secondarily to directional effects of panel emissivity and viewing geometry, thereby effectively mitigating the systematic cold bias in PST estimation. Notably, the land surface temperature of PV gaps (Tgap) is a key uncertainty source in the component decomposition, with PST changing by about 2 degrees C per 1 degrees C perturbation in Tgap, which may degrade the performance in winter under large inter-panel shadow conditions. This study provides a feasible approach for PV thermal monitoring from moderate-resolution TIR imagery in warm season.

Abstract Mesoscale convective systems (MCSs) are key drivers of the hydrological cycle over High Mountain Asia (HMA), delivering essential warm-season rainfall but also triggering flash floods and landslides. Their simulation over this complex region remains challenging for coarse-resolution models, and regional convection-permitting models cannot fully capture large-scale feedback. In this two-part study, we use global models at 25- and 3-km resolution (the latter storm resolving) to assess MCS characteristics and climate response over HMA. Part I provides a comprehensive evaluation against satellite observations. Both models capture the spatial distribution and seasonality of MCSs but overestimate warm-season frequency, with underestimation at low elevations and overestimation at high elevations. The storm-resolving model better reproduces the diurnal cycle. At the event scale, simulated MCSs are slightly larger, longer lived, and more intense than observed. Both reproduce the dominant eastward propagation and speeds but exaggerate a secondary southwestward mode. They capture broad precipitation patterns, including the dry zone north of the Himalayas, though the 25-km model retains a wet bias along the southern slopes that is reduced in the 3-km simulation. These findings highlight both the promise and limitations of current high-resolution global models in representing MCSs over complex terrain, providing a basis for assessing historical and future changes (Part II) and guiding future model development. Significance Statement Mesoscale convective systems (MCSs) are organized clusters of deep convection that play a key role in regulating the energy and water cycles over High Mountain Asia (HMA). They often produce hazardous weather, including flooding, strong winds, and hail. Accurately simulating and projecting these systems are essential for improving predictions of both mean climate conditions and extreme events. Because MCSs have fine-scale structures and complex interactions with large-scale circulation, high-resolution global models have become valuable tools for studying them. In this study, we conduct a detailed evaluation of a high-resolution global climate model and a global storm-resolving model in simulating key characteristics of MCSs over HMA. These evaluations provide the foundation for subsequent analyses of historical trends and future projections of MCS activity in the region.

[Objective] This study aims to reveal the temporal evolution patterns of PM2.5and O3pollution in urban and suburban areas of Nanchang, analyze their synergistic relationships across different temporal scales, and explore the driving mechanisms underlying these relationships. [Methods] Based on hourly observational data of PM2.5and O3from nine national control stations in Nanchang（2015-2023）, combined with ERA5 meteorological reanalysis data, percentile analysis and Pearson correlation coefficient methods were used to systematically investigate pollution characteristics, synergistic relationships, and influencing factors. [Results] From 2015 to 2023, the 90th percentile of PM2.5concentrations in urban-suburban areas decreased at an annual rate of 2.5 μg/（m3·a）and 1.4 μg/（m3·a）, while the 90th percentile of O3concentrations increased at 1.6 μg/（m3·a）and 5.0 μg/（m3·a）. Days with good or excellent air quality conditions for PM2.5and O3, respectively accounted for over 85% of the total days in the study period. PM2.5pollution days in urban-suburban areas were primarily concentrated in winter（January, February, and December）, whereas O3pollution days showed a“bimodal”distribution pattern dominated by May and September. Only four days of co-occurring PM2.5-O3pollution were observed in Nanchang city. On a daily scale, PM2.5and O3showed positive synergistic relationships in both urban-suburban areas, with the strongest correlation in summer. On an hourly scale, except for positive correlations in urban areas during summer, negative correlations were generally observed in other seasons. Mechanistic analysis, in which correlations between primary/secondary PM2.5and O3were seperately calculated, revealed that daily-scale synergistic relationships were primarily driven by secondary PM2.5formation, while hourly-scale relationships were influenced by secondary formation and boundary layer dynamics. Among meteorological factors, solar radiation and air temperature promoted positive synergistic effects, whereas precipitation and humidity weakened positive relationships between them or enhanced negative relationships. [Conclusion] PM2.5and O3pollution in Nanchang exhibited opposite trends:PM2.5pollution significantly improved, while O3pollution worsened. Daily-scale synergistic relationships were generally positive, primarily driven by secondary PM2.5formation, whereas hourly-scale relationships were closely associated with boundary layer height variations. Other meteorological factors also exerted regulatory effects on these relationships. [Significance] This study revealed the pollution transition characteristics of PM2.5-O3in Nanchang, providing a scientific basis for atmospheric multi-pollutant collaborative prevention and control, regional air quality improvement, and public health protection in the city.

Understanding the responses of wildfires to long-term climate variability is critical for improving fire management strategies and maintaining ecological stability. However, the spatial heterogeneity and drivers of wildfires in the East Asian monsoon (EAM) region on orbital timescales remain poorly constrained. In this study, wildfire dynamics across the last five glacial-interglacial cycles were determined based on a similar to 480 kyr wildfire history reconstructed from black carbon (BC) records in the Huangshan loess-paleosol sequence in northeastern China. The results revealed substantial glacial-interglacial cycles in northeastern China, with more intense fires during interglacials than glacials. Comparative analysis with other wildfire records revealed notable spatial heterogeneity of wildfires across the EAM region on the glacial-interglacial timescale. Wildfires in the eastern East Asian monsoon (EEAM) region were more active during interglacials, likely due to higher temperatures and greater biomass availability. Conversely, those in the western East Asian monsoon (WEAM) region were more frequent during glacials because of fuel desiccation caused by reduced precipitation. Overall, the large-scale spatial precipitation pattern, which gradually decreased from southeast to northwest in the EAM region, may serve as the core driver of wildfire spatial heterogeneity. Under future global warming, the EEAM region will likely face elevated wildfire risks and should be prioritized for fire prevention.

Since the pre-industrial era, human activities have drastically increased reactive nitrogen (Nr) deposition in lakes, altering nitrogen (N) cycles. To trace its continental-scale footprints, we synthesized dated sediment δ15N records from 51 remote lakes across North America, Europe, and East Asia. Results reveal that the accelerated declines in δ15N (indicating increased Nr deposition) occurred earlier in North American and European lakes (~1950 CE), coinciding with the Great Acceleration, while similar changes in East Asian lakes appeared around 1985 CE, paralleling China’s rapid socioeconomic development. δ15N in North American lakes reversed around 2005 CE (~10%), and earlier in European lakes (~1995 CE) with more pronounced (~40%) increases. Meanwhile, East Asian lakes showed no reversal (only a slowdown in the δ15N decline). These regional differences match the timing and implementation of N emission policies, underscoring the need for region-specific and multi-N species (e.g., NHx and NOy) mitigation strategies to protect lake ecosystems. Enhanced deposition of anthropogenic nitrogen in North American and European lakes emerged during the Great Acceleration (post-1950), while shifts in East Asian lakes appeared after the 1980s, based on isotopic analysis of sediment cores from 51 lakes across East Asia, North America, and Europe.

Assessing photovoltaic (PV) technical potential in cold alpine basins requires reliable solar-resource reconstruction, realistic module-temperature representation under strong thermal variability, and explicit consideration of ecological admissibility. Here, we develop a temperature-regulated LSTM-SAPM-ECM framework to assess historical evolution and future PV technical potential in the source region of the Yellow River, a solar-rich cold alpine headwater region and ecological barrier in western China. A unified data chain was constructed using station observations and radiation reconstruction for 1974—1978, fused station-CMFD-ERA5-Land data for 1979—2025, and bias-corrected climate drivers for 2030—2100 under RCP2.6, RCP4.5, and RCP8.5. The classical SAPM was improved by incorporating freeze-thaw, albedo, and wind-interaction effects into the thermal module, while a dual-layer ecological constraint module represented hard exclusions and soft ecological penalties. Solar resources exhibit a persistent northwest-to-southeast decreasing gradient, but high-resource areas do not necessarily correspond to high developable potential. Model performance improved from SAPM to LSTM-SAPM-ECM, with RMSE decreasing from 307.9 to 238.0 MJ m-2. Dual ecological constraints reduced average developable area by 32.8%. By 2100, ecologically constrained installed capacity and annual generation potential are projected to reach 10.5—14.5 GW and 16.4—23.1 TWh, respectively.

Amid rapid urbanization, China faces dual challenges of air pollution (AP) and carbon emissions (CE), urgently requiring synergistic governance—an area that remains underexplored due to limited dynamic and heterogeneous perspectives in existing literature. To address this gap, this study integrated spatiotemporal transition analysis with a two-way fixed-effects panel quantile regression model, identifying key synergistic drivers based on a novel classification of cities by pollution-carbon reduction urgency. Key findings revealed: (1) Distinct spatiotemporal co-evolution patterns: Annual PM2.5 concentration (APC) in Chinese cities followed an “increase-then-decline” trajectory, while per capita carbon emissions (PCCE) showed “rapid growth followed by moderation.” Both exhibited significant and intensifying spatial agglomeration, with CE demonstrating greater rigidity and path dependence. Three governance zones were identified, with priority zones—concentrated mainly in northern China—requiring the most urgent integrated action; targeted zonal strategies were accordingly proposed. (2) Complex synergistic reduction mechanisms: Drivers were classified into four categories—fully synergistic, non-synergistic, context-dependent synergistic, and one-dimensional. Based on this, a differentiated policy framework emphasizing “categorized implementation and systemic integration” was developed, offering theoretical and practical support for advancing the synergistic reduction of atmospheric pollution and carbon emissions (SRAPCE).

As the "water tower of Asia", the retreat of glaciers on the Qinghai-Tibet Plateau is reshaping the regional power system through hydrological processes, ecological feedback, and human activities. This study establishes a "Glacier-Hydrology-Ecology-Power" cascading analysis framework, incorporating multi-source data from 1974 to 2024. Various glacier retreat scenarios are simulated under RCP4.5 and RCP8.5 climate scenarios to project electricity demand and photovoltaic installation paths for 2030-2060. The results show that in the short term, increased meltwater may lead to localized hydropower gains. However, as ice reserves decline, the reduced water availability during dry seasons, enhanced runoff fluctuations, and the advancement of seasonal peaks will undermine hydropower system stability. Hydrological fluctuations serve as the key intermediary layer linking glacier changes with power system responses, while ecological constraints form the main amplification channel on the demand side. The primary challenge in the regional power system has shifted from simple electricity growth to structural constraints arising from clean energy fluctuations, rising local loads, and insufficient flexibility. By 2060, regional electricity demand is expected to increase from 2300 & times; 108-2480 & times; 108 kWh to 2880 & times; 108-3250 & times; 108 kWh, and photovoltaic installations will grow from 190-220 GW to 270-340 GW. This study demonstrates that the impact of glacier retreat on the power system is not only a supply-side shock, but also a cascading effect shaped by hydrological fluctuations, ecological feedback, and energy substitution.