Achieving deep decarbonization while maintaining power system operational feasibility represents a critical challenge in regions with a high penetration of renewable energy. In this study, we developed an integrated LEAP–NEMO modeling framework to evaluate long-term decarbonization pathways and power system flexibility in Guangxi Province, China, from 2022 to 2060. Four scenarios were designed, the Business-as-Usual Scenario (BAS), the Policy Strengthening Scenario (PSS), the Technology-Centered Scenario (TCS), and the Integrated Optimization Scenario (IOS), to assess the impacts of policy, technology, and system-level coordination. Our results show that under the IOS, renewable electricity reaches approximately 78% by 2050 and further increases to 88% by 2060, while coal consumption declines by over 60% compared with the baseline. Energy storage deployment under the IOS reaches 25 GW by 2060 (rising to approximately 28 GW under a higher-resolution 48-representative-day validation), approximately five times the level in the BAS (5 GW), significantly enhancing system flexibility and reducing renewable curtailment from 10.2% to 1.8%. Carbon emissions peak in 2027 at 228 Mt CO2, approximately 20% lower than the baseline, and decline to 52 Mt CO2 by 2060, achieving a cumulative reduction of 3.85 Gt CO2. Economically, the IOS pathway reduces total system costs to USD 815 billion, saving approximately USD 95 billion compared with the BAS, with the levelized cost of electricity declining to USD 63/MWh by 2060. These results demonstrate that integrated optimization of generation and storage resources is essential in achieving deep decarbonization while maintaining system deterministic capacity adequacy and economic efficiency.
Mega sporting events report carbon footprints, but rarely describe the air pollutants that accompany them. This study links the verified greenhouse-gas inventory of the Beijing 2022 Winter Olympics (713,788 t CO2e, 2016–2022) to co-emitted NOx, SO2, PM2.5, CO and NMVOC in a screening-level, emission-side assessment using eight emission carriers with Chinese pollutant-to-CO2 ratios, Shapley attribution, and Monte Carlo and Sobol analyses. Construction accounted for 63% of the CO2e, an estimated 70% of the 1335 t of NOx, and 76% of the 279 t of PM2.5, remaining the largest NOx source in 99.9–100% of uncertainty draws and under alternative attributions. The measures taken, chiefly the employment of green electricity, avoided 31% of no-measure emissions and 18% of NOx under market-based accounting, but only 8% of NOx if green certificates were not used. Under literature-based assumptions, eight additional levers could have cut CO2e by 46% (41–53%) and NOx by 49%; zero-emission construction machinery avoids the most NOx per tonne of CO2. A residual of 384,375 t would remain; the offsets retired (2.4 times the emissions) reduced no air pollutants. For construction-dominated events, such as this pandemic-era edition, climate- and air-quality co-benefits are decided largely at the design stage, through materials and machinery, rather than by offsetting.
Wind power forecasting is critical to grid stability and renewable energy integration. However, existing deep learning methods struggle to incorporate semantic domain knowledge from textual information, exhibit limited generalization with scarce training data, and require high computational costs for extensive fine-tuning. Large language models (LLMs) offer a promising solution through their semantic representations, few-shot learning capabilities, and multimodal processing abilities. This paper proposes LLM-WPFNet, a dual-modality fusion framework that integrates frozen pre-trained LLMs with time-series analysis for wind power forecasting. The key insight is encoding temporal patterns as structured textual prompts to enable semantic guidance from frozen LLMs without fine-tuning. LLM-WPFNet employs two parallel encoding branches to extract complementary features from time series and textual prompts, unified through asymmetric multi-head attention fusion that enables selective semantic knowledge transfer from frozen LLM embeddings to enhance temporal representations. By maintaining the LLM frozen, our method achieves computational efficiency while leveraging robust semantic representations. Extensive experiments on four wind farm datasets (36–200 MW) across five prediction horizons (1–24 h) demonstrate that LLM-WPFNet consistently outperforms state-of-the-art baselines by 11% in MAE and RMSE. Notably, with only 10% of training data, it achieves a 17.6% improvement over the best baseline, validating its effectiveness in both standard and data-scarce scenarios. These results highlight the effectiveness and robustness of the dual-modality fusion design in predicting wind power under complex real-world conditions.
Guangdong is China’s largest provincial electricity consumer (912.1 TWh in 2024). Using a Low Emissions Analysis Platform (LEAP) model calibrated to the 2024 provincial yearbook, this paper compares a Business-as-Usual (BAU) reference with three decarbonization pathways over 2024–2060: Policy-Instrument-Coupled (PIC), Deep-Electrification-and-Hydrogen (DEH) and Digital-economy-with-CCUS (DCN). From 862 Mt CO2 in 2024, BAU peaks at 1143 Mt in 2040, whereas the pathways fall to 139, 99 and 35 Mt of direct emissions in 2060. After BECCS removals, only DCN approaches neutrality (21 Mt net, comparable to a conservative provincial land-sink estimate). Forty-three control scenarios show that end-use efficiency and fuel switching deliver 85–91% of the 2060 reduction relative to BAU and the capacity mix 8–10%. In 10,000 correlated Monte Carlo draws over the tested ranges, the ranking DCN < DEH < PIC never changes. In an hourly check (three ERA5 weather years, stylized load), the pathways serve every hour with at most 17 GW of four-hour storage (20–32 GW with plan-level offshore wind and nuclear), whereas the BAU capacity path falls short from 2040. By 2060, 38–57% of the pathways’ must-run output is surplus, and electricity-supply cost per MWh rises in the reverse order of emission depth.
Provincial power systems must decarbonize while sustaining rapid demand growth and improving air quality; yet, few integrated assessments separate the contributions of renewable expansion and carbon capture, utilization, and storage (CCUS) for China’s less-developed regions. We apply the Low Emissions Analysis Platform (LEAP) to model Guangxi’s power system to 2060 under four scenarios—Reference (BAS), Renewable-driven (RES), CCUS-intensive (CCS), and an Integrated comprehensive-policy scenario (ICS). Under ICS, solar photovoltaic generation rises from 38 TWh in 2025 to 408 TWh in 2060 (close to half of all generations), non-fossil capacity grows by about 730%, and power-sector CO2 falls by roughly 95% relative to BAS. A counterfactual decomposition shows that CCUS provides about 75% of the CO2 reduction along the coal-retaining CCS pathway but only about 5% along the renewables-led ICS pathway and reduces neither SO2 nor NOx. The air-quality co-benefits—up to 82%, 78%, and 73% lower SO2, NOx, and PM2.5 than BAS—arise mainly from renewable substitution, which also abates carbon more cheaply (about 120–190 versus 300 CNY t−1 CO2) while sharply raising flexibility needs. Once avoided fuel and carbon-market costs are included, the renewables-led pathways become net cost-saving on a full-system basis and yield a monetized air-quality health co-benefit roughly twice that of the CCUS-intensive route. The findings give policy-relevant guidance for sustainable low-carbon transitions in Guangxi and comparable emerging regions.
Transport emissions are still rising in fast-motorising Chinese provinces. In Guangxi, the vehicle fleet grew 32% between 2020 and 2024, while the grid is already half non-fossil. A bottom-up model built on the Low Emissions Analysis Platform (LEAP) and calibrated to official 2024 statistics compares four pathways to 2060, with the grid emission factor calculated from each scenario’s generating mix. In 2024, the sector consumed 567 PJ and emitted 41.9 Mt CO2. Under business as usual, emissions peak at 48.0 Mt around 2036 and remain at 37.4 Mt in 2060; the three policy pathways peak by 2027 and reach 21.5, 16.8, and 11.7 Mt in 2060 (42–69% below business as usual), cutting cumulative emissions by 24–45%. An order-independent Shapley decomposition attributes 50% of the 2060 abatement to vehicle electrification, 36% to power-sector decarbonisation, and 11% to modal shift. The two leading levers are complements and jointly deliver 86% of it (Monte Carlo range 84–88%), but their relative size is not settled: if charging draws disproportionately on thermal generation, the grid’s share exceeds that of electrification. Even the deepest, net-zero-aligned pathway leaves 11.7 Mt that must be balanced outside the sector.