The Paris Agreement aims to limit global warming to below 2.0 °C, ideally 1.5 °C, above pre-industrial levels. However, current mitigation strategies may not suffice. China, as the world's largest developing country and carbon emitter, has pledged to peak carbon dioxide emissions by 2030 and achieve carbon neutrality by 2060. Previous studies show a rising mean temperature in China, impacting temperature and precipitation extremes, with increasing extreme events causing significant economic losses. Global climate models (GCMs) are useful tools for evaluating climate change, and CMIP6 provides improved projections with more realistic scenarios. While CMIP6 models offer better representation of physical processes, uncertainties remain concerning resolution, physical processes, and forcing conditions. Bias-corrected projections are needed to improve the reliability of regional and local predictions and to inform policymaking. This study uses bias-corrected CMIP6 GCM outputs to project changes in extreme temperatures in China under different global warming scenarios, focusing on the potential benefits of limiting warming to 1.5 °C.

Numerous studies have examined climate extremes using GCMs from CMIP phases. However, the reliability of future climate projections remains debatable due to the coarse resolution and uncertainties associated with these models. CMIP6 offers improved projections compared to CMIP5, better representing smaller-scale processes and exhibiting globally improved historical representations of climate extremes indices. However, uncertainties persist, highlighting the need for bias correction in regional and local scale projections to enhance the reliability of climate predictions and inform adaptation and mitigation policies. Few studies have focused on bias-corrected projections of temperature extremes across China using CMIP6.

This study employed the Equidistant Cumulative Distribution Functions (EDCDF) method to correct biases in outputs from 12 CMIP6 GCMs for historical and future SSP245 and SSP585 scenarios across China. The study compared the performance of original and bias-corrected CMIP6 models. Model evaluation involved analyzing temporal evolution using Supplementary Figure 1 (comparing bias-corrected CMIP6 multi-model ensemble with CN05 observations) and spatial distribution (Figure 2, Supplementary Figure S2). Metrics used for model evaluation included Inter-annual Variability Skill Score (IVS, Figure 1, Figure 3) and Root Mean Square Error (RMSE, Supplementary Figure S2). The arrival times of 1.5 °C and 2.0 °C warming levels were determined based on smoothed global average surface temperature changes (Supplementary Figure S3, Supplementary Table 3). Spatial distributions of extreme temperature indices changes at 1.5 °C and 2.0 °C warming levels were analyzed (Figure 4, Supplementary Figures 4 and 5, Table 1). The incremental changes from 1.5 °C to 2.0 °C warming levels were also assessed (Figure 5, Supplementary Figure 6, Figure 6). Twelve extreme temperature indices defined by the Expert Team on Climate Change Detection and Indices (ETCCDI) were used, encompassing extremal, absolute, and relative indices. A regridded procedure was conducted before bias correction. The EDCDF method, a quantile mapping technique, was used for bias correction (Equation 1).

After bias correction, CMIP6 models showed improved skill in simulating extreme temperature indices in China, particularly in temporal evolution and spatial distribution. Bias-corrected models had lower IVS scores than the original ones, indicating better reproduction of observed interannual variability. Spatial biases in the original models were significantly reduced after correction. Projections from bias-corrected CMIP6 models revealed an increasing trend in extreme temperatures across most regions of China under both 1.5 °C and 2.0 °C warming scenarios. Extremal temperature indices (TXx, TXn, TNx, TNn) showed an overall increase. Summer days (SU), tropical nights (TR), warm days (TX90p), and warm nights (TN90p) increased, while icing days (ID), frost days (FD), cold days (TX10p), and cold nights (TN10p) decreased. The magnitude of change was generally larger at 2.0 °C than at 1.5 °C. Spatial distribution of changes was similar under each SSP scenario, although regional averages differed slightly. Analysis of the incremental impact of the additional 0.5 °C warming (from 1.5 °C to 2.0 °C) showed that most extreme indices increased proportionately more during the final 0.5 °C warming than during the first 1.5 °C across most of China. For warm indices (TXx, SU, TX90p), the largest incremental changes were in the southwest. Limiting warming to 1.5 °C significantly reduces extreme temperature risks in China compared to 2.0 °C. Under SSP585, incremental changes were similar to SSP245 but smaller in magnitude and spatial extent.

The study's findings highlight the significant benefits of limiting global warming to 1.5 °C rather than 2.0 °C for mitigating temperature extremes in China. The consistent increase in most extreme indices and the disproportionately larger increase in the final 0.5 °C of warming underscore the importance of stringent climate action. The spatial variations in incremental changes highlight regional vulnerabilities and the need for targeted adaptation strategies. Although bias correction improved model performance, uncertainties remain, particularly in high-altitude regions with complex topography. The use of transient simulations rather than near-equilibrium scenarios also introduces uncertainties. Future research should incorporate more models, scenarios, and high-resolution data to refine projections and reduce uncertainties.

This study demonstrates that limiting global warming to 1.5 °C offers substantial benefits for reducing the risk of extreme temperatures in China. The disproportionate increase in extreme indices during the final 0.5 °C of warming emphasizes the urgency of climate action. Further research with more models and improved data is needed to reduce uncertainties, particularly in complex terrain and high-altitude regions.

While bias correction improved model performance, uncertainties remain, particularly in high-altitude and complex terrain areas due to the coarse resolution of GCMs. The use of transient simulations, rather than near-equilibrium scenarios, also introduces uncertainty. The study relied on a limited number of CMIP6 GCMs, and including a wider range of models and scenarios in future studies would be beneficial.