中国 10 公里超分辨率逐日气候变化及潜在蒸散量数据2015-2100 年（SRCPCN10）

全球气候变暖加剧了极端天气事件，对区域气候和水文生态构成严峻挑战，而现有CMIP6数据的空间分辨率不足制约了精细化评估。针对现有多气候指标和潜在蒸散发（PET）空间分辨率不足的问题，本研究发展了基于深度残差注意力机制的超分辨率融合框架，构建了中国首套10公里分辨率多模式-多情景的高分辨率气候和PET数据集（SRCPCN10）。通过5个CMIP6模式在SSP126-SSP585情景下的百年尺度数据，实现气温（Tas）、降水（Pr）、辐射（Rsds/rlds）等9个气候要素和PET的协同降尺度。残差通道注意力网络（RCAN）在气温、辐射及气压等要素上展现出卓越的降尺度性能（R²/KGE>0.99），而降水的空间不连续性导致其降尺度精度显著偏低（R²=0.897）。本研究揭示中国未来气候指标在SSP情景下呈现显著排放梯度响应，气温、辐射及降水增幅随辐射强迫增强梯度递增。极端气候事件呈现强烈空间分异，西北干旱区水文干旱持续恶化，华南洪涝与高温热浪风险陡增。本研究突破传统方法在气象要素非线性耦合及地形反馈表征的局限，为流域水文模拟、农业气候风险评估及碳中和路径优化提供了公里级多变量数据基础，显著提升区域适应策略的精准性。

Global warming has exacerbated extreme weather events, posing severe challenges to regional climate and hydrological ecology, while the insufficient spatial resolution of existing CMIP6 data restricts fine-grained assessments. Aiming to address the issue of insufficient spatial resolution for multiple climate indicators and potential evapotranspiration (PET), this study developed a super-resolution fusion framework based on deep residual attention mechanism, and constructed the first-ever 10-kilometer resolution multi-model and multi-scenario high-resolution climate and PET dataset for China (SRCPCN10). Using century-scale data from 5 CMIP6 models under SSP126 to SSP585 scenarios, collaborative downscaling was achieved for 9 climate variables including air temperature (Tas), precipitation (Pr), radiation (Rsds/rlds), as well as PET. The Residual Channel Attention Network (RCAN) exhibited excellent downscaling performance for variables such as air temperature, radiation and air pressure (R²/KGE > 0.99), while the spatial discontinuity of precipitation led to significantly lower downscaling accuracy (R²=0.897). This study revealed that future climate indicators in China show significant emission gradient responses under SSP scenarios, with the increments of air temperature, radiation and precipitation increasing along the gradient of enhanced radiative forcing. Extreme climate events exhibit strong spatial heterogeneity: hydrological drought continues to deteriorate in the Northwest Arid Region, while the risks of floods and heatwaves in South China have risen sharply. This study broke through the limitations of traditional methods in nonlinear coupling of meteorological variables and topographic feedback characterization, providing a kilometer-scale multi-variable data basis for watershed hydrological simulation, agroclimatic risk assessment and carbon neutrality pathway optimization, and significantly improved the accuracy of regional adaptation strategies.

搜集汇总

背景与挑战

背景概述

该数据集是中国首套10公里超分辨率逐日气候变化及潜在蒸散量数据，覆盖2015-2100年，基于深度残差注意力机制融合框架构建，包含气温、降水、辐射等9个气候要素和潜在蒸散发，数据来源于5个CMIP6模式在SSP126-SSP585情景下，旨在提升区域气候评估的精细化水平，支持水文模拟和农业气候风险分析。