大气科学学报2025，Vol.48Issue(5)：736-751,16.DOI:10.13878/j.cnki.dqkxxb.20241014001
不同云种类地表云短波辐射效应的主控因子及其不确定性研究——以美国南部大平原地区中心观测站为例
Dominant factors and uncertainties of surface shortwave cloud radiative forcing under different cloud types based on measurements at the central facility of the Southern Great Plains,USA
摘要
Abstract
Clouds play a crucial role in modulating the earth-atmosphere energy budget through their radiative cooling and heating effects.They are among the most significant yet uncertain components in climate projections due to their diverse types and complex cloud-radiation interactions.The surface shortwave cloud radiative effect(FCRE),defined as the difference between all-sky and clear-sky surface shortwave radiation,serves as a key indi-cator of the cloud-induced cooling effect.However,FCRE exhibits considerable variability and uncertainty across cloud types,attributable to large differences in macrophysical,microphysical,and optical cloud properties.Moreo-ver,the dominant factors controlling these effects remain insufficiently understood for different cloud types.This study analyzes the FCRE characteristics and their driving factors for eight distinct cloud types-shallow cumulus(Cu),stratus(St),altocumulus(Ac),altostratus(As),cirrostratus(Cs),cirrus(Ci),congestus(Co),and deep convective clouds(Dc)—using 14 years of 15-minute resolution cloud-type classification and concurrent surface shortwave radiation measurements from the central facility of the Southern Great Plains in the United States.A new metric,relative FCRE(FRCRE),is introduced to complement FCRE and to reduce angular dependence effects.Using the XGBoost(extreme gradient boosting)framework and SHAP(Shapley additive explanation)values,the study interprets the dominant factors and uncertainties associated with both FCRE and FRCRE across cloud types.Key findings include:1)While FCRE magnitudes exhibit statistically significant positive correlations with cloud fraction and albedo and a negative correlation with solar zenith angle across all cloud types,detailed analysis reveals complex,nonlinear relationships that vary by cloud type.2)At the 95％confidence level,FCRE amplitudes range from-350 to 206 W·m-2 for low cloud fraction/albedo clouds(Cu,Ac,and Ci),from-926 to 371 W·m-2 for intermediate types(St,Cs,As),and from-1 172 to 289 W·m-2 for high cloud fraction/al-bedo clouds(Co and Dc),indicating substantial inter-cloud-type variability.3)The proposed FRCRE metric-func-tionally similar to FCRE-has three dominant predictors:cloud fraction,cloud albedo,and direct irradiance.FRCRE minimizes the influence of solar zenith angle and reduces uncertainty by 2.6％to 66.0％(mean reduction of 10.7％)compared to FCRE,with the most notable improvement observed for convective cloud types.4)For high cloud fraction/albedo clouds(Co and Dc),cloud albedo is the dominant driver of FRCRE,whereas for low cloud fraction/albedo clouds(Cu,Ac,and Ci),cloud fraction is more influential.These results highlight both the simi-larities and differences between FCRE and FRCRE,demonstrate the value of FRCRE in reducing uncertainty in estima-ting cloud radiative effects,and offer a refined understanding of cloud-radiation interactions.This study provides important insights for improving the representation of cloud-radiative processes in climate models.关键词
短波云辐射效应/云种类/主控因子/不确定性分析/可解释性机器学习模型Key words
shortwave cloud radiative effect/cloud type classification/dominant factors/uncertainty analysis/in-terpretable machine learning model引用本文复制引用
刘唯佳,韩永翔,周鑫,李嘉欣..不同云种类地表云短波辐射效应的主控因子及其不确定性研究——以美国南部大平原地区中心观测站为例[J].大气科学学报,2025,48(5):736-751,16.基金项目
中国博士后基金项目(2024M751475) （2024M751475）
浙江省自然科学基金项目(LQ23D050001) （LQ23D050001）
