Comparative Analysis Of The Impact Of Cloud Cover On The Radiation Budget Of The Earth-Atmosphere System Based On CMIP5 And CMIP

Accurate simulation of the historic spatio-temporal variations in radiation budgets of the Earth-atmosphere system is a fundamental step for understanding and predicting the climate change.Recently,many studies have found that the radiation budgets are improved from Coupled Model Intercomparison Project phase 5（CMIP5）to CMIP6,however,there is no obvious reduction in the inter-model uncertainty.One of the main source of this uncertainty is the cloud effect on the radiation budgets in the Earth-atmosphere system.Based on Clouds and the Earth’s Radiant Energy System（CERES）Energy Balanced and Filled（EBAF）Dataset Edition 4.1,historical spatiotemporal variations in radiation budgets at the top of atmosphere（TOA）and the Earth surface under both all-sky and clear-sky conditions were compared between CMIP5 and CMIP6.Regions with high inter-model variability were also identified in two CMIPs.Then the relationship between radiation components simulation bias and cloud fraction simulation bias were investigated.The radiative effect of cloud was also quantificated in two CMIPs.Thos research aims to quantify the effects of cloud fraction on spatio-temporal patterns of the historical radiation budget in two CMIPs.The main conclusions are summarized as follows:（1）Under clear-sky condition,the ensemble means of radiation components,except outgoing longwave radiation at TOA in CMIP6 are better agreement with CERES and the inter-model uncertainty decreases.The surface downward shortwave radiation is overestimated by3.9 W/m² which decreases by 4.4 W/m² than CMIP5.In CMIP5,the surface downward longwave radiation is underestimated by 3.0 W/m²,which becomes slight overestimation of 0.3W/m²in CMIP6.Globally,the obvious uncertainity is found in clear-sky surface downward shortwave radiation in low latitudes.Simulation bias in other radiation components mainly locates in plateaus,arid regions and Antarctic ocean.Compared to CMIP5,the level and area of inter-model uncertainty in radiation components decreases in CMIP6.（2）Under all-sky conditions,the ensemble means of radiation components,except surface upward longwave radiation in CMIP6 are in better agreement with CERES EBAF 4.1.Except downward longwave radiation at the surface,the lower inter-model spread for other radiation components are found in CMIP6.The overestimation of the global mean surface downward shortwave radiation（2.8 W/m²）in CMIP5 decreases by 1.9 W/m² in CMIP6.And the underestimation of the surface downward longwave radiation(5.0 W/m²)in CMIP5 decreases by 3.3 W/m² in CMIP6..Except the surface downward radiation,the inter-model agreement of radiation components is improved in CMIP6.Globally,the simulation bias of CMIP6 is more obvious than that of CMIP5 in north pole and around 60^oN&S.Meanwhile,except in these areas,the simulation of radiation component in CMIP6 more agrees with the CERES than CMIP5.Compared to CMIP5,the level of inter-model uncertainty decreases in CMIP6,and its area is reduced by 22%～96%.However,the plateau and equatorial regions are still areas with poor inter-model consistency in CMIP6.（3）Compared to CMIP5,the cloud fraction increases by 6.5%with higher inter-model consistency in CMIP6,which is in better agreement with CERES observations.Except for cloud longwave radiative effects at TOA,the cloud radiative effects simulated by CMIP6 is closer to CERES with the better inter-model consistency.Such as the simulation bias of surface downward shortwave decreases by 2.3 W/m² than CMIP5.Globally,both the two-CMIPs’simulation bias of cloud radiative effects more often happens in deserts,plateaus and high-latitude seas.Besides,these areas are also the regions where the simulation inter-model uncertainty of cloud radiative effect locates in.In terms of the interannual variation and spatial patterns,both two CMIPs can reflect the physical relationship that the higher the cloud fraction,the stronger the cloud radiative effect,and the relationship is more significant in CMIP6.The quantitative relationship between the surface cloud shortwave radiative effect and the cloud longwave radiative effect simulated by CMIP6 is closer to the CERES observation.It indicates that the physical process of the impact of low cloud on the surface radiation simulated by CMIP6 is more accurate.