Zh Chn ,Minghui Wng ,* ,Hipng Zhng ,Shuhng Lin ,Zhun Guo ,Yiqun Jing ,Chn Zhou
a Institute for Climate and Global Change Research, School of Atmospheric Sciences, Nanjing University, Nanjing, China
b Department of Atmospheric and Oceanic Science & Earth System Science Interdisciplinary Center, University of Maryland, College Park, MD, USA
c School of Atmospheric Sciences, Sun Yat-sen University, Zhuhai, China
d Southern Marine Science and Engineering Guangdong Laboratory (Zhuhai), Zhuhai, China e State Key Laboratory of Numerical Modeling for Atmospheric Sciences and Geophysical Fluid Dynamics, Institute of Atmospheric Physics, Chinese Academy of Sciences,Beijing, China
Keywords:Low-cloud cover Large-scale circulation Hadley cell Pacific walker circulation
ABsTRACT Southeast China has comparable stratus cloud to that over the oceans,especially in the cold seasons (winter and spring),and this cloud has a substantial impact on energy and hydrological cycles.However,uncertainties remain across datasets and simulation results about the long-term trend in low-cloud cover in Southeast China,making it difficult to understand climate change and related physical processes.In this study,multiple datasets and numerical simulations were applied to show that low-cloud cover in Southeast China has gone through two stages since 1980 —specifically,a decline and then a rise,with the turning point around 2008.The regional moisture transport plays a crucial role in low-cloud cover changes in the cold seasons and is mainly affected by the Hadley Cell in winter and the Walker Circulation in spring,respectively.The moisture transport was not well simulated in CMIP6 climate models,leading to poor simulation of the low-cloud cover trend in these models.This study provides insights into further understanding the regional climate changes in Southeast China.
Low cloud has a striking cooling effect on climate systems because of its extensive coverage and effective reflection of solar radiation.The radiative effects of low cloud are supported by results from GCMs,on the basis of which Slingo (1990) estimated that a 4% increase in the amount of low cloud could offset the warming induced by a doubling of the CO2concentration.Besides radiative effects,low cloud has nonnegligible impacts on agriculture and the efficiency of renewable energy(i.e.,solar radiation) via its role in the hydrological and energy cycles,and thereby has an influence on the daily lives of human populations and associated socioeconomic factors (Köhler et al.,2017).
Despite its importance,the simulation of low cloud remains one of the largest uncertainties in current numerical models,especially in Southeast China,where continental stratiform cloud dominates in boreal winter and spring (i.e.,the cold seasons) (Ma et al.,2018).Therefore,a more comprehensive understanding of the long-term changes in low cloud during the cold seasons in Southeast China —a region where there is both a high human population density and large amounts of low cloud —will help greatly towards recognizing local climate changes and subsequently improving their predictability.
Although many studies have investigated low-cloud changes in Southeast China,the results of these studies pertaining to the trend of low-cloud cover are inconsistent.The discrepancies in annual lowcloud cover trends in Southeast China among these studies are partly owing to the disparities in cloud retrieval methods from different data sources.Several studies have been carried out to uncover the physical mechanisms relating to low cloud in Southeast China,mostly in association with the changes in large-scale circulation induced by global warming (e.g.,Song et al.,2022).These changes are expected to alter the atmospheric stability,subsidence strength,and moisture availability,and hence affect low-cloud formation/dissipation.For instance,Yu et al.(2004) proposed that destabilization and desaturation in the boundary layer under surface warming lead to a decrease in stratus cloud.Myers and Norris (2013) suggested that stronger subsidence is unfavorable for low-cloud formation by reducing the liquid water path.On the other hand,low-cloud cover changes over Southeast China might be related to the local Hadley circulation with its sinking motion close to Southeast China,which could modulate low-level moisture supplies for low-cloud formation (Murakami,1981).However,this mechanism has yet to be examined.
The present understanding in terms of how low cloud changes in Southeast China is still insufficient,mostly limited to statistical analysis.Therefore,more effort is needed to obtain a deeper understanding of low-cloud changes in Southeast China,especially at a processoriented level.To this end,in this study,we use multiple datasets to examine the long-term changes in low-cloud cover in Southeast China during the cold seasons and investigate the mechanism of low-cloud cover changes through numerical experiments with GCMs.Section 2 details the datasets,model experiments,and methods.Section 3 determines the long-term trend in low-cloud cover and investigates the underlying mechanisms.Finally,a summary and discussion are given in Section 4.
2.1.Data
A range of data from satellite observations and reanalysis products were used to detect the long-term trend in low-cloud cover in Southeast China.First,we adopted the empirically corrected satellite records from the ISCCP dataset (Norris and Evan,2015),covering the years 1982 to 2009 with a 2.5° × 2.5° horizontal resolution.A more recent dataset (2003—2017) from CERES (Wielicki et al.,1996),which has a 1°×1°horizontal resolution,was also used.Reanalysis data from ERA5(Hersbach et al.,2020) were employed in our study for a longer period,i.e.,from 1980 to 2017,with a 0.75°×0.75°horizontal resolution.Furthermore,results from seven AMIP model experiments (Gates et al.,1999) from CMIP6 (Eyring et al.,2016) were selected to evaluate the climate model simulations of low cloud in Southeast China.The lowcloud cover was calculated from these model results by determining the maximum cloud cover below 700 hPa.
2.2.Experiment design
To investigate the impact of large-scale circulation on the long-term changes in low-cloud cover in Southeast China,a nudging experiment using CESM2 (Danabasoglu et al.,2020) was performed from 1980 to 2017,which is similar to the standard AMIP experiment but with wind fields nudged toward MERRA2 (Gelaro et al.,2017) reanalysis data to represent more realistic large-scale circulation.The experiment details are described in Yue et al.(2021).
2.3.Moisture budget analysis
Following Kim and Ha (2015),a moisture budget analysis was applied in our study,and the moisture budget equation is written as
wherePis precipitation,Eis evaporation,TPW is total precipitable water,and -∇ ·Qis the convergence of vertically integrated moisture flux(CVIMF).CVIMF consists of three terms,as follows:
2.4.Index definitions
2.4.1.NorthedgeoftheHadleycell
The mean meridional mass stream-function (MMS) is used to characterize the Hadley Cell (Holton and Hakim,2013),and its north edge is identified as the latitude where MMS equals 0 kg s-1between 20°N and 50°N,calculated by linear interpolation.
2.4.2.Westedgeofthepacificwalkercirculation
The Pacific Walker Circulation (PWC) is expressed in terms of the zonal mass stream function as defined in Ma and Zhou (2016),and the west edge is identified as the longitude where the zonal mass stream function equals 0 kg s-1.
3.1.Long-term trend in low-cloud cover
The long-term changes in low-cloud cover in Southeast China during the cold seasons are shown in Fig.1 (a,b).The evidence derived from the two satellite datasets and ERA5 demonstrates a shift in the trend around the year 2008.The ISCCP dataset,mostly covering the former period,exhibits an obvious decreasing trend in low-cloud cover both in boreal winter (DJF) and spring (MAM),whereas an increasing trend is apparent in the latter period according to CERES.Although the absolute low-cloud cover values from ERA5 are lower,possibly due to the different definitions of low-cloud cover between ERA5 and the satellite data,they have little influence on our conclusion that the low-cloud cover declined before 2008 and then increased.Therefore,the results from the satellite data are generally consistent with those from the long-term ERA5 data.
3.2.Simulations of low-cloud cover by CMIP6 models and the nudging experiment
We further evaluate the performance of the AMIP simulations from seven CMIP6 climate models using the ERA5 reanalysis data to elucidate the physical mechanisms behind the low-cloud cover changes in Southeast China.Most of these CMIP6 models show little change in low-cloud cover from 1980 to 2017 (Fig.1 (c,d)).In DJF,all the models poorly reproduce the low-cloud cover,with correlations less than 0.5 and none statistically significant,and they perform much worse in MAM,even with some negative correlations (Fig.2 (c,d)).In short,these models cannot simulate the long-term changes in low-cloud cover well,let alone the turning point in the trend accurately.
To identify the main cause of the biases between the model results and reanalysis data,we examine the moisture transport,which is critical for cloud formation.The moisture transport is defined as the CVIMF(see Section 2.3),which is controlled by the specific humidity (q) and the wind field (UandV).As expected,there are commonly high correlations between the low-cloud cover and these four variables in most of the CMIP6 experiments,particularly in DJF (Fig.2 (a,b)).When these simulated variables are closer to the reanalysis data,like in the case of FGOALS-f3-L and MIROC6,such models simulate a more realistic lowcloud cover (Fig.2 (c,d)).This implies that a model is expected to better simulate the long-term changes in low-cloud cover if a more accurate CVIMF could be reproduced.
Fig.1.Time series of low-cloud cover (LCC) in Southeast China derived from ISCCP,CERES,ERA5,the CESM2 AMIP experiment,and the nudging experiment in(a) DJF and (b) MAM.(c,d) As in (a,b) but derived from selected CMIP6 experiments.Raster plots show the five-year moving linear trends in LCC derived from the CMIP6 results in (e) DJF and (f) MAM.The black squares indicate that the trend has passed the significance test at the 0.05 level.
To examine whether realistic wind fields help simulate the changes in low-cloud cover,a modified AMIP experiment with the wind fields nudged toward MERRA2 (nudging experiment) is compared with the standard AMIP experiment,both conducted by CESM2.Although the nudging method weakens the feedback between the cloud and largescale circulation,the long-term trend in low-cloud cover can still be reproduced well,as demonstrated by a substantial improvement in the simulation of changes in low-cloud cover in the cold seasons in Southeast China (Fig.1 (a,b)),with the correlation coefficients exceeding 0.8 significantly.This improvement indicates that large-scale circulation may be more important than other factors in simulating the long-term changes in low-cloud cover in Southeast China,which is the direction we follow next to elucidate the related physical mechanisms.
3.3.Physical mechanisms of low-cloud changes
As the CESM2 nudging experiment can capture the observed lowcloud cover trend well,this experiment is used to investigate the potential physical mechanisms underlying the long-term changes in low-cloud cover in Southeast China in the cold seasons.Moisture transport is an important factor for low-cloud cover in Southeast China,and CVIMF is used in our study to examine the effects of moisture transport due to large-scale circulation.CVIMF primarily consists of two terms: R1,dictated by the large-scale convergence,and R2,dependent on the horizontal wind advection and the distribution of the specific humidity.First,we calculated the percentages of variance explained by R1 and R2 in a multiple linear regression of CVIMF,as exhibited in Fig.3.In Southeast China (the purple rectangle),R1 and R2 play equally important roles in CVIMF in DJF (Fig.3 (a,c)),whereas in MAM (Fig.3 (b,d)),the scale tips to the former side,with R1 accounting for more than 70%.As such,next,we investigate how the large-scale circulation affects the changes in low-cloud cover through CVIMF —specifically,its two components in DJF and R1 in MAM —to understand the long-term changes in low-cloud cover in Southeast China during the cold seasons.
3.3.1.Low-cloudcoverchangesinDJF
Considering the vital role of large-scale subsidence in the maintenance of low cloud in Southeast China,the sinking branch of the Hadley Cell around this region is likely responsible for the long-term changes in low-cloud cover in DJF.Using reanalysis datasets and outgoing longwave radiation records,Hu and Fu (2007) depicted a significant poleward expansion of the Hadley Cell,implying a poleward shift of the north edge of subtropical subsidence.The Hadley Cell’s expansion can be explained by the idealized model proposed by Held and Hou (1980) that describes the range of the Hadley Cell as,whereHis the height of the tropical tropopause,ais the radius of the earth,ΔHis the fractional temperature difference between the equator and pole,Ω is the angular velocity of the earth,andgis the gravitational acceleration.Hhas risen by several hundred meters as the climate has warmed since 1979,suggesting a widening of the Hadley Cell.Such changes can exacerbate midlatitude warming and aridness,resulting in a decrease in CVIMF.
Fig.2.Bar charts of the correlation coefficients of low-cloud cover with CVIMF,specific humidity (q),zonal wind (U),and meridional wind (V) on the 850 hPapressure level derived from seven CMIP6 experiments and the nudging experiment during 1980—2014 in Southeast China in (a) DJF and (b) MAM.The slash patterns of the bars indicate a significant result at the 0.01 level.The Taylor diagrams display a statistical comparison of low-cloud cover (L),CVIMF (C),specific humidity(q),zonal wind (u),and meridional wind (v) changes in (c) DJF and (d) MAM between seven CMIP6 experiments or the nudging experiment and ERA5 with the latter as a reference.The black circular outlined letters indicate a significant result at the 0.01 level.
Fig.3.The percentage variance explained by (a,b) R1 and (c,d) R2 in a multiple linear regression of CVIMF in (a,c) DJF and (b,d) MAM from 1980 to 2017 derived from the nudging experiment.The purple rectangular frame denotes the region of Southeast China (23°—27°N,108°—118°E).The black dots indicate that the statistical analysis has passed the significance test at the 0.01 level.
Fig.4.Correlation coefficients between the vertical velocity at different height levels and (a) the north edge of the Hadley Cell (red dots) or CVIMF (green dots) in DJF in Southeast China.(b) As in (a) but for the west edge of the PWC (red dots) or CVMIF (green dots) in MAM in Southeast China (circles) and the SCS (triangles).The shapes outlined in black indicate a significant result at the 0.01 level.
Our nudging experiment shows that the north edge of the Hadley Cell experiences an upward and then a downward trend,with the turning point around 2007,as the low-cloud cover changes (Fig.S1(a)).To clarify the role of the Hadley Cell in the long-term changes in lowcloud cover,we examine the correlation of the large-scale vertical velocity with the north edge of the Hadley Cell and CVIMF at different height levels (Fig.4 (a)).Around the levels where low cloud forms,the large-scale vertical velocity is positively correlated with the Hadley Cell but negatively correlated with CVIMF.This reveals that when the north edge of the Hadley Cell shifts polewards in response to global warming,large-scale subsidence would increase in the lower troposphere,most likely due to the structure and intensity adjustment of the expansion of the Hadley Cell.The enhanced sinking prevents moisture near the surface from transporting upwards,decreasing the low-cloud cover there.When the north edge of the Hadley Cell shifts back,all these effects will reverse,i.e.,warm and moist air accumulates in the low-level atmosphere,leading to an increase in low-cloud cover.These processes explain how the shift in the north edge of Hadley Cell influences the long-term trend in wintertime low-cloud cover in Southeast China via CVIMF in the lower troposphere.
3.3.2.Low-cloudcoverchangesinMAM
A study of springtime rainfall (Li et al.,2016) suggested that a local circulation linking precipitation from the South China Sea and the Philippine Sea (SCS) (0°—15°N,110°—155°E;red rectangle in Fig.S2)with that from Southeast China may play a crucial role in the moisture supply for springtime low cloud in Southeast China.The change in the local circulation is likely related to the movement of the tropical PWC.
We first examine the change in the west edge of the PWC (the updraft branch,PWC_W),finding that it experiences a shift in around 2008 during the period 1980 to 2017.During this period,the PWC_W moves westward until it reaches 130°E (around the core of the SCS) near 2008,and then returns eastward (Fig.S1(b,c)).The movement of PWC_W contributes to changes in the large-scale vertical velocity over the SCS,supported by high correlations between the large-scale vertical velocity at several height levels and PWC_W (Fig.4 (b)).Meanwhile,as PWC_W moves back and forth,the precipitation over the SCS changes accordingly.When there are stronger upward motions over the SCS in response to the westward shift of PWC_W from 1980 to 2008,the precipitation there would increase due to stronger moisture convergence.After 2008,owing to the shift back of PWC_W,the precipitation would return to the original level.The consistent variations in the precipitation,CVIMF,and the convergence term R1 are demonstrated by Fig.S1.
For the former period (before 2008),the enhanced precipitation over the SCS because of the westward shift in PWC_W will release more latent heat,boosting the local circulation between the SCS and Southeast China.As a result,the falling branch of this local circulation located around Southeast China would also become stronger,supported by the significant negative correlations between the vertical velocity at several levels and PWC_W in Southeast China (Fig.4 (b)).This change leads to a divergence anomaly of the moisture in the lower troposphere (Fig.4 (b))and then a decline in the precipitation and thereby the low-cloud cover.These processes are manifested as opposite changes in precipitation,CVIMF and R1 in Southeast China (Fig.S1(b)) compared to those over the SCS (Fig.S1(c)).The fact that PWC_W has an opposite effect on the moisture divergence over the SCS and Southeast China is also demonstrated by Fig.S2(b).For the latter period (after 2008),when PWC_W undergoes the eastward shift,the local circulation will not be stimulated to the same extent as it was in the previous period,and the environment in Southeast China returns to being more suitable for low-cloud formation.
Combining the results above,what happened over these areas during 1980 to 2008 can be explained by the fact that,under global warming,PWC_W tends to move westward,which may enhance the vertical velocity or moisture convergence over the SCS.The stronger updrafts are conducive to the development of deep convection,leading to more precipitation over this ocean.More latent heat release reinforces a local circulation between Southeast China and the Philippine Sea,adding up to a subsidence anomaly in Southeast China.Over that region,its concomitant divergence of moisture transport is unfavorable for low clouds to form,and thus the low-cloud cover tends to decline in this period.During 2008—2017,the eastward shift in PWC_W reverses these variables,leading to the recovery of low-cloud cover in Southeast China.
This study examined changes in low-cloud cover in Southeast China using three datasets (ISCCP,CERES,ERA5) in combination with climate model experiments.We found that,before 2008,low-cloud cover decreased significantly from December to May (DJF and MAM);whereas after 2008,it generally increased.Our nudging experiment suggests that the moisture transport controlled by the large-scale circulation plays a crucial role in changes in low-cloud cover.
The changes in the Hadley Cell and PWC are the main causes of the long-term changes in low-cloud cover in DJF and MAM,respectively.In DJF,global warming shifts the north edge of the Hadley Cell polewards,strengthens the subsidence,and keeps the water vapor from entering the layer where low cloud forms.In MAM,the westward shift of the upward branch of the PWC enhances a local circulation by intensifying springtime deep convection over the SCS.The downflow of this local circulation locates around Southeast China and blocks the moisture supply for the formation of low cloud,leading to a decrease in low-cloud cover.
This study attributes changes in low-cloud cover in Southeast China during cold seasons to changes in moisture transport from large-scale circulation.Other factors,such as aerosols and changes in local stability may potentially affect low-cloud cover as well.Ding et al.(2021) found that biomass-burning aerosols aloft enhances low-cloud formation,but biomass-burning emissions did not show a significant trend shift around 2008,indicating that aerosol may not serve as the primary factor for long-term changes in low-cloud cover.As for local stability,our results based on an the CESM2 AMIP experiment showed positive correlations between lower-tropospheric stability,low-cloud cover and surface air temperature,but when we turned offthe radiative effects of clouds in the model,the correlation between low-cloud cover and surface air temperature became negligible,indicating that the strong correlations among lower-tropospheric stability,low-cloud cover and surface air temperature are caused by cloud radiative effects but not by the effects of lowertropospheric stability on low-cloud cover (not shown).While these effects may not be important for explaining historic long-term changes in low-cloud cover in Southeast China since the 1980s,future changes in low-cloud cover in Southeast China may involve changes in aerosols and local stability,which warrants further investigation in the future.
Funding
This work was supported by the Ministry of Science and Technology of China [grant number 2017YFA0604002],the National Natural Science Foundation of China [grant numbers 41925023,41575073,41621005,and 91744208 ],and the Collaborative Innovation Center of Climate Change,Jiangsu Province.
Acknowledgments
We thank the High Performance Computing Center of Nanjing University for providing the computational resources used in this work.We acknowledge the World Climate Research Programme for the promotion of CMIP6 through its Working Group on Coupled Modeling,and the Earth System Grid Federation for the production and availability of the datasets.The CESM code used in our work is available at(https://www.cesm.ucar.edu/models/cesm2/release_download.html).The ISCCP data can be obtained from the NARS Atmospheric Science Data Center (https://asdc.larc.nasa.gov/project/ISCCP).The CERES SSF1deg product was accessed from the NASA Langley Research Center CERES ordering tool (https://ceres.larc.nasa.gov/).The ERA5 data were obtained from https://www.ecmwf.int/en/forecasts/datasets/reanalysis-datasets/era5.
Supplementary materials
Supplementary material associated with this article can be found,in the online version,at doi: 10.1016/j.aosl.2022.100222.