Analysis Of Aerosol-Cloud Interaction And Its Radiative Forcing From Multi-Source Observations

Aerosol-cloud interactions have the largest uncertainty in assessing the anthropogenic contribution to present and future climate change.There still exist large differences in aerosol indirect radiative forcing between the satellite-based and model-based studies.Recent satellite-based studies have even found that an increase in aerosols will increase the size of cloud droplets,which is in contrast to conventional first aerosol indirect effect.Therefore,it is critical to investigate the mechanism of aerosol-cloud interaction in detail and realize the limitations of current observations in evaluating aerosol indirect forcing from the perspective of multi-platform observations.By using in situ aircraft measurements,satellite remote sensing as well as reanalysis data,we analyze aerosol–cloud relationships for different cloud types and over different regions,and explore the physical mechanism behind them.We also improve the current satellite-based method for estimating first indirect radiation forcing(RF_aci),and then reassess the RF_aci on a global scale.The main results are as follows:(1)The satellite-based results indicate that cloud effective radius(CER)is positively correlated with aerosol optical depth(AOD)/aerosol index(AI)over land(positive slopes),but negatively over oceans(negative slopes)even with small ranges of liquid water path(quasi-constant).The changes in albedo at the top of the atmosphere(TOA)corresponding to aerosol-induced changes in CER also lend credence to the authenticity of this opposite aerosol–cloud correlation between land and ocean.It is noted that potential artifacts,such as the retrieval biases of both cloud(partially cloudy and 3-D-shaped clouds)and aerosol,can result in a serious overestimation of the slope of CER–AOD/AI.Our results show that collision–coalescence seems not to be the dominant cause for positive slope over land,but the increased CER caused by increased aerosol might further increase CER by initializing collision–coalescence,generating a positive feedback.By stratifying data according to the lower tropospheric stability and relative humidity near cloud top,it is found that the positive correlations more likely occur in the case of drier cloud top and stronger turbulence in clouds,while negative correlations occur in the case of moister cloud top and weaker turbulence in clouds,which implies entrainment mixing might be a possible physical interpretation for such a positive CER-AOD/AI slope.(2)Our analysis on marine stratocumulus(VOCALS-REx)suggests that sub-cloud cloud condensation nuclei(CCN)number concentration is positively correlated with liquid water content(LWC)and droplet number concentration(N_d),but weakly correlated with cloud effective radius(R_e),indicating that it is essential to distinguish Twomey effect from the rapid adjustment of cloud by constraining LWC when using R_e as a cloud quantity.It is also found that an increase in aerosol concentration tends to simultaneously increase both N_d and relative dispersion(?),while an increase in vertical velocity(w)often increases N_d but decreases?.After constraining the differences of cloud dynamics,the positive correlation between?and N_dbecomes stronger,implying that perturbations of w could weaken the aerosol influence on?and hence result in an underestimation of dispersion effect.A comparative analysis of the difference of cloud microphysical properties between the entrainment and non-entrainment zones suggests that the entrainment-mixing mechanism is predominantly extremely inhomogeneous in the stratocumulus.(3)The analysis on continental stratocumulus and cumulus(RACORO)shows that with increasing CCN,N_d significantly increases in stratocumulus(Sc)while remains almost unchanged in cumulus(Cu).The increases of LWC and R_e,and the decrease of?with aerosols increasing are both found for Sc and Cu,but the extent of changes for the latter is less notable.By using a new approach to strictly constrain the dynamics in Cu,we found that neither simultaneously changing cloud dynamics nor dilution of cloud water induced by entrainment-mixing can explain the observed insensitivity of N_d.The different degree of reduction in cloud supersaturation caused by increasing aerosols might be responsible for the observed different aerosol indirect effect between Sc and Cu.(4)As known,AOD can only be retrieved on clear pixels.In the case that the clouds fully cover the pixel,AOD retrieval is missing,and thus these clouds will not be sampled for analysis,i.e.,sampling biases.It is thus expected that stratiform clouds with high cloud fraction(CF),which exerts much stronger colling effect,would be artificially and systematically excluded in current satellite-based investigations.By accounting for the sampling biases,the slope of ln N_dversus ln AOD increases,and the estimated RF_aci increases by 43%,which is closer to model not only in global-averaged value(-0.67 w m^-2)but also in its spatial distribution.This finding implies that previous satellite-based estimates have substantially underestimated the RF_aci,and further highlights the necessity of accounting for the sampling biases in future satellite investigations.