Comparison Of Cloud Liquid Water Path From Different Satellite-Based Measurements

Cloud Liquid Water in the atmosphere is of great significance in global water cycle and energy balance.With the advantages of high frequency and long-term detecting,satellite measurements served as an important component in retrieving cloud liquid water path(LWP).Two of the most common methods to retrieve LWP are passive microwave retrievals and visible/near-infrared retrievals.Comparing the retrievals from different retrieval algorithm is of great importance.In this study,we compared three kinds of LWP retrievals from satellite measurements.They are:LWPm from Wang's advance retrieval algorithm base on observation from Advanced Microwave Scanning Radiometer-EOS(AMSR-E),LWPR from Remote Sensing Systems(RSS)standard AMSR-E database,and LWPv from visible/near-infrared retrieval method base on the observation of Moderate Resolution Imaging Spectroradiometer(MODIS).we compared different retrievals and got the following results:The clear-sky brightness temperature is one of the most important input variables of Wang's retrieval method.First of all,we analyzed the spatial distribution characteristics and change law of brightness temperature at four channels of three frequencies,they are:10.6GHz vertical channel,10.6GHz horizontal channel,23.8GHz vertical channel as well as 36.5GHz horizontal channel.The brightness of four channels show the similar spatial distribution characteristics,which are higher in the tropics,Pacific warm pool and the Indian Ocean,and getting lower with the increase of latitude.The time series of the four channel also show the similar changes.We found that the correlation coefficients between the two lower channels and the two higher channels are very high and over 0.9,however,the correlation coefficients between the higher and lower channels are relative lower.This may due to the fact that different channels have different response to the same meteorological factors.The clear-sky brightness temperatures of each channels show similar changes during the seasonal variation,and seasonal variation is the most important factor to explain their changes.We compared the microwave retrievals LWPM and RSS AMSR-E retrievals LWPr,the results show that both LWPM and LWPR have similar distribution features on the globe oceans:higher values in places like the Pacific equator,the warm pool and the Indian Ocean,like the distribution features of water vapor.However,the mean LWPR is 0.05kg/m2 lower than mean LWPM,and the lower LWPR are concentrated in tropics and Pacific warm pool where the precipitation happened more frequently.This is because that the partitioning of rain-cloud is set to be 0.1Okg/m2 in retrieval algorithm of LWPr and the algorithm of rain-cloud is also empirical which has no guarantee of its accuracy.By removing the rain-cloud samples,we found that the difference between LWPM and LWPR is around-0.01kg/m2,and they may due to the simplified algorithm of cloud temperature in Wentz retrieval method.In non-precipitating clouds samples,there are still some regions with higher LWPM than LWPR.This is because the cloud-rain partitioning of Wentz retrieval algorithm was found to introduce a systematic bias in LWPR above 0.18kg/m2 as the algorithm erroneously assigned an increasing portion of the liquid water content of some thick non-precipitating clouds to rain.Wang's advance retrieval method can directly retrieve from precipitation samples and using infrared temperature as effective radiation temperature.So in theory,LWPM is closer to the true value in most situations.In this study,we also compared MODIS LWPV and LWPR in different conditions partitioned by MODIS cloud products:all the clouds,non-precipitation clouds,warm clouds and non-precipitation warm clouds.By comprehensive comparison and analysis,we found that in all clouds samples,mean LWPv is lower than mean LWPM,and there are two reasons lead to the result:the first is MODIS retrieving limitation of clouds with ice tops leads underestimate in mean LWPv;the second reason is the assumption of vertically homogeneous in MODIS retrievals may leads underestimate in LWPV of rain clouds.After removing the precipitation cloud samples,the area of regions with lower mean LWPv than LWPM is much less,and only concentrate in tropics and the warm pool.In other regions,mean LWPv is much higher than LWPM,which indicates that the overestimate of LWPv.For warm clouds including rain clouds,only in tropics and the warm pool the mean LWPM is higher than that of LWPv.And the regions with higher LWPM is nearly disappearing after removing the rain cloud samples.All of these results suggest that MODIS LWPV is overestimate in many warm non-precipitation clouds,and this is due to the vertically homogeneous assumption of visible/near-infrared retrieval method is not suit for vertical stratification clouds.Finally,this study gives an evaluation on this issue that MODIS' inability in retrieving the LWPv of ice-top clouds.We used the values of LWPM to estimate the LWPMISS of ice-top clouds.The MODIS detection failures are found to be frequent over global oceans,with an overall miss rate at?11.2%and MODIS underestimation on globally averaged LWP is?0.39 kg/m2.The miss rates and mean missed LWP values has a strong positive correlation with cloud ice water path(IWP).Clouds with colder tops are associated with more missed LWP.Among all missed LWPs,the highest miss frequency is found between cloud top temperature CTT 205-235 K and IWP 0.02-0.06 kg/m2,and the largest mean missed LWP is found at CTT<200 K and IWP>0.4 kg/m2.On the other hand,we evaluated how MODIS missed LWP effects on global radiative transfer modeling.The results show that globally radiative forcing due to the missed water clouds introduces a net warming effect to the earth-atmosphere system and the mean radiative force is-198.8 W/m2.