An unexploited opportunity in spaceborne cloud radar is the use of the surface reflection to constrain the CLWP. Exploiting this signal requires care in properly characterizing the surface roughness and the effects of gaseous attenuation on the radar beam. We have spent a great deal of effort to understand and characterize the errors resulting from these effects over ocean surfaces. Here we show how the surface reflection signal from CloudSat can be used to provide an estimate of the CLWP that has low bias in the absence of precipitation. Key to the approach are the precise cloud detection of the CALIPSO lidar and the precise precipitation detection of the CloudSat radar. We develop a climatology of CLWP using this surface refelctance approach. We demonstrate how the method is under-constrained in the presence of precipitation. By estimating physical bounds on the liquid water content in precipitating clouds we show that the treatment of precipitating pixels can affect the climatology by a factor of 2. Despite the uncertainty related to precipitation in the radar technique, we are able to conclude that the passive microwave climatologies of CLWP provide a better regional distribution of the CLWP than do reflected solar techniques. However, it is not yet possible to assess the absolute accuracy of the passive microwave data in regions of intense precipitation.
The data is used to specifically evaluate CLWP from the coincident MODIS-Aqua products. We find that the regional distribution of CLWP produces from radar surface reflection is correlated well correlated with a sub-adiabatic model of the CLWP that has the physically expected dependence on cloud thickness. This dependence is not found in observations from MODIS suggesting 3D radiative transfer effects may systematically reduce the CLWP from reflected solar methods. Secondly, we find that there is systematically more condensed liquid in clouds beneath cirrus than in cirrus-free conditions resulting in an ice-cloud masking bias in the MODIS CLWP product. Finally, we compare the method to partially cloudy MODIS pixels. These cloud-edge pixels are shown to have a substantial bias in the MODIS products, however they are also shown to have very low CLWP and therfore do not effect the aggregation of data to a large extent.