Deriving Non-Precipitating Cloud Water Path Using a Surface Reference Technique from CloudSat

A surface reference technique for the derivation of the integrated non-precipitating cloud liquid water path (W) from CloudSat observations is presented. A benefit of the technique is its simplicity. W is simply a linear function of the Path Integrated Attenuation due to hydrometeors (PIA). Uncertainty in W is nearly entirely governed by uncertainty in the PIA. Significant effort is made to characterize both random and bias components of this uncertainty. As a first step, an interpolation method is used to derive the 2-way Path Integrated Attenuation due to condensed water (PIA) in cloudy pixels. This estimate of the PIA is compared with an independent estimate that relies on an empirical Look Up Table (LUT) of the surface cross section. The comparison identifies biases in the LUT method of approximately 0.5 dB. This level of bias is tolerable when the remote sensing target is precipitation, however it is of the same order of magnitude as the signal from a typical non-precipitating boundary layer cloud. We therefore proceed using only the interpolation method, which we determine to be an imprecise but relatively unbiased estimate of PIA and hence W.

The relative lack of bias in the CloudSat W make it an excellent calibration source for more conventional cloud remote sensing instruments. In particular, W from CloudSat is compared with coincident estimates from the MODerate resolution Imaging Spectroradiometer (MODIS ) and the Advanced Microwave Scanning Radiometer for EOS (AMSR-E). These Correlations between all estimates are strong, however regional biases exceeding 100% are observed. These biases are a strong function of water vapor suggesting significant bias in the passive microwave retrievals. These comparisons are also used to quantify the well known under-catch of the MODIS estimates of W resulting from that instruments failure to retrieve W for very thin, isolated clouds and cloud edges.