(a) (b) |
It is always interesting to check the consistency of cloud measurements between active and passive sensors. MLS IWC is derived from 240 GHz radiance measurements in the upper troposphere where emission and scattering of ice crystals can both induce a detectable excessive radiance at limb. The scattering part of the signal is mostly from forward scattering whereby clouds act as a low quality lenses to broaden up the collimated MLS beam. On the other hand, CloudSat 94-GHz radar relies on backscattering of ice crystals to detect clouds, in which ice emission/absorption serves as attenuation and does not contribute to the cloud signal.
In the following examples (Figures 1a and 1b), roles of ice crystal sizes are characterized by the contribution functions of extinction and scattering coefficients at different size diameters. As expected, their roles depend on sensor’s frequency and particle size distribution since the contribution function is a result of convolution between the Mie efficiency and particle size distribution (which is the parameterization by McFarquhar and Heymsfield 1997).
In case (a) the 94-GHz and 240-GHz cloud extinction is dominated by ice emission, not by scattering, because of a large number of small particles. In this case CloudSat will have difficulty to detect clouds unless there is a large amount of ice masses. The scattering becomes dominated at 640 GHz, which is the main motivation for high-frequency microwave remote sensing of cirrus. If ice particles are high bimodal as suggested by the MH97, a better sensor will have sensitivities to both small and large particles.
In case (b), cloud extinction is dominated by scattering for all the three frequencies. However, the lower frequency sensors still lack sensitivity to small particles. Therefore, these sensors will depend more on particle size assumptions to infer cloud ice amount.
| (a) |
| (b) |