Figure 7. Vertical profiles of the cloud droplet effective radius from observation (solid line) and a 3.7 /m retrieval algorithm (dashed line).

re usually varies with height. As illustrated in Fig. 7, retrieval at a single near-IR channel like 3.7 ¡m may only represent a shallow layer near the cloud top.

Chang and Li (2002, 2003) have developed a new satellite retrieval method to account for the vertical variation of the re profile. It employs a combination of multiple near-IR spectral measurements at the 3.7 ¡m, 2.1 ¡m, and 1.6 ¡m channels, available from MODIS, to retrieve a linear re profile. The linear re profile is assumed to be a linear function of the in-cloud optical depth (t'), which is given by


where Ttotal is the total cloud optical depth and and re_base represent the re at the cloud top (t' = 0) and the cloud bottom (t' = Ttotal), respectively. Its retrieval principle lies in the fact that photons at the three near-IR wavelengths are sensitive to different depths within a cloud. This method is applicable to single-layer water clouds and is most effective for clouds with near-linear variation of re from cloud top to cloud base (Miles et al., 2000). However, the estimation of re for the cloud base would have a large bias if the re profile is nonlinear or if the cloud is thick, because the re signal from the cloud base becomes weak (Chang and Li, 2002).

Based on the linear re profile, Chang and Li (2003) show that the calculation of cloud

0 0

Post a comment