Purdue University
West Lafayette, IN 47907-1397
The goals of ISCCP and FIRE are, broadly speaking, to provide methods for the retrieval of cloud properties from satellites, and to improve cloud radiation models and the parameterization of clouds in GCMs. This study suggests a direction for GCM cloud parameterizations based on analysis of Landsat and ISCCP satellite data.
High-resolution Landsat data is degraded to the various resolutions used by ISCCP, and then a threshold used for clear/cloudy determination of pixels at those resolutions. For low-level single- layer clouds it is found that the mean retrieved liquid water path (LWP) in cloudy pixels is essentially invariant to the cloud fraction, at least in the range 0.2 - 0.8 (Figure 8.1). Decreasing the pixel resolution does not alter this conclusion - it only reduces the specific constant value of mean LWP within this range (Figure 8.2). At high cloud fractions (greater than ~0.8, say) the LWP can be considerably higher than the relatively constant value for lower fractions.
It appears that, at any instant, the mean LWP of the cloudy areas is the average of a population of clouds each of which has relatively constant LWP. The number of clouds in the population determines the cloud fraction, but as long as the number is not so large that clouds begin to fill the domain, the mean LWP for the cloudy area is the same. The closed and open cells associated with marine stratocumulus and trade cumuli respectively, are seen to exhibit such a feature. Trade cumuli are much narrower and deeper than are stratocumuli, and hence have a larger mean LWP per cloud. But for either type of cloud, increasing the number of clouds, and hence the cloud fraction, will not change the mean LWP of the clouds unless the clouds begin to saturate the domain.
The constancy of mean LWP with cloud fraction implies that the total volume of liquid in a box is a linear function of cloud fraction (Figure 8.3). This result is very important since it allows the cloud fraction to be estimated if the mean LWP of cloud in a GCM gridcell is known. The prognosed or diagnosed total liquid water in the gridcell can then be distributed according to this cloud fraction. The mean LWP for the clouds needs to be specified either empirically or using some process model.
