Clouds play a key role in regulating the Earth's climate. Real cloud fields are non-uniform in both the morphological and microphysical sense. However, most climate models assume the clouds to be Plane-Parallel Horizontal (PPH) plates with homogeneous optical properties. Three characteristics of 3D clouds have been found to be important for longwave radiative transfer. They are: (1) the 3D geometrical structure of the cloud fields, (2) the horizontal variation of cloud optical depth, and (3) the vertical variation of cloud temperature. One way to incorporate the 3D geometrical effect in climate studies is through the use of an effective cloud faction, for which a major component is the Probability of Clear Line Of Sight (PCLOS). The PCLOS also plays an important role in accounting for longwave 3D effects caused by variable cloud optical depth and vertical change of cloud temperature.

Aimed at improving the parameterization of longwave radiative transfer through 3D clouds, this study formulated a set of PCLOS models and tested the models with the Atmospheric Radiation Measurement (ARM) cloud observations.

In order to investigate the sampling issue that arises from attempting to obtain domain-averaged information from time series of observations, an evaluation technique was developed and tested with Cloud Resolving Model (CRM) and Large Eddy Simulation (LES) model data.

Various cloud properties that are necessary for the PCLOS models such as the absolute cloud fraction (N), cloud thickness, cloud spacing, and horizontal size were inferred from the ARM observations. A set of automated inference techniques were developed. The modeled PCLOS was then tested with the PCLOS inferred from time series of total sky images.