Role of Cloud Subgrid‐Scale Structure in Modulating Clouds Viewed by ISCCP, MODIS, and MISR Simulators

Abstract

AbstractThe role of cloud subgrid‐scale structure in modulating satellite views of clouds was investigated. This was realized by implementing a stochastic cloud generator into the CFMIP (Cloud Feedback Model Intercomparison Project) Observation Simulator Package together with surrogate clouds produced by a cloud‐resolving model (CRM). The subgrid‐scale structural parameters are decorrelation length for overlapping cloud fraction (Lcf), decorrelation length for overlapping cloud condensate (Lcw), and the shape parameter v for measuring cloud inhomogeneity. With the use of median values of Lcf, Lcw, and v derived from CRM, the simulated satellite views bear close resemblance to those using CRM‐inherent clouds. Varying these parameters in the range of lower and upper quartiles leads to differences that are about one‐fifth of those caused by changing cloud microphysics in CRM. While Lcf influences clouds throughout the whole troposphere, Lcw and v result in changes mostly within the upper layers. Increasing (decreasing) cloud inhomogeneity or overlapping degree leads to decreased (increased) occurrence of clouds, except for high‐topped clouds viewed by Multiangle Imaging Spectroradiometer. Care must then be exercised when interpreting model biases in comparison with different instruments. Sensitivity tests show changing condensate distribution from gamma to lognormal makes little impact on final results. Although the differences induced by any of the parameters alone are much limited, they are getting comparable to those seen between models and observations when all parameters are synergistically altered. This brings encouraging results to the modeling community that simulator‐diagnosed clouds can be potentially improved by tuning cloud subgrid‐scale parameters.