Quantifying Parametric Uncertainty Effects on Tropical Cloud Fraction in an AGCM

Abstract

The underestimation of cloud fraction, especially the low stratus cloud (LSC) fraction over the eastern oceans, remains a problem in most atmospheric global climate models. This study investigated potential improvements through perturbing nine moist physical parameters, using uniform sampling and Latin hypercube sampling methods, and quantified the parametric uncertainty and effects of non-linear interaction between parameters on the tropical cloud fraction in the Grid-Point Atmospheric Model of the Institute of Atmospheric Physics/State Key Laboratory of Numerical Modeling for Atmospheric Sciences and Geophysical Fluid Dynamics, version 2. Results showed that the uncertainty ranges of the tropical total cloud fraction and LSC associated with multiple-parameter perturbation were larger than those from any single-parameter perturbation. The total cloud fraction was significantly improved with multiple-parameter perturbation and the LSC also increased notably when using parameter values optimized for the total cloud fraction because of the indirect parametric effect on lower-tropospheric stability. Non-linear effects between parameters on simulating LSC were much stronger than those on simulating the total cloud fraction. Furthermore, the non-linear interaction reduced large values of total cloud fraction but had a strong incremental effect on large values of LSC over the southeast Pacific. The findings demonstrate the feasibility of improving the simulation of LSC over the eastern oceans by tuning moist parameters, and provide further insight into the non-linear effects between parameters on the simulation of cloud fraction.