A process oriented characterization of tropical oceanic clouds for climate model evaluation, based on a statistical analysis of daytime A-train observations

Abstract

This paper aims at characterizing how different key cloud properties (cloud fraction, cloud vertical distribution, cloud reflectance, a surrogate of the cloud optical depth) vary as a function of the others over the tropical oceans. The correlations between the different cloud properties are built from 2 years of collocated A-train observations (CALIPSO-GOCCP and MODIS) at a scale close to cloud processes; it results in a characterization of the physical processes in tropical clouds, that can be used to better understand cloud behaviors, and constitute a powerful tool to develop and evaluate cloud parameterizations in climate models. First, we examine a case study of shallow cumulus cloud observed simultaneously by the two sensors (CALIPSO, MODIS), and develop a methodology that allows to build global scale statistics by keeping the separation between clear and cloudy areas at the pixel level (250, 330 m). Then we build statistical instantaneous relationships between the cloud cover, the cloud vertical distribution and the cloud reflectance. The vertical cloud distribution indicates that the optically thin clouds (optical thickness 3.4) are composed of high and mid-level clouds associated with deep convection along the ITCZ and SPCZ and over the warm pool, and by stratocumulus low level clouds located along the East coast of tropical oceans. The cloud properties are analyzed as a function of the large scale circulation regime. Optically thick high clouds are dominant in convective regions (CF > 80 %), while low level clouds with low optical thickness (< 3.5) are present in regimes of subsidence but in convective regimes as well, associated principally to low cloud fractions (CF < 50 %). A focus on low-level clouds allows us to quantify how the cloud optical depth increases with cloud top altitude and with cloud fraction.