Spatial characteristics of the tropical cloud systems: comparison between model simulation and satellite observations

Abstract

A Lagrangian cloud classification algorithm is applied to the cloud fields in the tropical Pacificsimulated by a high-resolution regional atmospheric model. The purpose of this work is toassess the model’s ability to reproduce the observed spatial characteristics of the tropical cloudsystems. The cloud systems are broadly grouped into three categories: deep clouds, mid-levelclouds and low clouds. The deep clouds are further divided into mesoscale convective systemsand non-mesoscale convective systems. It is shown that the model is able to simulate the totalcloud cover for each category reasonably well. However, when the cloud cover is broken downinto contributions from cloud systems of different sizes, it is shown that the simulated cloudsize distribution is biased toward large cloud systems, with contribution from relatively smallcloud systems significantly under-represented in the model for both deep and mid-level clouds. The number distribution and area contribution to the cloud cover from mesoscale convectivesystems are very well simulated compared to the satellite observations, so are low clouds aswell. The dependence of the cloud physical properties on cloud scale is examined. It is foundthat cloud liquid water path, rainfall, and ocean surface sensible and latent heat fluxes have aclear dependence on cloud types and scale. This is of particular interest to studies of the cloudeffects on surface energy budget and hydrological cycle. The diurnal variation of the cloudpopulation and area is also examined. The model exhibits a varying degree of success in simulatingthe diurnal variation of the cloud number and area. The observed early morning maximumcloud cover in deep convective cloud systems is qualitatively simulated. However, the afternoonsecondary maximum is missing in the model simulation. The diurnal variation of the tropospherictemperature is well reproduced by the model while simulation of the diurnal variationof the moisture field is poor. The implication of this comparison between model simulation andobservations on cloud parameterization is discussed.