Precipitation characteristics of mesoscale convective systems in the tropical western Pacific inferred from satellite, radar reflectivity and a mesoscale numerical model

Satellite infrared (IR) and visible (VIS) images from the TOGA-COARE experiment are investigated through the use of Clustering Analysis. A new approach to classify high clouds using only IR and the clustering technique is proposed. Radar reflectivity field is compared to each portion of cloud previously classified in different events. Spatial correlation coefficient between radar and satellite classified cloud images is computed to identify the ideal number of classes (N) for the clustering procedure. This procedure indicates the best matching of N=6 classes with four classes associated to radar reflectivity pixels. The correspondence between cloud classification and radar reflectivity for evolving Mesoscale Convective Systems (MCS's) is investigated through Factor Analysis. Numerical simulations with the PSU/NCAR mesoscale model (MM5) for four case studies are performed to support the physical interpretation of each factor. It is shown that the first factor can be identified with stages when the convective region of a MCS is well developed, i.e. radar reflectivity pixels are related to the core of maximum convective portion of cloudiness (cold cloud top temperatures). Meanwhile, the second factor describes cloud characteristics associated with the anvil precipitating region and the formation of new convective elements.