| description abstract | Using GOES-7 ISCCP-B3 satellite data for 1987?88, the authors studied the evolution of the morphological and radiative properties of clouds over the life cycles of deep convective systems (CS) over the Americas at both tropical and middle latitudes. A deep convective cloud system is identified by adjacent satellite image pixels with infrared brightness temperatures, TIR < 245 K (?28°C), that at some time contain embedded convective clusters that are defined by pixel values of TIR < 218 K (?55°C). The first part of the analysis computes parameters for each convective system that describe the system areal size, number of convective clusters, fractional convective area, average and maximum size of the convective clusters, shape eccentricity and orientation, mean TIR, variance of TIR, TIR gradient, and the mean, variance, and gradient of collocated visible reflectances (when available). The second part of the analysis searches a 5° ? 5° region centered on each convective system, but in the subsequent image (3-h time separation), to locate possible candidates representing the same system at the later time. For each possible candidate, the method calculates the fraction of areal overlap with the target system and the implied speed and direction of propagation of the whole convective system and the largest convective cluster within the candidate system. The authors review previous studies of the sensitivity of CS statistics to the temperature thresholds used for identification and quantify the effects on these statistics produced by different ways of tracking convective systems. Comparisons of the results from several tracking methods explains how they work and why most of the life cycle statistics are not sensitive to tracking method used. The authors confirm that simple coincidence (as used in most previous studies) works as long as the time step between satellite images is smaller than the time required for significant evolution of the CS: since smaller systems evolve more rapidly, getting accurate results for CS smaller than about 50?100 km probably requires time resolution better than 3 h. The whole dataset has been analyzed by a tropical meteorologist who choses the best candidate at each time step by comparing listings of all the calculated parameters and visually examining each satellite image pair. The whole dataset has also been analyzed by a simple automated procedure. Based on about 3200 cases examined by the meteorologist and about 4700 cases obtained by the completely automated method, the statistical behavior of convective systems over the Americas is described: their mean cloud properties as a function of system size and lifetime, the evolution of their cloud properties over their lifetimes, and their motions. The results demonstrate a direct correspondence between the size and lifetime of mesoscale convective systems and exhibit many common features of their growth, maturation, and decay. Some differences between CS over tropical land and tropical ocean are also apparent. | |
