Abstract
A fully automated, objective classification system has been developed to analyze infrared satellite imagery. This automated system facilitates tracking and categorization of convective weather systems into various classes. The classes chosen reflect the maximum degree of organization attained by each weather system. Four classes of convective weather system are defined; tropical cyclones (TS; including prestorm clusters through to decaying storms), mesoscale convective complexes (MCC), convective cloud clusters (CCC), and disorganized shortlived convection (DSL). Systems are identified, tracked, and then classified. If a system satisfies the criteria for any of the organized convection classes (TS, MCC, or CCC) for at least two time periods, the entire track is allocated to that class. In cases where a system satisfies the criteria for more than one type of organized convection (commonly MCC and CCC), it is assigned to the “most organized” class (in this case, MCC). Thus, the characteristics of each class incorporate the life cycles of systems that satisfy the imposed criteria for at least a 6-h period.
Two satellite infrared-based (IR) rain-rate algorithms are applied to the convective areas in order to obtain precipitation amounts for the various classes of convection. The domain of interest extends from the eastern Pacific margin to the African coast (15°W) and 40°N–40°S.
In addition to the IR data, rain rates derived from Special Sensor Microwave/Imager data are compared with the infrared retrieved rain rates at available times for a subset of each of the three organized convection classes. Rainfall amounts obtained from these infrared algorithms are also compared with ground-based station observations over Florida. Comparison of the inferred rainfall with station data reveals that the TS precipitation is in approximate agreement (in the mean), whereas the precipitation contributions from the other forms of convection are somewhat overestimated. DSL is overestimated the most and CCCs are overestimated the least.
According to the infrared-based rain-rate algorithms, DSLs (short-lived systems) contribute the most total (basinwide, annual) precipitation, CCCs contribute the second largest amount, MCCs are third in the contribution of precipitation, and TSs contribute the least to the total precipitation.
