Geosynchronous infrared satellite data is commonly used to track and interpolate between microwave precipitation signals, and can also be useful for estimating warm precipitation over land where microwave techniques fail. The quality of such tracking and estimations depends on the correlation between infrared cloud features and precipitation features. In the current work the distribution of precipitation rates is examined as a function of cloud tower area and cloud top temperature. An IR satellite thunderstorm tracking algorithm developed at Meteo-France is used to track cumulus convective towers that are matched with radar data at 5 minute 1 km resolution. It is found that most (80%) of the precipitation occurs in the larger cloud mass that surrounds the towers, and when a tower is first detected the precipitation is already in progress 50% of the time. The average density of precipitation per area is greater as the towers become smaller and colder, yet the averaged shape of the precipitation intensity distribution is remarkably constant in all convective situations. This suggests that on average all convective precipitation events look the same regardless of location inside the cloud, despite the higher frequency of occurrence per area inside the convective towers.