Missing CI in multiple cloud layer decks

The GOES-R CI algorithm has trouble detecting Convective Initiation when there are clusters of clouds of multiple levels in the same area. Two areas of initiation occurred today, one near the Red River, and one near the south of the Dallas/Fort Worth area where multiple decks of clouds prevented the CI product from tracking the cells. The CI algorithm requires at least 10% overlap of objects between two successive time periods, and without being at the office to look at the overlap, it is not known if these were missed due to mistracking or if the algorithm just did not detect convective clouds due to the multiple cloud layers. Something we will need to look into for the future. In the scalloped areas from 1900 UTC and 2000 UTC, one can see the difficulty of pulling out distinct features in the visible imagery, yet by 2000 UTC, full blown thunderstorms had developed.