A supervised neural network algorithm is used to categorize near-global satellite retrievals into three mesoscale cellular convective (MCC) cloud morphology patterns. At constant cloud amount, morphology patterns differ in brightness associated with the amount of optically-thin cloud features. Environmentally-driven transitions from closed MCC to other morphology patterns, typically accompanied by more optically-thin cloud features, are used as a framework to quantify the morphology contribution to the optical depth component of the shortwave cloud feedback. A marine heat wave is used as an out-of-sample test of closed MCC occurrence predictions. Morphology shifts in optical depth between 65°S - 65°N under projected environmental changes (i.e., from an abrupt quadrupling of CO2) assuming constant cloud cover contributes between 0.04-0.07 W/m2/K (aggregate of 0.06) to the global mean cloud feedback.