The diurnal cycle of cold cloud cover is underestimated within Earth system models (ESMs) with the greatest underestimation in the afternoon. To better understand the diurnal cycle of tropical oceanic cloud cover, the diurnal cycle of deep convective system (DCS) initiation and the subsequent contributions to cloud cover resulting from systems initiating at earlier times is analyzed using newly developed DCSs Lagrangian tracking methodologies. Satellite infrared-based Tracking Of Organized Convection Algorithm through 3D segmentatioN (TOOCAN) DCSs are matched to Global Precipitation Measurement (GPM) mission precipitation and diabatic heating products. Matched data are then binned by their hour of initiation (in local solar time) to evaluate morphological characteristics and contributions to rain and cloud cover diurnal cycles. Analysis reveals an unexpectedly large peak in daytime DCS initiation that produce subsequent afternoon cloud cover, thus suggesting that the discrepancy between ESMs and observations is likely due, in part, to ESM misrepresentation of initiation or maintenance of daytime-initiated DCSs. Results also show that daytime DCSs produce less precipitation, but relatively more cloud shield compared to DCS that initiate overnight. As a framework to understand these diurnal variations in cold cloud production, we will apply a semi-empirical source-sink cold cloud area growth model that includes a convective area source term and latent heating source term. Vertical latent heating profiles from GPM, DCS morphology from TOOCAN, and atmospheric lapse rates and density from ERA5 are fit to the semi-empirical model to estimate cloud growth and decay timescales. Observation-estimated timescales and the source term variations will be evaluated to understand key drivers in the differences in DCS cold cloud production across the diurnal cycle.