Annual Cycle of Tropical Convective Clouds from 20-Year MODIS Observations with an Improved Convective Cloud Detection

Abstract This study utilizes a 20-yr (2003–22) dataset of tropical deep convective clouds (DCCs) from MODIS wide-swath observations with a new detection method to characterize the annual cycle of tropical DCCs, focusing on global patterns, regional variations, and day/night differences. Tropical mean daytime DCC occurrence peaks in December (1.90%) and reaches a minimum in May (1.55%). Notable features include the double ITCZ over the eastern Pacific in March–April and alternating northwest–southeast bands of day/night differences over the Amazon. Regional analyses across six tropical regions, using four DCC size classes (class 1 “small” to class 4 “giant”), reveal distinct behaviors: The western Pacific has the highest DCC occurrences, driven primarily by giant clusters; the eastern Pacific shows similar seasonal trends across all classes; the equatorial Atlantic lacks growth of class 4 clusters; Africa consistently shows larger nighttime cluster sizes; and the Amazon features an alternating day/night dominance in DCC occurrences, with larger nighttime occurrences from May to August but smaller values from October to December. Distributions of normalized DCC cluster number density and occurrence as a function of cluster size show that larger clusters (>10 4 km 2 ) dominate total convective area, despite the numerical dominance of smaller clusters (<10 3 km 2 ). Oceanic regions exhibit unique patterns of day/night differences, characterized by fewer clusters smaller than 10 2.5 km 2 and larger than 10 5 km 2 , but more clusters in the intermediate range during nighttime. These findings provide valuable information for model evaluations, highlighting the importance of resolving DCC cluster size variability in climate models.