In this study analysis of more than 50 high spatial resolution Landsat images over Central America has been made. Cloud cover, spatial pattern, and cloud size are correlated with local land use and elevation changes. The influence of land surface on spatial organization of cumulus cloud fields is examined using the Colorado State University Regional Atmospheric Modeling System. The spatial organization of simulated cumulus cloud fields over homogeneous surfaces with varying heat and moisture fluxes is examined. The simulations show strong tendencies towards regular spatial arrangement of cumulus clouds during the early stages. The spatial organization of cumulus clouds over surfaces with varying degrees of heterogeneity is also examined. Previous modeling studies have shown that checker board patterns of surface heterogeneity can result in hotspots for convection. The effect of random spatial distributions of surface heterogeneities on spatial distribution of cumulus are simulated. Preferential formation of clouds over areas of contrasting surface properties could account for the observations of cumulus clouds clustering during the early stages of cloud formation.