The Ekman dynamics of the ocean surface circulation is known to contain attracting regions such as the great oceanic gyres and the associated garbage patches. Less well-known are the extents of the basins of attractions of these regions and how strongly attracting they are. Understanding the shape and extent of the basins of attraction sheds light on the question of the strength of connectivity of different regions of the ocean, which helps in understanding the flow of buoyant material like plastic litter. Using short flow time trajectory data from a global ocean model, we create a Markov chain model of the surface ocean dynamics. The surface ocean is not a conservative dynamical system as water in the ocean follows three-dimensional pathways, with upwelling and downwelling in certain regions. Using our Markov chain model, we easily compute net surface upwelling and downwelling, and verify that it matches observed patterns of upwelling and downwelling in the real ocean. We analyze the Markov chain to determine multiple attracting regions. Finally, using an eigenvector approach, we (i) identify the five major ocean garbage patches, (ii) partition the ocean into basins of attraction for each of the garbage patches, and (iii) partition the ocean into regions that demonstrate transient dynamics modulo the attracting garbage patches. V C 2014 AIP Publishing LLC.[http://dx.doi.org/10.1063/1.4892530]Ocean dynamics operate and affect climate on timescales of months to millenia. In this paper, we investigate phenomena on the ocean's surface that manifest over very long time periods: we look for regions in which water, biomass, and pollutants become trapped "forever" (which we refer to as attracting regions), or for long periods of time before eventually exiting (which we refer to as almostinvariant regions). While attracting regions may be quite small in size or irregular in shape, they can nonetheless exert great influence on the global ocean surface dynamics if their basins of attraction are large.