Nitrogen atoms are adsorbed on a Cu(001) surface with c(2x2) periodicity. As shown in Fig.1, a grid-like nanopattern spontaneously develops at N coverages around 0.3ML. The self-organized pattern was first reported by Leibsle et al. in 1993 [1], but it has not been answered why and how the pattern is realized. We noticed two types of boundaries between N-islands, “monoatomic lines” and “multiatomic belts”, and their relations to N-atom alignments (“in-phase” or “out-of- phase”) between neighboring islands. We proposed structural models for the two boundaries and answered to the long-unanswered question [2]. Keys in our model are the relief of compressive stress induced by N adsorption and the stability of N- atoms coordinated laterally by four Cu atoms Recently we found that a ribbon-like nanopattern develops on a vicinal Cu(001) surface [3]. The ribbon pattern on a vicinal region coexists with a normal grid pattern on a flat (001) region, indicating that the surface symmetry determines which nanopattern develops (mirror symmetry of the vicinal surface due to parallel steps running with short spacing vs. square symmetry of the flat (001) surface). The N-islands in the ribbon pattern are deformed to rectangles. The deformation can be understood simply by taking into account the displacement of the second layer Cu atoms. In this talk, I would like to emphasize that simple hard sphere models are very effective in analyzing complex (or large-unit-cell) surface structures. [1] F.M. Leibsle et al., Phys. Rev. B 47 (1993) 15865-15868. [2] M. Yamada et al., Surf. Sci. 604 (2010) 1961-1971. [3] M. Yamada et al., e-J. Surf. Sci. Nanotech. 14 (2016) 43-47.