Density is a fundamental property of polar snowpacks and of particular importance for various applications such as remote sensing, surface-mass-balance estimates and paleoclimatic ice-core studies. As a result, there is growing interest in quantifying and understanding the stratigraphy of the snowpack and its spatial variability. Melting and refreezing significantly impact the density layering. Melt frequencies observed in polar ice cores are commonly used as markers for extreme summer temperatures. Moreover, the evolution of a seasonal cycle in density with depth is not well-understood as all of these aspects require extensive field data. To provide a larger sample size, we analyzed snow cores up to 5 m depth from two sampling sites on the Greenlandic and East Antarctic plateaus and determined the spatial distribution of melt features along two orthogonal trench walls of 60 m length and 4 m depth for the Greenland site. We quantify how the observed variability of melt features is strongly influencing the ability to interpret single-core melt records. In particular, prominent vertical melt pipes indicate deep penetration of water in cold Greenlandic plateau firn. For the spatial variability in density, we find a much larger homogeneity of the snowpack along the main wind direction and an imprint of surface features orthogonal to it. Thus, within a certain area (such as the footprint of an altimeter) the stratigraphic variability is directionally dependent and estimation of a representative profile (e.g. from snow cores or modeling) is not straightforward.