The interpretation of paleo-climate information gained from polar ice cores presumes a precise knowledge about the processes of signal formation, in particular that of air and greenhouse gas related signals. In contrast to all other climatic signals air is entrapped at the firn-ice transition 50–100m below the surface. Therefore, the air enclosure desires a broad understanding of all possible processes affecting firn densification over the 100–3000 years lasting compaction from snow to ice. Data obtained from one firn or ice core provide informations about one location of the ice sheet. However, these informations are not necessarily representative due to local processes, like for example accumulation, which are not constant over time and space. Thus, to understand the spatial and temporal variability in polar firn and its development, spatially extended high resolution investiga- tions of firn cores are required. Spatially distributed firn cores allow to bridge this gap between point measurements and area measurements. This thesis aims to contribute to the understanding of the development of spatial and temporal variability in polar firn. For this purpose, five firn cores distributed over only a few kilometres were analyzed. Additionally, the spatial and temporal variability of surface snow at the same site in Dronning Maud Land, East Antarctica, was retrieved. Important parameters for the study of the snow surface were the specific surface area (SSA) of the snow, its isotopic composition and the meteorological conditions. To investigate the densifica- tion process, high resolution density profiles of the firn cores were obtained, as well as their impurity content. For the first time the surface snow SSA development during two months in austral summer was observed. The expected decrease of SSA was altered by precipitation events. Only small amounts of precipitation were sufficient to rapidly increase the SSA, which has a high impact on the development of the albedo. In contrast to previous assumptions, the frequency of precipitation events is potentially more important than their amount. From the distinct spatial pattern of the isotopic composition of the snow, it was possible to char- acterize the snow surface of this region. In the deep firn spatially extending density layers are formed. The key find- ing is, that the density layering over a distance of a few kilometres develops remarkably coherent. The spatial variability between the firn cores decreases over depth. Potentially impermeable layers, that are spatially extending and prohibit the air movement through the firn column, were identified even above the transition from firn to ice. The present work forms the basis for modeling studies that quantify the influ- ence of impermeable density layers on the interpretation of the paleo-climate archive.