In this paper, an extensive dataset of snow in situ measurements, collected in support of airborne SAR-acquisitions in Sodankylä and Saariselkä test sites in northern Finland, is used to analyse the heterogeneity of bulk snow properties (snow depth, density and water equivalent) over different land cover types in northern taiga and tundra areas. In addition, the applicability of different spatial frequencies of snow sampling to estimate the true snow conditions is investigated. Overall, the highest variability in bulk snow properties was found over sparsely vegetated land cover groups, but the scale of variation was smaller in forested areas, as these areas exhibited a low correlation length in snow depth. This implies that more frequent measurements should be executed in forested (~ every < 5 m) than in open areas (~ every 7.5–12.5 m) to catch the true variability in snow depth. The results also indicated that the current spatial resolutions of space borne microwave radiometers and radars used for the remote retrieval of bulk snow properties are all well above the limit to fully describe the spatial variation of e.g. snow depth even in open areas. This conclusion supports the demand of research investigating high-resolution parameter retrieval in remote sensing of snow, e.g. using advanced SAR techniques.