Vernacular buildings, rooted in local traditions, are up-standing remains related to past activities of common people. The Slovenian karst region is characterized by such constructions of different types and functions (e.g. residence, shelter, storage place), built in the long-held building technique of dry stone. Today, many of these buildings are often difficult to access because they are located in the forest. To this end, a multi-scalar approach that joins airborne laser scanning (ALS) with meticulous ground-based three-dimensional (3D) acquisition techniques was set up to both inventory and (better) understand these specific archaeological constructions, with an emphasis on the shelters built by shepherds (in this region known as hiška or kjuta). First this paper will provide a comparison of various terrestrial techniques (such as image-based modeling and terrestrial laser scanning (TLS)) that were applied to document the 3D geometry of these vernacular constructions. Besides the specific characteristics (such as accuracy) of each dataset, the various documentation methods will also be analyzed with respect to their practical application in dense forests. Afterwards, the first results of this detailed ground-based documentation will be provided. More specifically, this terrestrial phase should allow to draw conclusions on the average dimensions of these buildings and the amount of building material and effort needed for their construction. Third, ALS datasets will be used to position and contextualize the terrestrial data. Both methods (airborne and ground-based) contribute to the inventory, interpretation and understanding of these structures in their own specific but complementary manner. ALS works as a broad brush approach and therefore enables one to contextualize the structures by fitting them into the landscape, while terrestrial datasets facilitate a detailed examination of the individual structures. Using the detailed 3D datasets provided by terrestrial acquisition techniques, it is also hoped to develop a specific ALS filtering and object recognition strategy that allows to semi-automatically detect such buildings that remained hitherto unknown.