The concept of "integrity", as currently used in the analysis of World Heritage sites or cultural landscapes mainly consists of 1) the composition of a given site, regarding its origins and its current state of conservation; 2) the visual and functional relationships between its components (attributes). One of the major questions is "what defines the origin period?". The integrity analysis has to clearly understand and evaluate which tangible components exist, in order to correctly identify their origin period, estimate how much of the original structure and function remain. An additional difficulty rises in case of scientific and/or technical heritage assessment: the very historical necessity of updating and implementing technical/scientific innovations. These follow advances in science and/or technique(s), and lead to frequent successive changes, impacting the site's structure. Hence, for living sites, one cannot see the origin of the project as the only reference state, as it could be for classical heritage; it is instead requisite to enlighten a chronological series of major reference states. The Pic-du-Midi Observatory (France) is a remarkable example as high-mountain scientific station (2860 m). Its various and successive scientific uses led to a series of reshaping and structural evolutions over around 150 years, till nowadays. We have there a series of reference states very conform to an active scientific station and observatory. The project aims to combine historical data and measurements on the current site's state, to build a succession of 3D models for a series of 5 or 6 'reference state' of the observatory within associated database, and to document the integrity analysis more accurately than usual. The final uses of the digital achievement will also serve for the management of the site and for further development of