Hybrid biotope database mapping

SNIC 2017/5-35


SNAC Small

Principal Investigator:

Helle Skånes


Stockholms universitet

Start Date:


End Date:


Primary Classification:

10503: Multidisciplinär geovetenskap




Today, Stockholm along with major cities in general, are in constant expansion and their municipalities face the challenge to do it in a sustainable way to maintain biodiversity and healthy ecosystems. This development makes the assessment of biodiversity and ecosystem services in urban green structure increasingly important This om turn calls for updated, efficient and advanced mapping support emphasizing ecological assemblage of the landscape, where planners can pinpoint the urban and rural values to preserve or strengthen. There is also an increasing demand from the landscape ecology research community for high detail spatial data on biotopes and their properties in terms of land use, moisture regime, morphology, etc. for spatial analyses. To meet this demand, Stockholm University is building a biotope database method targeting Stockholm County. While this applied project is in close partnership with end users, it also relies on refinement of advanced remote sensing and mapping techniques. Indeed, a trade-off has to be found between the needs of detailed information and efficiency. How can we detect consistent qualities of the landscape at a detailed scale without losing the efficiency in terms of time, money, computer power? We propose a hybrid method based on the interaction between human perception and automatic remote sensing methods using satellite imagery, LiDAR, and CIR high resolution air photo photogrammetry and visual interpretation in photogrammetric stereo-environment. Our goal is to bridge the objectives of nature conservation and the capacity of modern remote sensing techniques in a clever semi-automated way. The project is currently financed through innovation funding from Naturvårdsverket and Boverket. Our method sets very high demands on computer capacity to handle all the detailed data needed for automatic and semiautomatic classification. And to be cost efficient and feasible, we need to be able to run analyses over large areas at the same time. This is impossible to achieve on simple desktop computers.