Semi-automatic open source geoprocessing for change-detection in federal geodata

This project addresses the vast complexity of keeping federal geodata up to date according to the official regulations. Cooperating with the Hessian state agency for land management and geoinformation (Hessisches Landesamt fÜr Bodenmanagement und Geoinformation - HLBG) concepts and solutions have been developed in a two-year project. It is supposed to enhance the efficiency of the geodata update processes within state agencies.
With the help of the concept of change detection datasets are supposed to be updated. The aim of the project is to provide a thematic layer of change indicators for the staff dealing with updating datasets. To assess the feasibility of this approach in a federal agency surrounding two exemplary datasets were defined. The long-term goal is to use semi-automatic change detection processes for increasing numbers of datasets within the agency.
Lake boundaries as well as relevant (meaning: above a Delta z-threshold defined by the agency) elevation changes were processed. Light detection and ranging DEM (Lidar) is supposed to be kept up to date without subsequent flights. This paper describes the process of updating the lake boundaries.
According to the needs of the state agency processing rule sets (e.g. accuracy tolerances or minimum area size) were developed. The rule sets were then implemented using the Python programming language to create geoprocessing scripts as post-processing algorithms for raster database queries, which can tackle the data amount for Hesse with its 21,000 km².
This paper presents the methods which are being used to detect objects and derive features to be used in the process of change detection. Due to the high topicality of the oriented aerial images their derivatives (photogrammetric point clouds, orthophotos) serve as the core element for the detection of change features in the more recent epoch