Landslides are recurrent natural hazards in mountainous regions of the Earth, claiming people’s lives and destroying man-made infrastructures. Society and policy makers are interested in knowing the location and spatial extent of landslides immediately after a triggering event, as well as in predicting landslides. Detailed, accurate and complete landslide maps are essential to meet these interests. Promoted by the increased availability and quality of Earth Observation (EO) data, attempts to automate the production of landslide maps have recently been undertaken. As demonstrated by several case studies, especially object-based image analysis (OBIA) provides methodologies that are valuable for automated landslide mapping. However, existing OBIA mapping routines have been targeted to specific data sets and study areas, and can therefore be hardly transferred to other data sets or geographical settings. In order to achieve a high degree of automation, objectivity, and general applicability of an object-based landslide mapping system, the proposed project aims at the following developments:

• Automated knowledge-based and statistical OBIA routines that are optimised for application to VHR (very high resolution) and HR (high resolution) optical EO data sets for the polygon-based extraction of landslides after a triggering event;
• OBIA mapping routines that are automated and at the same time robust and flexible tochanging input optical EO data sets (and thus spatial resolution), as well geographical setting;
• Automated knowledge-based and statistical OBIA routines that are optimised for application to multi-temporal VHR and HR optical EO data sets for the polygon-based detection of landslide changes;
• An innovative and efficient web processing chain which comprises robust OBIA mapping routines for feeding a pre-operational service that allows for automated polygon-based landslide mapping based on VHR and HR optical EO data, as well as the identification of landslide-affected infrastructure.

For accomplishing the above-mentioned developments four to six study areas with different environmental characteristics will be defined in Austria and South Tyrol, Italy. Various HR and VHR optical EO data sets will be acquired for these areas, including also very recent satellite data such as Sentinel-2. A comprehensive analysis of different optical data for distinct geographical areas is highly innovative and will lead to the development of a robust and reliable landslide mapping system using OBIA. This system is essential for the implementation of a pre-operational service. It is expected that this service will be close to the needs of potential users, and will provide users with extensive and reliable information on landslides. All products of the proposed projects, i.e. maps, routines, and the service, will be thoroughly evaluated: the spatial and thematic accuracy of landslide maps will be quantified; routines will be evaluated by innovative robustness indices; the usability and relevance of the service will be validated by potential users. A comprehensive validation report will document major findings.