During the last years, remote sensing data along with GIS have been largely employed for supporting emergency
management activities. In this context, the use of satellite images and derived map products has become more common
also in the different phases of humanitarian crisis response.
In this work very high resolution satellite imagery was processed to assess the evolution of Za’atari Refugee Camp, built
in Jordan in 2012 by the UN Refugee Agency to host Syrian refugees. Multispectral satellite scenes of the Za’atari area
were processed by means of object-based classifications. The main aim of the present work is the development of a semiautomated
procedure for multi-temporal camp monitoring with particular reference to the dwellings detection.
Whilst in the emergency mapping domain automation of feature extraction is widely investigated, in the field of
humanitarian missions the information is often extracted by means of photointerpretation of the satellite data. This
approach requires time for the interpretation; moreover, it is not reliable enough in complex situations, where features of
interest are often small, heterogeneous and inconsistent. Therefore, the present paper discusses a methodology to obtain
information for assisting humanitarian crisis management, using a semi-automatic classification approach applied to