Multisource Big geo data for Earth Observation: challenges and open issues

Geo data are heterogeneous by nature comprising both georeferenced images acquired by remote sensing and their derived products (think at down-stream services of remote sensing images such as those provided by the major interna-tional organizations for Earth Observation, namely the ESA Earth Online cata-logues for Sentinel data products, and the NASA catalogue for MODIS data products) and cartographic maps published as open data by public and private organizations. Furthermore, thanks to the Web 2.0 revolution and wide spread diffusion of IoT and smart devices equipped with GNSS sensors, the availability of new and real-time sources of geo data is rapidly increasing. Let us think at Volunteered Geographic Information (VGI) created by citizens eager to participate in citizen science initiatives, such as Open Street Map project, GNSS tracks, and passive crowdsourced geotagged posts created within social networks, or a variety of low-cost sensors data. The huge amount of this heterogeneous multisource geo data constitutes a chal-lenge for the future data economy, although to convert data into value, we need to face some open issues related with the discovery of the relevant geo data among huge repositories, the control and assessment of geo data questionable quality, and finally the cross-analysis of geo data from multiple sources. All such tasks involve the management of the imprecision and uncertainty of both geo data and the user needs. The talk will analyze some applications of multi source geo data management focusing on the roles of both standard Web services as technological basis for enabling multi sources geo data sharing, and fuzzy approaches to model the representation, discovery, quality assessment and geo temporal cross analysis of geo data tolerating imprecision and uncertainty.