Implementation of software components to build web-based decision support systems: the case of the land take pipeline within a geospatial cyberinfrastructure
Co-organized as ESSI1.20/HS12.12/SSS13.41
Convener: Giuliano Langella | Co-conveners: Peter Baumann, Francesco Vuolo
Mon, 08 Apr, 14:00–15:45
Room -2.85

Nowadays, researchers have to tailor their models, data and results into systems which can be used by non-experts, such as policy makers, stakeholders, farmers and the many professionals in need of clear answers to land management questions.

One way ahead to bridge the gap between R&D and real-life applications is the development of decision support systems (DSS) on top of geospatial cyberinfrastructures (GCI) that can handle end-user requests in real time with all the complexity being transparent to the user.

The short course will cover some developments carried out within the EU H2020 LandSupport Project. The implementation of an indicator of land-take is showed, both presenting how to deal with the technical steps on a more general level and proposing hands-on sessions on the implementation of specific components of the whole land-take workflow.

First an introduction is presented, covering a general overview about the GCI and the requirements of pipelines.
A brief description of the main tasks follows:

• Big spatio-temporal raster data are managed by means of rasdaman. Here a workflow is presented showcasing how to import and query multi-band Sentinel-2 data based on the OGC Big Data Standards.
• Cloud masking and filtering. Copernicus Sentinel-2 data are processed to obtain bottom of the atmosphere, cloud free, reflectance data. A theoretical and a hands-on session in R will be presented.
• Classification. A spectral-temporal datacube of Sentinel-2 data are used to get a binary map of imperviousness (1: urban pixel, 0: non-urban pixel). At least one classification model will be presented with hands-on in R and/or MatLab.
• Land-take. An algorithm to calculate land-take using a low-level programming language is showed, with more advanced insights about the opportunity to face GPU calculations.

Altogether, we motivate how the LandSupport approach aims at providing decision support based on multi-source spatiotemporal data in a user-centric manner.
Ample time will be available for answering questions and discussion.