Super Resolution Reconstruction – Artificial intelligence improves satellite data to address ecological challenges
Project led by VRVis, with partners AMA, EOX and Uni Salzburg uses AI to refine freely available satellite imagery of agriculturally used land.
Remote sensing is one of the key technologies for modern digitized agriculture. Remote sensing is the process of obtaining information about the state of our earth (land, water, atmosphere) using sensors that do not have to be on site. Particularly in the field of agriculture, this approach is suitable for observing local phenomena such as vegetation, soil or biodiversity via satellites from space, thus contributing to sustainable management strategies and, in a global context, to food security.
The European Copernicus Earth Observation Program is of particular importance here. The freely available Sentinel-2 satellite data are multispectral and have a high temporal resolution. Only the spatial resolution is not sufficient to reliably identify small-scale agricultural land - in Austria, one third of all agricultural land is too small or too complex in shape to be analyzed in detail with these satellite images. Led by VRVis Zentrum für Virtual Reality und Visualisierung Forschungs-GmbH, the FFG-funded SMAIL project, with the help of partners Agrarmarkt Austria (AMA), EOX IT Services GmbH and the Department of Geoinformatics - Z_GIS at Paris-Lodron University Salzburg, addresses precisely this point and uses Super-Resolution Reconstruction (SRR) to improve the spatial resolution of time-series imagery using artificial intelligence.
AI in the application of digital, sustainable agriculture
The area-wide, digital monitoring of agricultural areas (Area Monitoring System) is to support area controls within the framework of the EU's Common Agricultural Policy (CAP) from 2023. Therefore, in the project "Super-resolution-based Monitoring through AI for small Land parcels (SMAIL)", AMA is focusing on new, fast solutions for data-driven agricultural management to achieve higher spatial resolution of the freely available Sentinel-2 satellite data while maintaining spectral quality with the help of AI-assisted SRR.
"Digitization offers many opportunities in agriculture," says Gerd Hesina, Managing Director of VRVis, "and Austria's top-level research is already creating the best solutions in this field, which focus on the sustainable development of the agricultural sector. We are pleased that AMA is focusing on AI-based optimization of freely available satellite imagery." Integrating current satellite data into a geo-information system also opens up a great many possibilities for gaining further information - such as mapping tree populations, forest monitoring or land use changes. In the future, there will be many more fields of application for AI-based SRR, Hesina is convinced.
The SMAIL project is funded under the FFG's AI for Green, a funding program that promotes the development of innovative AI technologies to address environmental challenges in Austria.