ForschungForschungsbereichePhotogrammetrische Bildanalyse
Usage of remote sensing data for national surveying tasks

Usage of remote sensing data for national surveying tasks

Team:  M. Voelsen, F. Rottensteiner, C. Heipke
Jahr:  2019
Förderung:  Landesamt für Geoinformation und Landesvermessung Niedersachsen (LGLN)
Laufzeit:  2019 - 2022

The main objective of this project is to utilize Sentinel image data for the automatic updating and enrichment of geospatial databases of the surveying administration of lower Saxony (LGLN), especially with regard to land cover and land use. Currently, this is still done using images from aerial flights, which are acquired every three years in an expensive manner (see link). Sentinel data by contrast are acquired every 5-6 days for each area in Lower Saxony, but those images have a limited geometric resolution. Sentinel data are continuously collected by the European Space Agency (ESA) within the framework of the Copernicus programme for several years now and they are made available free of charge. In this project radar data from Sentinel-1 and optical data from Sentinel-2 will be merged and used jointly: optical data for a reliable acquisition of geoinformation, while radar data are weather-independent and therefore allow a high temporal resolution.

The data is analyzed using deep learning techniques. In concrete terms, convolutional neural networks are used in order to assign a class to each pixel using semantic segmentation. Such a class can be, for example, the type of land cover. The training data needed for this method are obtained from existing geodata of the LGLN provided that their actuality is linked to temporally suitable sentinel data.