A Fragile Watermarking Approach for Earth Observation Data Integrity Protection

Volumes of data generated from remote sensing activities have grown larger in the last years due to the increasing number of satellites orbiting the earth and the development of drones and their sensors. Multiple online providers allow obtaining the data of a geographical location from a given time period. Furthermore, free downloading of data sets that store information in different formats is possible. The easy access, portability, and distribution make it possible to perform data tampering and spread inaccurate information, which leads to wrong data-based conclusions. This work proposes a fragile watermarking approach to detect Earth observation data contradictions and avoid economic losses that could compromise organizations’ performance. We design an architecture to link attributes and generate a signal that is stored in other fragments of the same data, allowing double-checking of data quality before proceeding with decision-making. Our method is proven to be very effective in spotting unauthorized updates.