The project

For the very first time, we plan to use a combination of satellite images and artificial intelligence to show how many areas of sealed soil exist across three cities in Germany: Hamburg, Leipzig and Stuttgart. Alongside the data analysis and visualization, we plan to investigate whether these areas are predominantly in private or public hands. The project is implemented as a cooperation between the non-profit media house CORRECTIV and the data agency Vertical52. We also plan to work together with local journalists. Focus of our publication will be an interactive map.

Why is this relevant?

Sealing means that water for example due to heavy rains cannot drain into the ground. The more sealed areas there are, the more the risk of floods and drought increases. And sealed areas ensure that it gets hotter. Unsealing is one of the most important measures to protect people from the consequences of the climate crisis. And it is the task of the municipality. Since elections take place in June and September, we are planning several publications.

Methodology

For the detection of sealed areas, we use several detection procedures in combination with existing data from the municipalities (property cards, aerial photographs). The required data is obtained from geoportals and other available sources and normalized for further use. In addition to the existing information (buildings), we generate new information based on training data and models mechanical learning. For this purpose, buildings and roads are recognized, for example, and compared with existing data in order to identify unknown sealed areas.