1. The Netherlands: Groningen
The first use case is the Netherlands, especially the northern part of the country with a Groningen gas field and the Wadden Sea. This region is active due to both human activities like oil/gas extraction and geo-processes like coastal erosion. The Groningen region due to human activity is affected by land subsidence and earthquakes. For this case, the co/cross polarimetric information from SAR datasets and applying supervised machine/deep learning (ML/DL) methods will be tested for land use and land cover (LULC) classification. The analysis will be supported by topographic basemap data.
2. Poland: Upper Silesian Coal Basin
The Upper Silesian Coal Basin (USCB) is a mining and sinkhole-prone area in southern Poland and one of the most human-altered regions of the European Union. Significant land subsidence has been observed in this area as a result of decades of mining, primarily of hard coal but also non-ferrous mineral ores. Poland's energy transition of closing underground mines has resulted in flooding, and thus into groundwater rebound in the aquifer system, as well as into land uplift and several to a dozen sinkholes recorded each year. The area is also seismically active due to anthropogenic transformations, with at least a few high-energy tremors of mining origin, registered each year, causing land surface displacement. For this case, we attempt to define new SAR signatures for SARbd to serve human-induced change detection and early warning with ML/DL.
3. Indian Coast: Chennai Oil Spill
The Ennore port, near Chennai, is located in the southeastern part of India. In January 2017, due to the collision of two tankers, the water and beaches in this area were affected by a large oil spill. In this use case, the polarimetric SAR features and ML/DL algorithms will support the detection of the oil spill and the assessment of its trajectory, including the prediction of the harmful extent along the beach.
Sample datasets
The sample datasets for those three cases have been uploaded to EOTDL database, and are freely accessible. They are at Q0 and Q1 quality levels.