Automated feature extraction by combining polarimetric SAR and object-based image analysis for monitoring of natural resource exploitation

Kurzfassung

An automated feature extraction procedure based on the combination of a pixel-based unsupervised classification of polarimetric synthetic aperture radar data (PolSAR) and an object-based post-classification is presented. High resolution SpotLight dual-polarimetric (HH/VV) TerraSAR-X imagery acquired over the Doba basin, Chad, is used for method development and validation. In an iterative training procedure the best suited polarimetric speckle filter, processing parameters for the following entropy/anisotropy/alpha (H/A/α) decomposition and the heron based unsupervised Wishart classification are determined. By considering feature properties such as shape and area, the subsequent object-based post-classification increases the user’s and producer’s accuracy of the feature extraction procedure by an order of a magnitude to finally 59% to 71% in each case (valid for an area based accuracy assessment), or to even 74% to 89%, taking only the numbers of correctly/falsely detected features into account. In addition, the high transferability of the methodology is verified by an application to a second TerraSAR-X acquisition. The feature extraction procedure is developed for monitoring oil field infrastructure. For developing countries, several studies reported a high correlation between the dependence of oil exports and violent conflicts. Moreover, land use and environmental issues also occur in countries characterized by a peaceful development of their oil industry. Consequently, to support problem solving, an independent monitoring of the oil field infrastructure by Earth observation is proposed, enabling monitoring of large areas within a short time to compare the real amount of land used by the oil exploitation and the companies’ contractual obligations. The developed method focuses on the monitoring of the oil well pads, characterized by rectangular, approximately 50 m x 100 m large patches of bare land.