Mapping land-based oil spills using high spatial resolution unmanned aerial vehicle imagery and electromagnetic induction survey data

摘要: Natural oil and gas are important sources of energy worldwide their exploration exploitation have significantly increased due to the global demand. The transportation these valuable resources greatly depends on pipelines; however, pipeline leakages huge economic environmental impacts warranting an effective operational methodology for monitoring. We proposed a method mapping soil contamination leakage in Dixonville, Alberta, Canada. In particular, very high-resolution unmanned aerial vehicle (UAV) imagery electromagnetic induction (EM) surveying data were analyzed using hierarchical object-based random forest (RF) algorithm. level-1 classification, land cover map was produced UAV data. Next, all classes, excluding contaminated soil, masked out. level-2 class further partitioned into three subclasses representing varying degrees contamination. Specifically, we salinity index, named normalized detect areas index herein, as well several other indices bands, used input features classification. An overall classification accuracy about 77% achieved method. results demonstrate that synergistic use high spatial resolution EM is promising detecting examining ecosystem disturbance leakage.