Development of analytical approaches for monitoring pipeline corridors for leak detection using open satellite Earth Observation data

点击翻译按钮获取中文摘要

Pipelines move 95% of Canada's natural gas and crude oil, totaling a network of some 540,000 kilometres ofpipeline. Over the 2010 to 2014 period, an average of 147 pipeline incidents were reported to theTransportation Safety Board of Canada, with a majority of these incidents involving the release of products inthe environment. Improved detection of pipeline leaks is of critical importance as oil/waste-water leaks canlead to environmental impacts on rivers, lakes, wetlands, vegetated areas, and cause risk to human health andsafety, as well as having significant economic impacts. Given that pipeline infrastructures are frequentlylocated in remote areas and are difficult to monitor regularly, new Earth Observation (EO) satellite sensors,providing frequent and open data, has great potential to complement existing techniques. Then, the mainobjective of this project is to develop analytical methods for a scientifically rigorous and semi-automatedpipeline leaks surveillance system in snow-off conditions based on time series of open, cost-free Sentinel-2 andLandsat-7 and Landsat-8 imagery. The selection of satellites images will be based on past pipeline incidents.The recent 795,000-litre oil spill from the Keystone pipeline in South Dakota could be a study case. Theimages will be pre-processed using Hatfield Automatic Processing System. Different enhancement techniqueswill be explored to detect pipeline leaks. We will test under which conditions oil or waste water can be seen inwater courses, on soil surface or by stressed vegetation. INRS-ETE will provide research expertise in remotesensing while Hatfield Consultants will contribute its experience in automated images processing andenvironmental monitoring in the oil and gas sector.