Detection of hazardous gas leakage from pipelines using anomalous spectral reflectance features of vegetation

Detection of leakage of underground gas pipelines is an important environmental issue in many countries. This paper investigates a non-destructive method of detecting gas leakage from pipelines by investigating spectral response of leaves and canopy in areas affected by gas leakage. In an experiment in a greenhouse we test the effect of small dozes (hence no oxygen shortage) of natural gas (ethane, methane) on leaf spectra. It is concluded that ethane affects vegetation stronger than natural gas and methane. The location of the red edge and red absorption pit are not very strongly affected by any of the gases. In an outdoor experiment we add high dozes of gas to canopy. The experiment showed a clear effect of gas on vegetation growth in terms of reduced height in the gas plots. We also saw a difference between applying ethane (severely affects plant growth) or natural gas (slightly affects plant growth). In the field experiment a clear difference in red edge position on the canopy was detected for gas versus no gas plots. This was supported by a difference in the leave chlorophyll content as well. In a third experiment we measured a 1 km transect over a leaking benzene condensate pipeline in Northern Netherlands. At four different test locations several survey designs were tested to reach the objective of locating variations in health of vegetation. All survey designs proved to be useful, and several anomalies were found in three of the four different locations. Without ground validation it can only be determined if vegetation stress is present but not the exact cause of any anomaly. In the near future we will apply our models to HyMAP imagery and not only look at spectral indicators but also apply spatial-contextual image analysis approaches. In general it is concluded that reflectance spectroscopy (or hyperspectral remote sensing) proved to be a tool that offers a non-destructive investigative method to identify anomalous spectral features in vegetation.