LAUSR

ABSTRACT The leakage of natural gas from underground storages can cause economic, safety, and ecological problems. However, when the leakage is slight, the gas concentration in surface atmosphere cannot reach the threshold for direct monitoring. Hyperspectral remote sensing can indirectly detect it through the spectral changes of vegetation in the leak area. A field experiment was performed, which took bean, corn and grass as test plants, to study the characteristics of canopy spectral changes under natural gas leakage stress. The results showed that the spectra of stressed vegetation changed significantly after three days of gas injection, which manifested an increase in the visible bands of 570–700 nm and a decrease in the near-infrared of 760–1000 nm. By performing ANOVA (Analysis of Variance) on the spectra of vegetation in the control and experimental plots, the bands with p< 0.05 and bigger F values were selected as the sensitive bands. And the natural gas stress index was constructed using the selected bands. A quantitative test by JM (Jeffries-Matusita) distance showed that the index could well identify (JM > 1.8) the stressed vegetation throughout the phenological plant cycle. This study can be applied to detect underground natural gas storage leakage spots by hyperspectral remote sensing.