With the global warming, people now pay more attention to the problem of the emission of greenhouse gas (CO2). Carbon capture and storage (CCS) technology is an effective measures to reduce CO2 emission. But there is a possible risk that the CO2 might leak from underground. However, there need to research and develop a technique to quickly monitor CO2 leaking spots above sequestration fields. The field experiment was performed in the Sutton Bonington campus of University of Nottingham(52.8N, 1.2W) from May to September in 2008. The experiment totally laid out 16 plots, grass(cv Long Ley) and beans(Vicia faba cv Clipper) were planted in eight plots, respectively. However, only four grass and bean plots were stressed by the CO2 leakage, and CO2 was always injected into the soil at a rate of 1 L·min-1. The canopy spectra were measured using ASD instrument, and the grass was totally collected 6 times data and bean was totally collected 3 times data. This paper study the canopy spectral characteristics of grass and beans under the stress of CO2 microseepages through the field simulated experiment, and build the model to detect CO2 microseepage spots by using hyperspectral remote sensing. The results showed that the canopy spectral reflectance of grass and beans under the CO2 leakage stress increased in 580～680 nm with the stressed severity elevating, moreover, the spectral features of grass and beans had same rule during the whole experimental period. According to the canopy spectral features of two plants, a new index AREA(580～680 nm) was designed to detect the stressed vegetations. The index was tested through J-M distance, and the result suggested that the index was able to identify the center area and the core area grass under CO2 leakage stress, however, the index had a poor capability to discriminate the edge area grass from control. Then, the index had reliable and steady ability to identify beans under CO2 leakage stress. This result could provide theoretical basis and methods for detecting CO2 leakage spots using hyperspectral remote sensing in the future.