Carbon capture and storage (CCS) is one of the critical techniques for mitigating global climate warming which can capture and store the CO2 released during industrial processes into deep geological sites. However, there is a possible risk that the CO2 might leak from under the ground. Therefore, it is important that CO2 leakage spots on the surface of sequestration fields be detected at large scale and on a long‐term basis. The purpose of this study was to design a new model to identify CO2 leakage spots through monitoring the spectral change of vegetation on the surface of the sequestration fields. A field simulation experiment was performed at the Sutton Bonington campus of the University of Nottingham to collect the leaf and canopy reflectance of beans, which grew in soil with different levels of CO2 concentration. Visible symptoms of bean stress caused by CO2 leakage may include the reducing bean vigor and the bean leaves turning yellow. The canopy spectra and leaf spectra were processed by a smoothing, continuum removal and first derivative method. We found that, while the CO2 concentration of the soil increased, the sum of the first derivative reflectance in the 500–550 nm (SDg) increased and that in the 680–760 nm (SDr) decreased gradually compared with that of the control plots. Moreover, the newly developed ratio index SDr/SDg can be effectively used to distinguish gassed and ungassed beans at both leaf and canopy scales. Significant correlation was found between the SPAD values and the SDr/SDg indices (R2 = 0.5687, n = 197). This study indicates that hyperspectral remote sensing techniques can be used to detect CO2 leaking spots by investigating the spectral change of vegetation reflectance in the sequestration fields.