Accurate vegetation destruction detection using remote sensing imagery based on the three-band difference vegetation index (TBDVI) and dual-temporal detection method

Satellite remote sensing, as an important tool for Earth observation, has been widely used to monitor various vegetation destruction events (VDEs), such as logging, wildfires and insect infestations. However, due to the spectral diversity of VDE and the complexity of background environments (BE), achieving accurate VDE detection remains a challenge. To overcome this limitation, this study developed a novel index, called the threeband difference vegetation index (TBDVI), which fully considered the spectral characteristics of both various BEs and multiple VDEs, for the accurate detection of vegetation destruction in complex scenarios. Three experiments were chosen to prove the performance of TBDVI, including (1) various possible vegetation changes; (2) various possible background changes; and (3) multiple real vegetation destruction events. The results showed that TBDVI was suitable for various vegetation change scenarios and complex background conditions, with F1 scores of 0.906-0.979. Moreover, TBDVI accurately identified the extent of VDE caused by logging, insect infestation, landslides, wildfires, and floods, with F1 scores of 0.922-0.965. Compared with existing spectral indices (VIs) (i. e., normalized difference vegetation index (NDVI), normalized difference moisture index (NDMI) and normalized burn ratio (NBR)), TBDVI has obvious advantages in reducing the impact of the background environment. In addition, TBDVI exhibits cross-sensor applicability and has potential for large-scale and high-frequency vegetation monitoring. In conclusion, TBDVI is an effective and robust spectral metric that is important for the conservation and management of vegetation resources.