ABEESI: A remote sensing index for the fine detection of multitype abnormal bareness events on Earth’s surfaces

• Proposed an index for the fine detection of multitype abnormal bareness events on Earth’s surfaces. • The proposed index is less affected by different background environments. • The proposed index can effectively detect multitype abnormal bareness events on Earth’s surfaces. Frequent abnormal bareness events on Earth’s surfaces (ABEES) seriously threaten ecosystem stability and the safety of people’s lives and property, and it is urgent to carry out fine detection and timely warning. However, existing remote sensing indices for ABEES detection are only applicable to one or two types of ABEES and easily confuse different background information with ABEES, which makes it difficult to achieve fine detection of multiple types of ABEES in complex and unknown scenes. Therefore, this study proposed an index for the fine detection of multitype ABEES (ABEESI). Specifically, this study first conducted a comparative analysis of the shape and position differences in spectral curves between various ABEES and different ground objects. Then, representative bands that effectively capture these differences were selected to construct ABEESI. To evaluate the performance of the proposed index, six common types of ABEES and a large unknown scenario (Mongolia) were selected for testing. The validation results indicated that ABEESI achieves an F1-score exceeding 0.84 for detecting multiple types of ABEES, which is significantly better than traditional remote sensing indices, especially in cases of landslides, debris flows, and sand boils. Meanwhile, ABEESI still maintains good performance in a large unknown scenario, with a detection accuracy F1-score still exceeding 0.81. Overall, ABEESI is less affected by different background environments, effectively overcoming the limitations of traditional remote sensing indices that easily confuse ABEES with bright barren lands, artificial surfaces and other background environments. Moreover, ABEESI can effectively reflect different ABEES states, which makes it possible to provide an early warning in the early stages of ABEES and terminate warnings when the anomaly areas begin to recover.