An NDVI-Based Framework Integrating Spatial Heterogeneity Dynamics for Detecting Grassland Degradation

Abstract

Grassland degradation severely undermines the functions of grassland ecosystems. Inconsistencies in defining grassland degradation have spawned various estimation methods, leading to ongoing controversies. We argue that grassland degradation is a process-oriented phenomenon and should thus be detected by identifying specific stages and statuses within this process. Based on this assumption, we quantified the grassland changes and mapped the grassland degradation in the Three-River Source Region (TRSR) by incorporating three-dimensional indicators: greenness, spatial heterogeneity, and temporal stability. We found that most grasslands exhibited a greening trend (78.59%), while the remaining 21.41% showed a browning trend. Meanwhile, a decreasing trend in spatial heterogeneity was observed in 56.93% of the grasslands, whereas 43.07% displayed the opposite trend. By combining the above two indicators, potential preliminary stages of grassland degradation were identified. The results indicate that 52.50% of the total grasslands are experiencing indicator-based signals consistent with improving conditions, while 47.50% remain in a potential deteriorating state. After incorporating stability, we found that 52.54% of the improving grasslands are in an unstable state, whereas 55.86% of the deteriorating grasslands are in a stable state. Although the observed greening trends coincide temporally with the implementation of ecological restoration projects in the TRSR, and this temporal overlap is consistent with the interpretation that these initiatives may have supported the observed improvement, the remotely sensed stable deteriorating trends hidden beneath unstable improvements require ongoing, targeted attention.