Revealing the causal response in landslide hydrology with MT-InSAR and spatial-temporal CCM: A case study in Jinsha River

Abstract

Convergent Cross Mapping (CCM) is a powerful tool for analyzing causality in complex dynamic systems. However, standard CCM and Geographical CCM (GCCM) focus exclusively on temporal or spatial attributes, failing to integrate both dimensions. This study introduces a spatial-temporal CCM that quantifies the state of convergence to enable batched analyses of large-scale spatial datasets. The proposed method captures variations in causality and delayed responses across different spatial locations, thereby enhancing spatial-temporal data utility and the efficiency of causal inference. Using this model, we analyzed the relationship between landslides and hydrology. The results revealed that Areas with High Displacement (AHDs) responded more rapidly to hydrological factors than stable regions, with deep-layer soil moisture (100–289 cm depth) exhibiting the strongest causality and the fastest response. Building on these findings, we identified zones of minimal instability within each AHD (areas that displayed the quickest response to hydrological changes).