Jihong Zhang, Xiaoqing Zuo, S J Guo, Cheng Huang, Xuefu Yue
Interferometric Synthetic Aperture Radar (InSAR) tropospheric delay (TD) correction in alpine-canyon regions has long presented severe technical challenges. This study focuses on the spatio-temporal heterogeneity and multi-factor quantitative analysis of InSAR TD in these regions, aiming to provide a theoretical basis and scientific support for developing high-precision TD models. Taking two typical alpine-canyon regions, Yongsheng and Fugong, as study areas, this study employed spatio-temporal fitting functions and coupled with multiple environmental factors by integrating Pearson correlation analysis, principal component analysis (PCA) and the geographical detector (GeoDetector) to quantitatively analyse the complex stratified delay (Dstra) and turbulent delay (Dturb). The results demonstrate that the spatio-temporal heterogeneity of Dstraand Dturbin alpine-canyon regions is significantly higher than in flat terrains. Without vertical correction, the bias of Dstracan exceed 10 cm, and using average vertical correction coefficients introduces a bias of up to 1.5 cm. The combined model of the Gaussian function (GF) and cubic spline function (CSF) effectively balances fitting accuracy and model complexity. Joint analysis via Pearson correlation analysis, PCA and GeoDetector reveals that the explanatory power of distance to water bodies, dTemp and dPWV for Dturbis significantly higher than that of slope and relief undulation. Among these factors, dPWV exhibits the highest stability in explanatory power. Furthermore, the relative explanatory power of various environmental factors fluctuates across subareas. Therefore, a zonal modelling strategy is recommended to mitigate the impacts of local micro-environmental differences. These findings contribute to improving the accuracy of InSAR surface deformation monitoring, thereby supporting early identification and risk warning of regional geological hazards.