Early detection of landslide hazard threatening oil pipeline in UNESCO-protected Hyrcanian Forests (Iran) using DInSAR and MaxEnt modeling

Abstract

Landslides in the Hyrcanian Forests, a UNESCO World Heritage site, pose a serious threat to oil pipelines, potentially resulting in catastrophic environmental consequences. The dense canopy cover of this unique forested massif also poses a challenge to conventional methods, such as field survey and optical remote sensing which makes it difficult to accurately assess and mitigate landslide hazards. This study aims to assess landslide susceptibility zonation and its implications for oil pipeline risk assessment using a synergistic approach that combines differential interferometric synthetic aperture radar (DInSAR) with the maximum entropy (MaxEnt) model. Landslide movements in the dense forest were extracted and measured by processing L-band ALOS-2 PALSAR-2 synthetic aperture radar images using the DInSAR technique. A total of 120 landslide patches were detected and these Satellite-derived landslide data, along with nine landslide conditioning factors, were used in the MaxEnt model for landslide susceptibility zonation. The generated landslide susceptibility map with the area under the receiver operating characteristic curve of 0.845 revealed that approximately 26.74 km of the pipeline crosses areas of high or very high landslide hazard. Field surveys and subsequent investigations validated the accuracy of landslides detected by DInSAR and MaxEnt predicted susceptible areas, confirming the reliability of our integrated approach. This research pioneers an integrated radar imaging and predictive modeling approach for landslide risk assessment in densely forested hilly regions.