Object-based Mapping of Landslides with Remotely Sensed Data and Python Programming in Taq Taq, Iraq

Abstract

This paper involves object-based landslide detection, classification and susceptibility mapping with a specific focus on how Python scripting can complement automation, reproducibility and efficiency of analysis. A combined Geographic Object-Based Image Analysis (GEOBIA) and Python based Analytical Hierarchy Process (AHP) model was used to ascertain the susceptibility to landslides in the Taq Taq, Erbil Governorate, Kurdistan Region of Iraq. Data was standardized, reclassified, and weighted with AHP, and Python scripts automated the processing of the raster and weighted overlay and weighted statistical computation. The findings indicate that landslide is most controlled by LULC (24.97%), slope (14.86%), soil type (12.06%) and NDVI (11.47%) with the least contribution of aspect and stream proximity. The resulting Landslide Susceptibility Index (LSI) map suggests that about 91.8, 8.1, and almost 0.002 percent of the study area belong to the low, moderate and high susceptibility categories respectively. The high susceptibility areas taking up an insignificant portion of the study area. The suggested GEOBIA-Python-AHP system is highly promising in the reproducible landslide occurrence evaluation and would be of great help in land-use planning and hazard reduction in mountainous areas. The study concludes that Python programming has intersected with GEOBIA, and GIS-based mapping is a paradigm shift towards dynamic and open-source and reproducible landslide modelling.