LEVERAGING MACHINE LEARNING ALGORITHMS TO IDENTIFY POTENTIAL GEOSITES FOR GEOTOURISM PROMOTION IN ZIZ UPPER WATERSHED IN SOUTHEASTERN MOROCCO

Abstract

The present study aims to tackle the complex task of identifying optimal areas for defining geomorphosites in large regions, considering various influencing factors. The study focuses on Ziz Upper Watershed (ZUW), southeast Morocco, and evaluates the effectiveness of the commonly used machine learning classifier (MLC) in mapping potential geomorph osite areas. The identification and mapping of such areas are crucial for attracting and enhancing geotourism in the region. Initia lly, a comprehensive inventory of 120 geomorphosites was conducted, and precise measurements of three topographical parameters were taken at each site. Subsequently, the machine learning algorithm, namely Bagging was employed to develop predictive model. The performance, achieving an area under the curve (AUC) of 0.935. This models successfully identified highly favorable areas, encompassing approximately 12% of the study area. These favorable areas were predominantly situated in the western region of the study area, characterized by mountainous terrain with relatively shorter slope lengths and high altitudes. The findings of this research provide valuable guidance to decision-makers, offering a roadmap for improving the chances of discovering geomorphosites.