Identification of vulnerable areas to flash floods using weighted sum analysis and unsupervised machine learning in arid regions of the northern Atacama Desert

• Autores: Pino-Vargas, Edwin, Huayna, German, Tapia, Ángel, Pocco, Víctor, Espinoza-Molina, Jorge, Cabrera-Olivera, Fredy, Huanacuni-Lupaca, César, Acosta-Caipa, Karina, Ramos-Fernández, Lía
• Formato: artículo
• Fecha de Publicación: 2025
• Institución: Universidad Nacional de Trujillo
• Repositorio: Revistas - Universidad Nacional de Trujillo
• Lenguaje: inglés; español
• OAI Identifier: oai:ojs.revistas.unitru.edu.pe:article/6341
• Enlace del recurso: https://revistas.unitru.edu.pe/index.php/scientiaagrop/article/view/6341
• Nivel de acceso: acceso abierto
• Materia: Flash floods; Caplina Basin; Weighted Sum Analysis; Unsupervised Machine Learning; Atacama Desert

In recent decades, global warming has triggered significant changes in the hydrological cycle, leading to various disasters, especially contrasting events such as droughts and floods. These occurrences have also been recorded in the Atacama Desert, resulting in considerable economic losses worldwide, in Latin America, in Peru, and within the study region. The primary objective of this study is to obtain fundamental morphometric parameters, including basic spatial, linear, shape, and landscape aspects through the integration of GIS tools and artificial intelligence, enabling the identification of flood-prone areas within micro-watersheds. The studied basin is located at the head of the Atacama Desert, in southern Peru, where various lithological and hydro-geomorphological structures influence its vulnerability to floods. To assess flood vulnerability in the Caplina River micro-watersheds, 16 morphometric parameters were precisely analyzed, identifying areas of high vulnerability that require basin management measures. The results show that the hydrological response of the Caplina Basin is strongly influenced by its morphometric characteristics, with micro-watersheds in the middle and lower sections exhibiting higher susceptibility to flash floods. These findings aim to support urban planning and watershed management, offering insights for policymakers to develop flood mitigation strategies and enhance infrastructure resilience.