Speaker Bio & Abstract
American University of Sharjah
BiographySamy Elmahdy graduated from Ain Shams University, Cairo, Egypt. Then he obtained his Master degree in Remote Sensing from the Faculty of Geoinformatics, University Technology Malaysia, Malaysia where he also obtained a PhD degree from the Faculty of Engineering, University Putra Malaysia on the application of GIS and Geomatics Engineering in Geotechnical Engineering related issues. His achievements were a gold medal awarded the first prize in category II: professional paper in Space application 2011 of the National Space Agency Malaysia (Angkasa). Professional paper title: Application of terrain analysis to the mapping and spatial pattern analysis of subsurface geological fractures of Kuala Lumpur limestone bedrock, Malaysia. In 2017, he also won the first prize of Rashid Bin Humid Award for Scientific Outstanding, UAE. Currently, Dr. Elmahdy is working as geospatial scientist at GIS and Mapping Center for American University of Sharjah, which focuses on the Remote Sensing applications in hydrology and environmental related issues. He is successfully published more than 35 research articles on different study areas in Malaysia, Sudan, United Arab Emirates and Egypt in international conferences and peer reviewed journals.AbstractFlash flood prediction across the northern emirates, UAE using ensemble machine learning and remote sensing dataSamy Elmahdy1,* , Tarig Ali 1 and Mohamed Mohamed 2,3
(1) GIS and Mapping Laboratory, Civil Engineering Department, College of Engineering, American University of Sharjah
(2) Civil and Environmental Engineering Department, United Arab Emirates University
(3) National Water Center, United Arab Emirates University
In the northern Emirates, flash floods (FF), as a response to climate change, are the largest natural hazardous causing massive destruction and losses in human lives and infrastructures. Thee machine learning models in flash floods prediction, including Random forest (RF), boosted regression tree (BRT), classification and regression trees (CART). These models were fed with seven flash floods condition factors such as altitude, slope, minimum curvature, topographic relief, lithology, distance from streams, stream density. This process was followed by accuracy metrics (F, precision and recall) and the receiving operating characteristics (ROC) curve. This validation has shown that ensemble machine learning can produce an accurate flash flood prediction map. This is the first attempt that flash flood has been predicted and mapped across the northern emirates of the United Arab Emirates. The results exhibited that the mountainous and narrower wadis (e.g., RAK, Masafi, Fujairah, and Rol Dadnah) have the highest probability of occurrence of flash floods. The proposed approach is an effective approach to enhance flash flood prediction and prevention, produced using ensemble machine learning, which is used widely in the literature.