Enhancing precision flood mapping: Pahang's vulnerability unveiled
Dr Tahmina Afrose Keya, Siventhiran S B, Maheswaran S, Sreeramanan S, Low J An, Leela A, Prahankumar R, Lokeshmaran A, Boratne AV, Abdullah, M. T
Abstract
Flooding in Malaysia is considered one of the most impactful natural disasters. Annually, Pahang experiences substantial destruction due to floods. The aim of this research is to address the urgent issue of flood susceptibility in Pahang, Malaysia. To achieve this, a combination of Geographic Information System (GIS) and Ensemble Machine Learning (EML) will be utilized. By considering nine factors from a geospatial database that contribute to flooding, the areas prone to floods will be mapped. The mapping process will be carried out using the ArcGIS environment, and a model called Random Forest (RF)-embedding will be developed using the Ensemble Machine Learning (EML) technique. To determine the most influential factors in flooding, Feature Selection (FS) will be employed. The accuracy of the flood susceptibility models will be assessed by analysing the Area Under the Curve (AUC). Flood susceptibility mapping is a complex procedure with uncertainties. Hence, our research can contribute to flood management in vulnerable regions by improving flood models and providing spatial outcomes to help decision-makers implement risk reduction strategies.
Before start
GIS is essential for spatial data analysis and decision-making, particularly in flood susceptibility mapping. It integrates geospatial data to examine spatial relationships and visualize vulnerable areas. Machine Learning, specifically ensemble methods like Random Forest, provide advanced techniques for analysing complex datasets and improving the accuracy of flood susceptibility predictions.
Attachments
Steps
Conceptual framework
Develop an Integrated GIS-Based Framework .
The objective is to establish a robust GIS-based framework for flood susceptibility mapping in the Pahang
State. This involves compiling and integrating geospatial datasets related to topography, hydrology, land use, and climate variables to create a comprehensive database for analysis.
Apply Ensemble Machine Learning Algorithms .
The objective is to apply ensemble machine learning algorithms, such as Random Forest (RF) and Gradient Boosting Machines (GBM), to the integrated dataset to develop predictive models of flood susceptibility.
This objective includes feature selection, model training, validation, and evaluation to ensure the accuracy and reliability of the susceptibility maps.
Generate Actionable Insights for Decision-Making :
The objective is to generate actionable insights from the flood susceptibility maps to support informed decision-making and disaster management strategies. This involves identifying vulnerable areas, assessing the factors contributing to flood risk, and recommending targeted interventions and mitigation measures to reduce the impacts of floods on communities, infrastructure, and the environment.
