Developing a GIS-Based Mobile App for Real-Time Landslide Risk Assessment

Thesis Abstract

Landslides pose significant risks to communities situated in hilly and mountainous regions, resulting in loss of life, property damage, and socioeconomic disruptions, particularly in areas where real-time monitoring and early warning systems are inadequate. This study aims to develop a GIS-based mobile application that provides real-time landslide risk assessments to enhance early warning capabilities and inform timely decision-making among local authorities and residents. The specific objectives include identifying key geological, hydrological, and meteorological indicators associated with landslide occurrences, designing an integrated GIS framework capable of dynamic data acquisition, developing a user-friendly mobile interface, and evaluating the effectiveness of the app in real-world scenarios. The research adopts a mixed-methods approach, combining quantitative and qualitative techniques for comprehensive analysis. Quantitatively, the study utilizes a stratified random sampling technique to select 250 residents from landslide-prone communities, along with 30 local geologists and disaster management officials selected through purposive sampling, ensuring expertise and relevant experience. Data collection instruments include structured questionnaires for community members, semi-structured interviews with experts, and field data on historical landslide events compiled from government hazard records. The app’s development is informed by spatial analysis of landslide susceptibility factors, leveraging Geographic Information Systems (GIS), remote sensing data, and real-time sensor inputs. Quantitative data are analyzed using regression analysis to identify significant predictors of landslide risk, while thematic analysis is employed for qualitative interview data to understand community perceptions and decision-making processes. The app's algorithm incorporates validated models such as the factor of safety and probabilistic risk assessments, calibrated using historical landslide data. The validation process involves pilot testing in two communities over a six-month period, capturing over 50 real-time sensor readings and community responses, allowing assessment of the app’s predictive accuracy and usability. Data analysis includes sensitivity testing of the risk prediction model, using receiver operating characteristic (ROC) curves to evaluate its discriminative power, and reliability testing of the mobile interface through user feedback surveys analyzed via descriptive statistics. Expected findings of this research include the identification of key predictors, such as slope gradient, soil moisture, rainfall intensity, and land cover types, that significantly influence landslide susceptibility. The GIS framework is anticipated to produce a dynamic risk map, updated in real time through sensor integration, which accurately reflects current hazard levels. Early testing is expected to demonstrate an 85% accuracy rate in risk prediction, with positive feedback from community users regarding usability and comprehension of risk alerts. The app is projected to facilitate faster response times by local authorities and improve community awareness, thereby reducing potential casualties and damages. The study contributes to existing knowledge by integrating advanced GIS methodologies with mobile technology to address the persistent challenge of real-time landslide risk assessment in resource-constrained settings. It offers a replicable model for developing countries with similar geographic and hazard profiles, aligning with theories such as the Risk Homeostasis Theory and the Socio-Technical Systems Theory to understand how technological interventions influence community risk perceptions and behaviors. In conclusion, the research underscores the pivotal role of GIS-enabled mobile applications in disaster risk management, advocating for their integration into national disaster response strategies. Recommendations include scaling the app deployment across broader geographic regions, incorporating additional sensor technologies, and fostering community training programs to enhance resilience. Future research avenues should explore artificial intelligence algorithms for improved predictive accuracy and evaluate long-term social impacts of mobile-based disaster warning systems in vulnerable communities.