ABCDEFGHIJKLMNOPQRSTUVWXYZAA
1
2
SNTopic TitleShort decription of this MSc topic (e.g. the problem issue at hand)Thematic AreaType of ResearchObjectives and Methodology proposed for this MSc topicReference for further Reading
3
4
1Building Footprint Extraction and Geometry OptimizationUrbanization in the Global South is accelerating, resulting in diverse building patterns and typologies that differ significantly from those in the Global North. Current AI models for building footprint extraction are predominantly trained on datasets from developed regions, limiting their ability to adapt to localized architectural diversity and irregular urban layouts. This mismatch leads to significant inaccuracies in footprint extraction, particularly in densely built environments where precise geometry is critical. Additionally, there is a scarcity of localized datasets that reflect the unique structural characteristics of urban areas in the Global South. These gaps hinder essential applications such as urban planning, disaster management, and infrastructure development, all of which require reliable building footprint data. Without research to optimize AI models for these contexts, cities in the Global South will continue to face challenges in managing rapid urbanization effectively. An academic inquiry is needed to address these limitations, leveraging machine learning and geospatial technologies to improve accuracy in footprint extraction and geometry optimization.Urban DevelopmentApplied ResearchThe objectives of this project are: 1) to develop methods for accurate building footprint extraction in dense urban areas; 2) to improve geometric accuracy for planning purposes; and 3) to validate the models using urban case studies. The methodology involves geospatial data analysis, machine learning, and real-world validation.
5
6
2Building Damage Assessment and Categorization / Automated Detection of Rural Road Issues Using AIAssessing building damage following disasters such as earthquakes and floods is a critical yet labor-intensive process in developing countries, where manual surveys remain the norm. These surveys are time-consuming, costly, and prone to delays, impacting timely disaster response and recovery efforts. While AI and machine learning (ML) hold great promise for automating damage detection, existing global models are predominantly trained on datasets from developed contexts and often fail to account for the unique architectural characteristics of disaster-prone regions in the Global South. These models struggle with identifying localized damage patterns, particularly when applied to unstandardized or unconventional building structures. Moreover, there is a significant lack of localized datasets that can train AI systems to perform accurately in these contexts. This research gap delays disaster recovery and reduces the ability to allocate resources effectively. Hence , there is a research need to develop AI and ML models tailored to regional needs, enabling faster, more accurate building damage assessments and improving disaster resilience in under-resourced areas.Disaster Risk ManagementApplied ResearchThe objectives of this project are: 1) to develop an AI-based system for damage assessment and categorization; and 2) to enhance disaster response and risk management practices. The methodology involves AI modeling, image data analysis, and pilot field tests in disaster-prone areas.
7
8
3Road Network Data Classification and CategorizationEfficient transportation planning and management depend on accurate road network data, which remains scarce or unreliable in many developing countries. Existing AI models for road network classification and categorization are primarily trained on structured datasets from developed cities, making them unsuitable for the irregular, fragmented, and unplanned road patterns that dominate rural and peri-urban areas in the Global South. Additionally, the absence of standardized spatial data for roads leads to challenges in infrastructure planning, disaster response, and transportation system optimization. This gap in data quality and AI adaptability hinders access to critical services, delays connectivity improvements, and complicates emergency planning. Without reliable road network data, developing countries face significant obstacles in enhancing transportation systems or responding to infrastructure demands during crises. Research is needed to create AI models that can accurately classify and categorize road networks in developing contexts, improving mobility, urban planning, and disaster response capabilities.Transportation and InfrastructureApplied ResearchThe objectives of this project are: 1) to create models for road network data classification; 2) to categorize road types and assess infrastructure quality; and 3) to validate these models using real-world datasets. The methodology involves GIS-based analysis, machine learning, and extensive dataset validation.
9
10
4Socio-Technical Research on Humanitarian Open SpacesHumanitarian open spaces are crucial for disaster response, serving as temporary shelters, resource hubs, and emergency operation centers. However, existing global standards, such as Sphere Standards, primarily focus on identifying open spaces in urban contexts, leaving a critical gap in addressing the unique needs of rural, disaster-prone areas. Rural spaces often include agricultural land or informal communal areas, which lack standardized criteria for classification or effective use in disaster preparedness. The absence of defined parameters for rural humanitarian open spaces creates challenges for local governments and humanitarian organizations in identifying, utilizing, and integrating these spaces into disaster management plans. This gap hinders equitable and effective disaster response in vulnerable rural regions, where open spaces are often the primary refuge for displaced populations. Academic research is essential to develop criteria and guidelines that address the distinct socio-technical characteristics of rural open spaces, enabling their optimized use for disaster preparedness and response.Humanitarian and Social ResearchMixed-Methods Research- Objectives:
1) Identify gaps in open space guidelines for rural settings.
2) Develop criteria for classifying rural open spaces.
3) Adapt urban standards to rural disaster contexts.
- Methodology:
Analyze IOM reports, conduct field surveys, and use GIS to refine criteria.
11
12
5Development of a Composite Vulnerability Index for Household-Level VulnerabilityCurrent disaster risk assessments in Nepal and other developing regions predominantly focus on community-level evaluations, overlooking the granular vulnerabilities of individual households. These assessments often fail to account for localized factors such as proximity to floodplains, landslide-prone areas, or infrastructure weaknesses, which are critical for targeted disaster preparedness. Additionally, there is limited integration of survey-based and spatial data, which are essential for developing accurate multi-hazard risk assessments. The lack of detailed household-level data leads to generalized risk assessments, reducing the precision and effectiveness of disaster response efforts. This gap in disaster management planning impacts not only emergency response but also broader developmental initiatives aimed at building long-term resilience. Academic research is urgently needed to develop a household-level composite vulnerability index that integrates multiple dimensions of risk. Such a framework would improve disaster preparedness, inform equitable resource allocation, and support sustainable development planning in disaster-prone areas.Disaster Risk ManagementQuantitative Research- Objectives:
1) Define a framework for household-level composite vulnerability and risk assessment.
2) Integrate multihazard parameters for vulnerability computation.
3) Develop a household level preparedness plan template
- Methodology:
Conduct household surveys, integrate spatial datasets, and use GIS-based analysis to compute and validate the index.i
13
14
6Automated Landslide Detection Using AI ModelsLandslides are among the most frequent and destructive natural hazards in mountainous regions like Nepal, where steep terrain, fragile geology, and changing climatic patterns exacerbate their occurrence. The situation is worsened by climate change, which increases the frequency of extreme rainfall events, making landslide management even more critical. However, effective landslide management is hindered by the lack of comprehensive, real-time, and accurate data on landslide-prone areas. Traditional methods of landslide mapping and prediction are often time-consuming, expensive, and reliant on manual surveys that cannot keep pace with rapidly changing conditions. Current AI models are rarely adapted for the complex topography and geologic variations in regions like Nepal, further limiting their effectiveness. Moreover, existing models fail to integrate environmental variables such as meteorological data, geology, and socioeconomic pathways to predict future risks. There is a pressing need to develop AI-driven automated landslide detection and prediction systems that leverage satellite imagery, such as Sentinel-1, and advanced change detection techniques to provide accurate, real-time insights for disaster preparedness and mitigation efforts.Environmental Risk AssessmentMachine Learning and AI Research- Objectives:
1) Develop an AI-based landslide detection system.
2) Create a dynamic landslide inventory.
3) Simulate future landslide risks under varying climate scenarios.
4) Engage stakeholders through capacity-building workshops.
- Methodology:
Use Sentinel-1 imagery, train ML models, incorporate SSPs, and develop a public platform.
15
16
7Community-Based Early Warning SystemsFloods and other hydrometeorological disasters pose significant threats to disaster-prone regions, particularly in developing countries. Effective early warning systems (EWS) are essential for mitigating flood risks and reducing the loss of life and property. However, existing early warning systems are often cost-prohibitive, technologically complex, and inaccessible to rural or underserved communities. These systems rely heavily on advanced infrastructure and central monitoring, which limits their applicability at the community level. Furthermore, a lack of community engagement in the design and implementation of these systems reduces their effectiveness, as local populations are not fully aware of or involved in using these tools. The absence of participatory approaches and low-cost technology in EWS leaves vulnerable communities without timely and actionable warnings. There is a critical need to design community-based early warning systems that are cost-effective, use simple technology, and actively involve local populations in their development and deployment. Such systems can enhance resilience by empowering communities with tools tailored to their specific needs and resources.Disaster Risk ManagementParticipatory Research- Objectives:
1) To develop cost-effective, community-based early warning systems.
2) To engage local communities in the design and implementation process.
3) To evaluate the effectiveness of participatory approaches in improving disaster response.
- Methodology:
Develop and deploy prototypes, conduct community workshops, and measure outcomes through field trials.
17
8Suitable House Addressing System for Urbanizing Cities in Developing CountriesUrbanization in developing countries is advancing rapidly, often outpacing the development of essential governance infrastructure. One major challenge in urban governance is the lack of a systematic house addressing system, which is critical for navigation, service delivery, and disaster response. In rapidly urbanizing areas, informal settlements and unplanned growth exacerbate the problem, leading to inconsistent and inefficient address systems. This absence creates significant barriers for residents, businesses, and service providers, affecting access to essential services such as healthcare, postal delivery, and emergency response. Existing address systems are often designed without considering local cultural, geographic, or technological contexts, making them unsuitable for dynamic and fragmented urban settings. These gaps highlight the need for research into scalable and efficient house addressing systems tailored to the specific needs of developing countries. Such systems could facilitate navigation, enhance service delivery, and improve urban planning processes in cities experiencing rapid growth and informal development.Urban PlanningApplied Research- Objectives:
1) To assess the challenges of current addressing systems in urbanizing cities.
2) To propose a culturally and geographically contextualized addressing framework.
3) To validate the system’s feasibility through pilot implementations.
- Methodology:
Conduct case studies, stakeholder interviews, and prototype testing in selected cities.
18
9A Hybrid Approach for Participatory GIS Mapping to Support Project PlanningParticipatory Geographic Information System (GIS) mapping is a valuable tool for project planning in disaster management, urban planning, and community development. Traditional participatory GIS methods rely heavily on paper maps or manual ground-based drawings, which, although widely used, are time-intensive, prone to inaccuracies, and challenging to integrate with modern digital systems. At the same time, emerging technologies like drone-based mapping provide high-resolution data and efficiency but lack the inclusive engagement of participatory approaches, which are critical for community acceptance and ownership of projects. This disconnect between advanced technical precision and local community engagement limits the effectiveness of planning processes in resource-constrained regions. The absence of a methodology that integrates both approaches creates a significant gap in project execution, stakeholder involvement, and data accuracy. There is a need to develop a hybrid participatory GIS mapping framework that combines drone-based aerial monitoring with community-driven data collection. Such a system could significantly enhance the accuracy of mapping outputs while ensuring inclusivity, ownership, and sustainability in project planning and implementation.Urban Planning and Community MappingParticipatory Research- Objectives:
1) To compare traditional and hybrid participatory GIS mapping methods.
2) To evaluate the effectiveness and community perception of hybrid approaches.
3) To recommend best practices for integrating digital tools into participatory mapping.
- Methodology:
Conduct comparative studies, facilitate participatory workshops, and analyze mapping outputs and participant feedback.
19
20
10Effectiveness of Virtual Reality for Disaster Preparedness Education: Case Study of our VR Tool for Disaster Preparedness EducationEducation plays a vital role in disaster preparedness, equipping individuals and communities with the knowledge and skills to respond effectively during emergencies. However, traditional educational approaches often fail to fully engage participants or ensure long-term retention of critical information. This challenge is particularly acute in resource-limited settings, where populations frequently face diverse and localized disaster risks. Virtual Reality (VR) has emerged as a transformative educational tool, providing immersive and interactive experiences that have been shown to enhance learning engagement and retention. Despite its potential, there is limited research evaluating VR’s effectiveness in the context of disaster preparedness education, particularly in comparison to conventional teaching methods. Key questions remain about how VR can improve skill acquisition, knowledge retention, and community-wide engagement in disaster-prone areas. This research gap highlights the need to examine VR’s applicability and impact through a focused case study. By evaluating a specific VR tool designed by our team disaster preparedness awareness and education, this research could provide actionable insights into the broader potential of VR to enhance resilience-building education globally.Disaster Preparedness EducationMixed-Methods Research- Objectives:
1) To assess the effectiveness of VR in enhancing disaster preparedness education.
2) To compare VR-based learning with traditional methods in terms of knowledge retention and engagement.
3) To evaluate the usability and scalability of the VR tool for diverse user groups.
- Methodology:
Conduct pre- and post-training surveys, usability testing, and participant interviews to gather qualitative and quantitative data.
21
22
11Drones for Landslide Study: Using Drone Images from Two Different Periods to Compute Change in Landslides and VolumesLandslides are a recurrent hazard in Nepal, exacerbated by the region’s complex topography, fragile geology, and erratic climatic conditions. The Ghyapche landslide, one of Nepal’s largest and most significant, serves as a critical site for understanding landslide progression and risk dynamics. Over the years, substantial drone-based aerial imagery and ground survey data have been collected at Ghyapche, spanning different time periods between 2017 and 2023. However, these datasets remain underutilized, leaving gaps in understanding the spatial and volumetric changes in landslide activity over time. Without detailed analysis and modeling, valuable insights into the progression, triggers, and mitigation strategies for landslides remain inaccessible. Additionally, the integration of aerial drone imagery with ground-based data is an underexplored area of research that could significantly enhance landslide risk assessments. Research is needed to analyze these datasets, compute volumetric changes, and model landslide activity to inform effective disaster risk management strategies. This study could contribute to building community resilience and improving landslide mitigation efforts in Nepal.Environmental MonitoringApplied Research- Objectives:
1) To analyze temporal changes in landslide areas using drone imagery.
2) To compute landslide volume changes and identify factors influencing debris movement.
3) To engage local communities and stakeholders in understanding landslide risks.
23
24
12Using Drones for Glaciological Research: Drone Images to Compute the Glacial Melting Rate and Identify Other Indicators of Climate ChangeThe retreat of glaciers due to climate change poses severe risks for high-altitude regions, including the formation of unstable glacial lakes and increased likelihood of Glacier Lake Outburst Floods (GLOFs). Monitoring glaciers in mountainous areas like the Himalayas is challenging due to their remote locations, dynamic nature, and extreme environmental conditions. Traditional methods for glacial monitoring rely on satellite-based observations, which, while useful, lack the precision required to detect localized changes such as glacial thinning, ice crevasse formation, and lateral expansions. Drone technology offers a transformative opportunity to bridge these gaps by capturing high-resolution imagery that can detect small-scale changes in glacial dynamics. Despite its potential, there remains limited research on integrating drone-based monitoring with ground-based measurements to assess glacial melting rates and secondary indicators of climate change. Conducting research on this integration can provide a more comprehensive understanding of glacial responses to climate change, contributing valuable insights to disaster risk reduction and climate adaptation strategies for vulnerable regions.Climate Change and GlaciologyApplied ResearchObjectives: 1. To compute the rate of glacial melting using high-resolution drone images. 2. To identify secondary indicators of climate change such as changes in glacier morphology, lake volume, and surface crevasses. 3. To develop a methodological framework for integrating drone imagery with ground-based data.
Methodology: 1. Conduct aerial surveys using drones equipped with high-resolution cameras and sensors to collect glacier imagery. 2. Process and analyze data to create Digital Elevation Models (DEMs) and GIS maps of glaciers. 3. Integrate aerial and ground-based data for cross-validation. 4. Conduct temporal analysis by comparing current and historical datasets to measure changes in glacier size, volume, and melt rate.		Note: much of this (DEM generation with UAV, multi-temp change detection, volume change assessment) was already done 10 years ago by collegues in Utrecht: Immerzeel, W. W., Kraaijenbrink, P. D. A., Shea, J. M., Shrestha, A. B., Pellicciotti, F., Bierkens, M. F. P., and de Jong, S. M., 2014, High-resolution monitoring of Himalayan glacier dynamics using unmanned aerial vehicles: Remote Sensing of Environment, v. 150, p. 93-103.
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100