AI-Driven Land Tenure Mapping For Sustainable Land Governance: A Review With Reference To Nepal

Abstract

Land tenure insecurity is a major challenge in Nepal, affecting millions of households across its diverse socio-economic and geographic landscape. A large portion of Nepal's land system is composed of informal, customary, and communal tenure arrangements, which are challenging to fully capture using traditional cadastral mapping techniques and are also often slow and expensive. To improve the land tenure system and sustainable land governance in Nepal, this paper examines the use of AI-driven methods integrating Machine Learning (ML), Deep Learning (DL), Geographic Information Systems (GIS), and participatory mapping. The study analyses how Artificial Intelligence (AI) can automate parcel boundary detection, land-use classification, change detection, and the integration of informal tenure into formal land administration systems, drawing on secondary data from government and international sources, evidence from comparative international practices, and Nepal's national land reform context. Secondary data indicate that approximately 35% of Nepal's land remains unregistered in the cadastral system; informality rates are particularly high in mountainous and hilly regions. The study suggests that AI-driven mapping can enhance evidence-based policymaking, reduce costs, and increase mapping efficiency. However, issues such as algorithmic bias, gaps in digital infrastructure, limited institutional capacity, and unclear legal frameworks for AI-generated outputs must be addressed immediately. The study concludes that AI-based land tenure mapping offers a viable route to just and sustainable land governance in Nepal when incorporated into inclusive governance frameworks and backed by legislative reforms.