Land Suitability Modelling for Integrated Mangrove Aquaculture to Adapt to Salinity Intrusion Using Machine Learning and Deep Learning Approaches in Southwestern Bangladesh

Abstract

The Sundarbans mangrove forest, located in southwestern Bangladesh, neighbors numerous vulnerable coastal communities. The mangrove forests play a crucial role in sustaining local livelihood structures and providing key ecosystem services. Climate-induced change risks contributed saline waterlogging, and salinity intrusion in low-lying lands within polders through extreme weather events, such as cyclones, storm surges, as well as coastal flooding inundation. Furthermore, the uncontrolled expansion of brackish shrimp farming has further exacerbated soil and water salinization, posing severe threats to self-sufficient community-based livelihood options. Nature-based Solutions (NbS) offer essential tools for protecting the Sundarbans mangroves and addressing environmental challenges and livelihood threats faced by surrounding communities, including climate change, ecological degradation, loss of biodiversity, poverty, and food insecurity. The proposed Integrated Mangrove Aquaculture (IMA) practices, aligned with NbS standards and community involvement, are considered as a solution that balances mangrove ecosystem conservation and sustainable community livelihood management. Specifically, IMA provides adaptive responses to soil salinization and saltwater flooding caused by coupled climate risks, both of which increase the vulnerability of local livelihoods, aiming to develop a production model in the local saline environment that is both ecologically and economically beneficial. This study develops a machine learning-enhanced framework to optimize IMA for combating climate-driven salinity intrusion in Bangladesh’s coastal communities. We used four machine learning models that integrated 18 biophysical, environmental and socioeconomic variables to assess land suitability for IMA development. Additionally, the data-driven evaluation model provides scientific support for future planning and management, facilitating implementation of IMA in vulnerable communities and ultimately contributing to the Sustainable Development Goals (SDGs).