Integrating multi-index remote sensing and machine learning for mangrove dynamics assessment using sentinel-2 imagery

Area of Science:

• Ecology
• Environmental Science
• Remote Sensing

Background:

• Mangrove ecosystems are crucial for coastal protection, carbon storage, and biodiversity.
• Indus Delta mangroves face threats from environmental changes, necessitating monitoring.
• Understanding mangrove dynamics is key for conservation and management.

Purpose of the Study:

• To assess mangrove dynamics in the Indus Delta from 2018 to 2024.
• To evaluate the effectiveness of integrating remote sensing and machine learning for mangrove monitoring.
• To provide data for conservation and restoration planning.

Main Methods:

• Utilized Sentinel-2 imagery and nine spectral indices.
• Employed random forest machine learning for multi-temporal classification.
• Conducted change detection analysis to assess land-cover transitions.

Main Results:

• Mangrove extent increased by 65.6%, from 70.69 km² in 2018 to 117.05 km² in 2024.
• Regeneration (50.54 km²) significantly surpassed degradation (3.43 km²), indicating ecosystem recovery.
• Expansion occurred via colonization of tidal flats and succession from other vegetation types.

Conclusions:

• Integrated remote sensing and machine learning provide accurate mangrove monitoring.
• The Indus Delta shows strong mangrove ecosystem recovery and expansion.
• Findings support effective mangrove conservation, restoration, and management strategies.