Remote sensing and GIS for mapping post-disturbance forest vertical structure, recovery, and management typologies.
Education
Professional Experience
At FRRC Working on satellite-based vegetation height mapping in post-disturbance forest landscapes, focusing on a ~9,000 km² Czech study area severely impacted by bark beetle outbreaks (2018–2022). This research evaluates whether AI-derived geospatial foundation model embeddings — AlphaEarth Foundations (AEF; Google DeepMind) can outperform conventional Sentinel-1/2 multi-sensor workflows when combined with Random Forest and U-Net deep learning architectures. Validated against UAV-derived canopy height models across 40 field sites, the AEF+U-Net configuration achieved R² = 0.83 and RMSE = 1.30 m, significantly outperforming all other tested configurations while reducing workflow complexity by up to 40% through the elimination of manual feature engineering.
Alongside this, I am developing a large-scale framework for mapping post-disturbance forest regeneration attributes across Central Europe using AEF embeddings, drawing on field plots distributed across the region. The work addresses the structural, compositional, and functional dimensions of forest recovery, with the aim of producing a wall-to-wall regeneration mapping pipeline that can inform forest management and resilience assessment under ongoing climate-driven disturbance regimes.
A related strand extends this from forest condition to management, mapping whether disturbed sites have been cleared or left to recover on their own. Because management choice shapes the regeneration trajectory that follows, this makes it possible to ask which post-disturbance interventions actually support recovery.
Originally from Bangladesh, I love to travel the globe and explore new culture. I enjoy cooking and nature photography. I also run a science communication YouTube channel, Canopy to Climate, aimed at making forest remote sensing research accessible to broader audiences.
