Sahar Sharifi
About
My research project
Using Earth Observation for the assessment of Urban, Peri-Urban and Wetland SitesMy PhD research focuses on leveraging Earth Observation data for the assessment and monitoring of urban, peri-urban, and wetland environments. The project explores how innovative EO and AI capabilities -particularly Very High Resolution (VHR) satellite imagery- can be integrated with ground-based datasets to evaluate key habitat attributes such as quality, connectivity, and ecological networks. These analyses aim to improve our understanding of ecosystem structures and contribute to nature recovery and biodiversity enhancement across selected research sites.
In recent years, machine learning, particularly deep learning, has made substantial progress and has become a transformative component of EO-based environmental monitoring. The integration of these data-driven methods with high-resolution remote sensing imagery enables more accurate, efficient, and scalable analyses of complex ecological systems. This fusion between AI and EO advances scientific understanding and provides actionable insights for sustainable urban planning, ecosystem restoration, and adaptive environmental management.
Supervisors
My PhD research focuses on leveraging Earth Observation data for the assessment and monitoring of urban, peri-urban, and wetland environments. The project explores how innovative EO and AI capabilities -particularly Very High Resolution (VHR) satellite imagery- can be integrated with ground-based datasets to evaluate key habitat attributes such as quality, connectivity, and ecological networks. These analyses aim to improve our understanding of ecosystem structures and contribute to nature recovery and biodiversity enhancement across selected research sites.
In recent years, machine learning, particularly deep learning, has made substantial progress and has become a transformative component of EO-based environmental monitoring. The integration of these data-driven methods with high-resolution remote sensing imagery enables more accurate, efficient, and scalable analyses of complex ecological systems. This fusion between AI and EO advances scientific understanding and provides actionable insights for sustainable urban planning, ecosystem restoration, and adaptive environmental management.
ResearchResearch interests
- Earth Observation
- Remote Sensing
- Machine Learning
- Deep Learning
- Data analysis
- GIS
- Climate Change
Research interests
- Earth Observation
- Remote Sensing
- Machine Learning
- Deep Learning
- Data analysis
- GIS
- Climate Change
Sustainable development goals
My research interests are related to the following:
Publications
Tropical forests act as major carbon sinks and regulate the atmospheric carbon content. Conventional methods for quantifying carbon stocks are highly dependent on the accuracy of spatial mapping of land use and land cover (LULC). Recent developments in high-resolution remote sensing technology have increased the potential to produce accurate LULC classifications as a prerequisite for assessing such carbon stocks. This research examines the advantages of remote sensing techniques to support accurate LULC mapping via satellite imagery at 3 m spatial resolution with eight spectral bands. PlanetScope images were used for the LULC classification for the Budongo Forest Reserve (BFR) area in Uganda. Using the detailed LULC map, the InVEST model was employed to estimate carbon stocks. Aboveground biomass estimation was achieved by combining GEDI LiDAR data with vegetation indices derived from PlanetScope imagery. The study produced a 3-metre-resolution LULC map. The classification performance was validated for accuracy based on ground-truth data, yielding a kappa coefficient of 0.80. Aboveground biomass mapping achieved 3 m resolution with an R² value of 0.84. The total carbon stock estimate for the BFR area, derived from these approaches, was 11 120 727 Mg C, with an average density of 136 Mg C ha−1. These findings underscore the importance of high-spatial-resolution satellite data in enhancing our understanding of carbon stock estimation and can be utilised to inform comprehensive strategies for the effective management of terrestrial carbon stocks.