Dr Milto Miltiadou (She/her/hers)

My research focus on unlocking the potential of Earth Observation data, including time-series and LiDAR, by advancing algorithms, optimising processing pipelines, and implementing tools for forest monitoring. So far, I have implemented two versatile tools for generating feature vectors for machine learning applications: DASOS (https://github.com/Art-n-MathS/DASOS), included in the GitHub Arctic Vault, manages both full-waveform and discrete LiDAR and addresses variations in point cloud density; and PlotToSat (https://github.com/Art-n-MathS/PlotToSat) offers a practical, time-efficient way to extract time-series from multi-source EO data across numerous plot regions spread across a landscape. In previous work, I introduced extraction of structural elements from 3D-Windows for detecting dead standing Eucalypt trees without tree delineation in low density LiDAR data. I also proposed a new category of data structures, integral trees, for efficient LiDAR data management during 3D polygonal model creation. Additionally, I showed that the SAR phenological cycle of the Paphos forest in Cyprus contains two peaks: the summer peak is associated with pine needle regeneration, and a February trough is potentially linked to the activity of the pityocampa pest.

Research Interests:

• Creating and exploiting time-series of multi-source Earth Observation (EO) data
• Optimising and advancing algorithms for processing and handling large-scale EO and LiDAR datasets
• Generating feature vectors from EO and LiDAR data for use in machine learning models
• Machine learning applications and advancements for forest monitoring
• 3D structural analysis and harmonisation of LiDAR data
• Phenology of multi-source EO data with particular interest in Synthetic Aperture Radar (SAR)