Research
My research sits at the intersection of geospatial artificial intelligence and ecology, developing computational methods to understand how forests and landscapes change and what those changes mean for the planet.
Research Areas
Forest Disturbance Mapping
Automated detection of deforestation, degradation, and recovery using satellite time series and machine learning classifiers. Dense Landsat and Sentinel-2 stacks are used to pinpoint both the timing and the magnitude of change, so that gradual degradation is separated from abrupt clearing.
Change Detection Landsat Sentinel
Landscape Connectivity
Modeling how landscape structure and fragmentation affect species movement, gene flow, and ecosystem resilience. Graph-theoretic and circuit-based approaches turn land cover maps into networks whose weak links reveal conservation priorities.
Connectivity Graph Theory
Spatial Modeling
Developing geospatial machine learning models for ecological prediction, including species distribution models and above-ground biomass estimation. Emphasis is on spatially aware validation, so that reported accuracy survives contact with new landscapes.
GeoAI Prediction
Biodiversity Assessment
Using multi-sensor remote sensing to characterize habitats and predict biodiversity patterns across spatial scales. Fusing optical, radar, and LiDAR data captures both the composition and the three-dimensional structure that species respond to.
Biodiversity LiDAR
Approach
Code, data, and methods are released openly wherever licensing allows. Analyses are built as reproducible pipelines rather than one-off scripts, so results can be re-run, checked, and extended by others.
The methods here are designed around openly licensed Earth observation data โ Landsat, Sentinel, SoilGrids, LANDFIRE, ESA WorldCover โ so that they remain usable by researchers and practitioners without commercial imagery budgets.
Research outputs are packaged as tools people can actually run: QGIS plugins, R packages, and documented workflows, rather than results that stop at the figure in a paper.
Publications
Software
forestPulse: Analyse Forest Structure and Disturbance from Field and Remote Sensing Data
iLAND Workbench: QGIS-based iLAND Workbench for hassle-free installation and reproducible analysis workflows
Journal articles & preprints
Peer-reviewed articles will be listed here as they appear. In the meantime, the most current list is on Google Scholar.
Data & Code Availability
Analysis code and, where licensing permits, derived datasets accompanying this research are published openly on GitHub, alongside The CEDAR Lab organization. Software releases are archived on Zenodo so they can be cited with a persistent DOI.
Interested in Collaboration?
Iโm always open to collaborations at the intersection of GeoAI and ecology โ joint analyses, method development, student co-supervision, or applying these tools to your study system.