Mary Muthee

Detected center-pivot irrigation in Egypt using zero-shot embedding similarity search with AlphaEarth (Google DeepMind), then transferred the same detections to Saudi Arabia with no additional labels. Built entirely in Google Earth Engine using self-supervised geospatial foundation model embeddings

This project develops an end-to-end automated workflow that integrates news media-based event detection with Sentinel-1 SAR imagery for near-real-time flood mapping. The resulting flood maps are published through a web-based platform to support rapid visualization and dissemination.

This project applies Sentinel-1 SAR interferometry to map ground displacement following the October 2025 landslide in Chesongoch, Elgeyo Marakwet County. Using pre- and post-event SLC data, differential InSAR was implemented to detect surface deformation associated with slope failure.

An interactive web map to explore the spatial relationship between access to basic services and disaster impacts across Africa in 2022. The project integrates multiple geospatial datasets to support visual comparison and pattern discovery.

This infographic visualizes the relationship between basic service access and disaster impacts in Africa. The design integrates visual hierarchy and eye-tracking feedback to guide interpretation

SatCLIP embeddings to predict crisis areas, model a severity index, and estimate populations in Crisis and Emergency phases across Kenya.

Monte Carlo Maximum Likelihood on KDHS 2022 data revealing spatial variation of physical IPV at approximately 6 km resolution with 9 covariates.

Python OBIA workflow on Sentinel-2 imagery benchmarked against eCognition multiresolution segmentation for land-cover delineation.

Interactive NDVI, NDMI, MNDWI, LST, and LULC visualization of landform changes in Kampong Plam supporting sustainable land management decisions.

Flood response dashboard showing affected households, WASH distribution, food aid, and health outreach to guide humanitarian planning.

Pre and post event Sentinel-2 imagery with MSAVI to assess vegetation disturbance and surface changes from the 2025 landslide event.

DSM, DTM, and CHM generated in R using lidR and terra to classify vegetation into trees, shrubs, and herbaceous strata.

Bioclimatic predictors indicate a future southward habitat shift; precipitation of the warmest quarter emerges as the most influential variable.

Sentinel-5P analysis across pre-lockdown, lockdown, and post-lockdown phases. Nyeri Town and Mukurweini recorded the highest concentrations.

GIS risk mapping following the October 2020 M6.6 earthquake. Key hotspots near Izmir city and Seferihisar identified for response planning.

Worldwide parliamentary representation infographic based on UN SDG 5.5 data, designed with visual hierarchy and eye-tracking principles.

Attractions, protected areas, hotels, and fuel stations across Kenya with an integrated distance tool for trip planning.

County-level interactive map of partial and full vaccination rates alongside mapped health facilities across Kenya.