Finding Clandestine Graves: Using Geospatial Analysis to Search for Missing Persons in Baja California, Mexico

Abstraction: Three-method geospatial ML framework to prioritize search areas for clandestine graves

Key points:

• Over 100,000 people are missing in Mexico; 2,000+ clandestine graves recorded since 2007; Baja California has 12,000 disappearances since 2007
• Study uses 52 georeferenced grave sites (from freedom-of-information requests) as training data for three combined spatial methods
• Point pattern analysis (ANN + Ripley's K) found graves cluster significantly: observed mean distance 7 km vs. expected 16 km; new graves likely 18-28 km from known sites
• Accessibility + viewshed analysis defines "clandestine space" as high-accessibility/low-visibility areas; 41 of 52 known graves fall in this zone, covering 32% of Baja California territory
• Hyperspectral analysis uses Sentinel-2A/B satellite imagery to compute a Nitrogen Accumulation Index (NAI) detecting accelerated plant chlorophyll growth from decomposing remains
• Combining all three methods reduces the potential search zone by an additional 10%; results integrated into a Google Earth Engine visualization app

Connections: Amnesty International · Sentinel 2 · Geospatial Analysis · Machine Learning · Remote Sensing

Source: https://citizenevidence.org/2022/06/30/finding-clandestine-graves-using-geospatial-analysis-to-search-for-missing-persons-in-baja-california-mexico/