Illegal Mining Detection in Peruvian Amazon
Geospatial analysis combining satellite imagery and machine learning
Overview
This project combines remote sensing, machine learning, and spatial econometrics to detect and evaluate the impact of illegal gold mining in the Peruvian Amazon (Madre de Dios region).
Research Question
How can satellite imagery and machine learning techniques be leveraged to detect illegal mining activity and quantify its environmental and social impacts?
Methodology
Remote Sensing & ML: - Satellite imagery processing (Landsat 8, Sentinel-2) - Random Forest classification for land cover detection - Time-series analysis for mining expansion patterns
Causal Inference: - Difference-in-Differences with spatial fixed effects - Geospatial buffer analysis for policy evaluation - Spatial spillover effects modeling
Key Findings
- Illegal mining expanded 15% annually between 2018-2023
- Significant impacts on deforestation rates in affected zones
- Mercury contamination detected in nearby water sources
- Spatial spillover effects extend up to 10km from mining sites
Policy Impact
Analysis directly informed targeted enforcement strategies and environmental protection policies at the regional government level in Madre de Dios.
Implementation
Technologies: - Python (GeoPandas, Scikit-learn, Rasterio) - Google Earth Engine for satellite data processing - Stata for econometric analysis
Citation
@unpublished{medrano2024illegal,
title={Illegal Mining Detection in Peruvian Amazon: Geospatial Analysis Combining Satellite Imagery and Machine Learning},
author={Medrano, Maykol},
year={2024},
note={Working Paper},
url={https://maykolmedrano.github.io/projects/mining-analysis.html}
}APA Format: Medrano, M. (2024). Illegal Mining Detection in Peruvian Amazon: Geospatial Analysis Combining Satellite Imagery and Machine Learning [Working Paper]. https://maykolmedrano.github.io/projects/mining-analysis.html
Full working paper and replication code available upon request.