A new predictive model for more accurate electrical grid mapping

What this post added

Developed a new predictive model for mapping medium-voltage (MV) electrical grid infrastructure using publicly available data sets, achieving approximately 70% accuracy within one kilometer of known grid locations. This model leverages nighttime radiance data from VIIRS, processed to remove natural and non-settlement light sources, and a modified Dijkstra's shortest path algorithm to estimate grid paths. The output is released in CSV and GeoTIFF formats for six countries, with accompanying documentation and code for replication.