[DRAFT - IN PROGRESS]

This reports the findings of a brief feasibility into the use of satellite night-time imagery data to aid the identification of rural settlements for humanitarian mapping purposes. The data in question is from the US Government’s Earth Observation Group. I thought two products looked potentially useful: listed as Visible Infrared Imaging Radiometer Suite (VIIRS) and Defense Meteorological Satellite Program (DMSP). DMSP-OLS is a composite of visual imagery that’s been filtered for e.g. cloud cover and moonlight interference (see metadata). VIIRS searches a specific spectral band to identify night-time fires, and comes both as a composite raster surface and a daily list of discrete points of identified fires.

To evaluate these I compare them to the OSM buildings layer of Unity State in South Sudan*, which was recently augmented quite thoroughly by a Missing Maps mapathon. The various EOG datasets come in tiff, png and kml formats, and have been visualised using R. The code for importing, reformatting and plotting the data is posted at Github. Results follow:

First a code demo - plotting the crop of London from the same data:

File F182012.v4c_web.avg_vis.tif

• obtained using the HOT Exports utility