Slums are home to a large fraction of urban residents in cities of developing nations, but little attempt has been made to go beyond a simple slum/non-slum dichotomy, nor to identify slums more quantitatively than through local reputation. We use census data from Accra, Ghana, to create an index that applies the UN-Habitat criteria for a place to be a slum. We use this index to identify neighborhoods on a continuum of slum characteristics and on that basis are able to locate the worst slums in Accra. These do include the areas with a local reputation for being slums, lending qualitative validation to the index. We show that slums also have footprints that can be identified from data classified from satellite imagery. However, variability among slums in Accra is also associated with some variability in the land cover characteristics of slums.