Estimating population from photographically determined residential land use types

The U. S. Bureau of Census conducts a census of population characteristics every ten years, a major element of which is existing population size. These latter data are essentially out of date by the time of census publication and become increasingly so as time progresses. In order to supply much needed intercensal urban population size data, a methodology is proposed which relates land use information from aerial photography to characteristic spatial population density figures obtained from U. S. Census data to generate accurate and timely urban population estimations. This proposed methodology was applied to four California cities and produced results ranging from an underestimate of 9.17% to an overestimate of 7.00%. While the error factors encountered in individual cities were significant, the composite error for the four cities together was only 4.51% (underestimation). Analysis of the suspected sources of error indicated that several minor modifications of the proposed methodology might increase the accuracy of the method. The relative ease and low cost associated with the application of the proposed methodology to four California cities suggests that the proposed methodology may be highly useful for planning and administrative purposes at local and regional levels.