STUDY OF SOME APPROACHES FOR POPULATION ESTIMATION FROM REMOTE SENSING DATA

Abstract

Knowledge of the size and spatial distribution of human population in an urban area is essential for understanding social, economic, and environmental issues. Conventionally, census has been the primary source of information on population distribution and demographic characteristics. Due to the cost and frequency problems associated with census, the utility of remote sensing for population estimation has been continuously explored since the 1950s. The main objective of this dissertation is to explore the potential of the medium A resolution remote sensing data for population estimation in India at regional level using a number of existing statistical approaches and a non-parametric approach. The study area belongs to Dehradun region lies between 30° 15' 58" N to 300 24' 16" N latitude and 78° 06' 05" E to 77° 58' 56" E longitude. The study is mainly based on remote sensing data from the Landsat Thematic Mapper (TM) and Enhanced Thematic Mapper (ETM+) sensors. For population estimation, four different approaches i.e. pixel based, allometric growth, sub pixel based impervious surface and artificial neural network approach have been implemented. All the approaches are based on the relationship between the various remote sensing variables and population. The overall error in all approaches was found to be between 0.043% and 10.58%. The pixel based approach produced the least accuracy. Integration of all the variables in artificial neural network resulted in improved accuracy of population estimation from remote sensing data. Allometric growth and impervious surface, approach produced population estimates with similar