Thesis (M.Sc.) - Asian Institute of Technology, 2002

A thesis submitted in partial fulfillment of the requirements for the degree of Master of Science, School of Advanced Technologies

The DMSP/OLS city lights images of Thailand were generated from 110 scenes between January to March and September to December, 2001. The city lights images were compared with stable lights images database and a land use map. It was found that threshold technique at 33.33% of intensity of lights bared the best results. This thresholding technique removes the lights from ephemeral sources and rural infrastructures, which has been classifying as city lights. The area of lit was analyzed at provincial level. It is most highly correlated to provincial gross value of service products that produced by service sectors, and next highest is the gross provincial products (GPP), which is a measurement of provincial income. In order to analyze the potential of DMSP/OLS city lights image, multivariate linear regression methodology were used to develop a model for achieve this purpose. The results are the area lit and population estimation models. Best model to estimate area of lit from socio-economic data were proposed with accuracy of R2 equal to 0.783. Another best model to estimate population from area of lit and other socioeconomic data were propose with R2 equal to 0.715. The benefit of both models can be used to check and estimate population change, migration pattern or city expansion using comparison of two dates of DMSP/OLS images. It is concluded that it is possible to use potential of DMSP/OLS by statistical modeling to produce population estimation of Thailand quickly from space.