Population Estimation Of Kunming With Remote Sensing Based On Impervious Surface

Demographic data,mainly obtained by census held once every ten years in our country and other countries of the world,is an important factor of social and economic development.The census can obtain more accurate demographic data,but it usually cost a lot of labors,material resources,financial resources and time,and the demographic data is not available within the intercensal period,which usually goes against the city planning and socioeconomic development.With the development of remote sensing technology,the remote sensing technology is applied gradually to the population estimation.Namely,the regression model between the total population and the impervious surface is established on the correlation between them,which makes up for the shortcomings of the traditional census.But the traditional population estimation model with remote sensing does not give full consideration to the correlation between other classes of the population and the impervious surface,the regression model between them is established directly,which is lack of pertinence and affects the population estimation accuracy.Based on the traditional population estimation model,the algorithm of model is improved in this paper.Firstly,The population of every town in Kunming is divided into six categories: total population,male population,female population,0-14 years old population,15-64 years old population and older than 64 years old population.Then,the correlation coefficient of Pearson of each category of population with impervious surface area is calculated respectively and the category population of having biggest absolute value of correlation coefficient is selected as the dependent variable of the model.Next,the proportion of total population and population of the selected category is selected as the model coefficient.Finally,the population estimation model is established to estimate the total population of each town in Kunming.Specific steps are following: Firstly,TM images and DEM data are used to obtain the impervious surface area of each town in Kunming in 2010.Then the traditional and improved population estimation model,between the impervious surfaces area attained before and the census data of each town in Kunming in 2010,is established.Finally the precision of the two population estimate model is compared.Main conclusions are as follows:(1)There is a negative correlation between the population and the impervious surfaces of each town in Kunming in 2010,but the correlation between the population older than 65 and impervious surfaces is most relevant.(2)Estimating population model with remote sensing is different in different region,different regions need to build a suitable population model with remote sensing.(3)Improved population estimation model has a higher population estimation accuracy than traditional model,with the mean relative error reducing nearly 10% on the whole.(4)The population of town is overestimated,but there are more population is underestimated in city region.