| Age-specific Death Rate is the key parameter to calculate life expectancy by abridged life table.However,in the statistical yearbook,the subsections of different age groups are normally larger than those in abridged life table and the death population of the corresponding subsection is lacking,which makes the Age-specific Death Rate cannot be calculated directly.Though population data of sampling years(including year of Census or year of population sampling survey(l%))can provide age groups of average population(death population included)which are the same as those in abridged life table,sampling interval is too long.How to calculate life expectancy at provincial level in non sampling years scientifically is of great significance.Besides,in academia,stable lights are mainly used to reflect and analyze the development of regional economies and the level of energy consumption and seldom monitor human health.The life expectancy reflecting the health status of the population comprehensively is influenced by many factors,such as the level of economic development(quality of medical service),habits and lifestyle and(natural)social environment.Thus,it is worth studying deeply whether there is a correlation between life expectancy and night lights closely related to night illumination(lifestyle and habits),energy consumption and light pollution(reflection of environmental conditions).In terms of the above two points,the following studies have been carried out in the paper.(1)An enhanced density algorithm for time-series life expectancy which estimates life expectancy at provincial level in non sampling years by Lagrange Interpolation is proposed for the first time by making full use of statistical yearbook combining census data.(2)Spatiotemporal coupling between sum of lights and life expectancy at provincial level can be verified on the basis of panel regression and spatial correlation.31 provinces in China are selected in the experiment except Hong Kong,Macao and Taiwan.Life expectancy of different provinces and years estimated by enhanced density algorithm for time-series life expectancy and corresponding sum of lights are analyzed in the perspective of time and space from 1997 to 2013.The results are as follows:(1)The accuracy of life expectancy estimated:Lagrange Interpolation(two order)on ratio of age group population>Lagrange Interpolation(first order)on ratio of age group population>Lagrange Interpolation(first order)on official life expectancy.The maximum,minimum and average life expectancy estimated of 31 provinces in 17 years is respectively 91.63 years old(Tianjin),65.23 years old(Yunnan)and 76.05 years old.(2)Panel data are formed by life expectancy estimated and sum of lights extracted at provincial level.It fits the fixed-effects variable-coefficient model,that is,time series expressions between life expectancy and sum of lights at provincial level vary from province to province.(3)In 17 years,spatiotemporal analysis between life expectancy and sum of lights at provincial level shows that:for time series,the expressions are mainly parabolic models(including exponents and power functions);for space distribution,positive spatial cross-correlation is formed.Spatiotemporal coupling is obvious.Details are as follows:(a)Life expectancy at provincial level,for time series,a steady growth is revealed.Moreover,there is a negative correlation between the degree and level of growth in the majority of cases(Tibet,Guizhou,etc.are not included);for space distribution,positive spatial autocorrelation is clear:high-high values are mostly located at the East Coast,low-low values are distributed in the Central and Western Regions.(b)Sum of lights at provincial level,for time series,the overall trend is increasing and there does exist obvious differences in the range of increase:the eastern coastal provinces are generally higher than the western provinces;for space distribution,positive spatial autocorrelation is observed:high-high values are mostly distributed in the north and north-east of China;low-low values are appeared at the Midwest.(c)Sum of lights and Life expectancy at provincial level,for time series,the regression curve is mostly convex(except for Xinjiang,Fujian,etc.);for space distribution,positive intercorrelation is revealed:high-high values are mostly located at the northeast and Jiangsu or Zhejiang;low-low values are distributed in Western Regions.Enhanced density algorithm for time-series life expectancy,which can effectively estimate life expectancy at provincial level in non sampling years,provides reference for solving the problems of missing time series data in scientific research.Spatiotemporal coupling of life expectancy and sum of lights at provincial level,which is verified,offers a new perspective for estimating and analyzing life expectancy at provincial level from the area of remote sensing.It promotes spatio-temporal data mining of remote sensing and development/integration with medical geography. |
PDF Full Text Request