A Study On The Spatial Distribution Of Village - Level Poverty Population

The data of poverty people is often collected using the administrative unit as statistical unit in China and it is unable to meet the demands of making pro-poor policies for the government. Therefore, it is necessary to carry out small-scale spatial distribution of poverty simulations to provide guidance for the development of pro-poor policies of the government. Land use data, elevation data, socio-economic data, and statistical data were used to conduct the poverty population distribution. Population distribution was made firstly. In the process of population distribution, IDW, multiple regression and BP network model were performed based on the correlation between land use and population density index. And then a detailed comparative was made to analyze the accuracy of the different methods. Secondly, the poverty rate is selected as the dependent variable and OLS model, GWR model, and MGWR model are taken to build MGWR_SL model to conduct spatial distributing of poverty population.In the process of poverty population distribution, logistic model and correlation analysis were performed to screen out factors affecting the poor people of the study area from the administrative village attribute and population attribute. This is a case study in which data are collected from Qianjiang District in Chongqing city and a500m×500m spatial database of the study area was built. The main conclusion is as follows:(1) In the process of the spatial distribution of rural people, by comparing the estimated value with the actual value of population in each village, and taking10%as the allowable error restriction, the reliability of IDW method is40%, while multiple regression is70%and BP neural network is83.3%;(2) collective economic income of administrative village, natural disasters, average elevation and per cultivated area are the significant influence factors on village attribute while labor illiterate/semi-literate proportion of migrant workers, the proportion of migrant workers of the labor force, agricultural machinery penetration, population density, the proportion of high school students, the prevalence of long-term and the average annual income of farmers are population attribute;(3) In the process of poverty spatial distribution, considering the spatial heterogeneity and spatial autocorrelation together, OLS, GWR, MGWR and MWGR-SL were performed to conduct the poor people distribution. Taking10%as the allowable error restriction, the reliability of OLS method is50%and GWR is60%, while MGWR is72.3%and MGWR-SL is83.3%;(4)80percent of the impoverished administrative villages distribute across mountainous terrain with RDLS more than2. The incidence of poverty is not spatially independent, but highly agglomerate. The administrative villages that have higher incidence of poverty are mostly distributed in southeast and northwest of Qianjiang District, which are rough mountainous areas.(5) The incidence of poverty which is around the Qianjiang District and near the high way is lower while the higher average elevation is, the higher poverty incidence is. So speeding up the urbanization rate, strengthening the construction of the road network, and transferring the poor people located in high elevation areas can serve as an important direction for future poverty alleviation and development of Qianjiang District.