Abstract: It is of great significance to effectively identify the poverty-stricken areas and appreciate the mechanism of their evolution based on remote sensing to strengthen the coordination between poverty alleviation and rural revitalization. In this paper, a method of constructing multidimensional poverty index by integrating nighttime imagery with the correlation coefficient and analytic hierarchy process is proposed. And at the same time, the accuracy of the model is tested. With the spatial analysis technology of geographic information system （GIS） and the characteristics of temporal continuity of nighttime light imagery, this paper makes a detailed analysis of the dynamic development of poverty-stricken counties from the perspective of spatial-temporal evolution. The results demonstrate that from 2003 to 2013, the proportion of the poverty-stricken counties in Sichuan dropped from 46.45% to 28.42%. The variation coefficient illustrates a decreasing trend, which proves that the poverty gap within Sichuan province is slowly narrowing. In the spatial distribution evolution map from 2003 to 2013, most districts and counties in Ganzi, Aba and Liangshan autonomous prefecture are in long-term multidimensional poverty. The increase of multidimensional poverty index is not miraculous in the extreme poverty and extremely affluent counties, but the increase is miraculous in the middle counties. In the hot spot analysis, the hot spot area and the cold spots gradually show an east-west trend. The results of this study can provide a prospective basis for the accurate formulation of poverty alleviation policies in areas with complicated local alienated poverty situation.

Key words: nighttime light imagery, analytic hierarchy process, multidimensional poverty index, space-time evolution, hot spot analysis