Poverty estimation at the county level by combining LuoJia1-01 nighttime light data and points of interest

Abstract

To reduce poverty, it is important to obtain accurate information on poverty conditions in a timely manner. Previous studies indicated that nighttime light products are helpful for poverty estimation. However, there exist no studies that have investigated the potential of LuoJia1-01, a new-generation nighttime light satellite with a much finer resolution (∼130 m), for analyzing poverty. In addition, nighttime light data may not reflect daytime-only socio-economic activities. To address these problems, our study first compared the performance of LuoJia1-01 and NPP-VIIRS products in poverty estimation. Next, we incorporated point of interest (POI) information so that daytime-only socio-economic activities can be considered. Our experiments indicated that LuoJia1-01 performs slightly better than NPP-VIIRS in terms of poverty estimation at county level. More importantly, we revealed that the combination of POIs and nighttime light products can moderately increase the average estimation accuracies. These findings are expected to support poverty monitoring over large regions for long periods, which cannot be fulfilled by traditional household surveys and census.