Implementing night light data as auxiliary variable of small area estimation

Along with the growing popularity of the small area estimation method, the need to utilize good auxiliary variables also increases. Remote sensing data, such as night light imagery, offers advantages such as time-cost efficiency and global coverage but is easily accessible. This research aims to implement night light intensity as an auxiliary variable for the EBLUP model to estimate per capita consumption expenditure at West Java in 2018. This research employs three scenarios of auxiliary variables usage in EBLUP model construction: official data, night light intensity, and the combination between both data. The results show that night light intensity is an efficient auxiliary variable for estimating per capita consumption expenditure. Furthermore, the EBLUP model with a combination of official data and night light as auxiliary variables gives the best accuracy with coefficient of variation (CV) as evaluation.随着小域估计（Small Area Estimation）方法的日益普及，对优质辅助变量的应用需求也不断提升。遥感数据（如夜光影像）兼具时效成本效益与全球覆盖范围的优势，且易于获取。本研究旨在将夜光强度作为辅助变量，应用于经验最佳线性无偏预测（EBLUP）模型，以估算2018年西爪哇省的人均消费支出。本研究在EBLUP模型构建中设置了三种辅助变量使用场景：仅使用官方统计数据、仅使用夜光强度数据，以及同时结合两类数据。研究结果表明，夜光强度是用于估算人均消费支出的高效辅助变量。此外，以官方统计数据与夜光数据结合作为辅助变量的EBLUP模型，在以变异系数（CV）作为评价指标时表现出最优的预测精度。