The team proposed a cross sensor (DMSP-OLS and NPP-VIIRS) night light data correction scheme based on the self encoder. The self encoder was modified. The EANTLI data in 2013 was taken as the input data, and the verification set of the output data was the combined NPP-VIIRS NTL data of the same year. After iterative training, EANTLI from 2000 to 2012 will be input into the model after training in order to obtain "NPP-VIIRS like NTL Data" of each year. This data set has the same parameter attributes as NPP-VIIRS night light data and data quality similar to NPP-VIIRS. In terms of temporal consistency, the first set of new night light dataset with a global resolution of 500 meters can truly reflect the temporal changes of population and light brightness at different scales, and even reflect the global economic depression from 2007 to 2010. The new night light data set solves the problem that DMSP-OLS and NPP-VIIRS cannot be used at the same time, and provides a new data source for urban problem research and other related fields.