
The asset wealth dataset is generated using a remote sensing–based asset wealth estimation model. Compared with similar studies, this model demonstrates higher accuracy, explaining 84% of the spatial variation in asset wealth at the community or village scale and 71% of the temporal variation at the subnational level. By effectively integrating nighttime light imagery with Landsat data, the model incorporates information on nighttime economic activity while compensating for the limitations of nighttime lights in unlit areas through the inclusion of daytime imagery.