Estimating population of Java Island based on nighttime lights data

Visible Infrared Imaging Radiometer Suite (VIIRS) instruments produce nighttime light (NTL) images showing artificial light emissions, which are closely related to human existence as an indicator of built-up areas, especially settlements. This study was designed to determine the capability of NTL data to estimate population based on its correlation with the intensity of artificial light emission and lit area by conducting multivariate linear regression analysis using Python in Google Colaboratory. The research area consisted of regencies/cities on Java Island, home to the largest population in Indonesia, that had different rates of development. The samples were city/regency population data divided randomly with a 7:3 ratio into training and testing samples. The model was created using a training samples with correlation coefficients of 0.857 for 2015, 0.855 for 2017, and 0.852 for 2019 and then validated by calculating the percent error ( error) between the estimated and actual populations using the testing samples. The results showed an average of 1.44 error, and from this high accuracy indicator, the study concludes that NTL can be used to estimate the population. However, this estimate only serves as an overview because the model was developed based on small-scale cases, resulting in less detailed outcomes. © 2021 SPIE.