Indian census using satellite images: Can DMSP-OLS data be used for small administrative regions?

India conducts its census every ten years. Census data is collected manually in India with enumerators visiting every household in the country. Being such a vast country (in terms of area) and with a population of more than 1 billion, manual data collection is a laborious and expensive process. In response, this paper proposes a surrogate census method using DMSP-OLS night-time images. The study focuses on smaller administrative regions such as sub-districts (or taluks as they are known in the country) in the state of Maharashtra. Models are proposed using selected census metrics, and mean and standard deviation of stable lights and brightness information as obtained from the satellite images. The adjusted r2 values range from 0.2 to 0.8 at 95% confidence interval, with the majority of the metrics being moderately correlated (with r2 between 0.4 and 0.7). Generally it was found that the observed lights and brightness of big rural settlements from DMSP-OLS images have the potential for predicting certain census metrics. However, unlike larger areas such as districts where DMSP-OLS night-time images adequately predict census metrics, at the sub-district level the results need to be supplemented and validated with other information sources such as survey reports.