Authors: Greg Yetman*, , Kytt MacManus, CIESIN, Columbia University, Jane Mills, CIESIN, Columbia University
Topics: Quantitative Methods, Population Geography, Economic Geography
Keywords: satellite, population, viirs
Session Type: Paper
Start / End Time: 8:00 AM / 9:15 AM
Room: Plaza Court 4, Sheraton, Concourse Level
Presentation File: No File Uploaded
Nighttime lights have been used as the basis for urban extent delineation and the correlation between nighttime lights and socioeconomic characteristics such as GDP and income is a basic assumption in global poverty mapping. Correlations with population distribution are also a long standing subject of study. The paper will outline a series of case studies in the application of VIIRS data at NASA SEDAC using new methodologies. One effort aims to predict total population density using NASA’s Black Marble VIIRS nightly nighttime light satellite imagery with machine learning methods and predictive algorithms. The machine learning methods include a multilayer perceptron model, a convolutional neural network model, model evaluation test harness, and a persistence model forecast. In a separate effort, VIIRS data are used to inform the measurement of the UN’s Sustainable Development Goal 7, Sustainable Energy for All, which necessitates new methods for measuring the proportion of the population with access to high quality and sustainable electricity (Indicator 7.11). The case study includes an evaluation of increases in electrification by comparing nighttime lights with nighttime population counts and settlement data sets. In a third effort, estimations of nighttime lights per capita (NLPC) on a 30 arc-second grid are produced. The underlying assumption is that there is a correlation between low NLPC values and poverty or deprivation. While this data set does not replace small area estimates it can provide an important additional, proxy, data layer with consistent methods and global coverage for estimating poverty.