World population distribution: smoothed representations

摘要

Population densities and latitudes and longitudes for more than two thousand regions of the nations of the world are collected. Spline functions are fit to the logarithms of the densities. For the most smooth spline, more than 80% of the variation can be explained by latitude alone by predicting steady increases from south to north. For the most rough, more than 92% is accounted for by latitude and longitude. The data set is then expanded to 6,520 observations, primarily through the incorporation of detailed units for the United States, Canada, Brazil and India—and smoothed. Among the highest fitted values are those in the Tokyo area. Denver, Winnipeg and Salt Lake City, population concentrations in broad sparsely settled areas, are among the largest residuals. Local maxima (Mexico City, Calcutta and Düsseldorf, for instance) and minima (points in the Sahara and Northern Tibet) are identified, after fitting the spline on a fine spherical grid.