CNN-Based Spectral Super-Resolution of Panchromatic Night-Time Light Imagery: City-Size-Associated Neighborhood Effects.

Nataliya Rybnikova,Alexander N Gorban,Evgeny M Mirkes

doi:10.3390/s21227662

Abstract

Data on artificial night-time light (NTL), emitted from the areas, and captured by satellites, are available at a global scale in panchromatic format. In the meantime, data on spectral properties of NTL give more information for further analysis. Such data, however, are available locally or on a commercial basis only. In our recent work, we examined several machine learning techniques, such as linear regression, kernel regression, random forest, and elastic map models, to convert the panchromatic NTL images into colored ones. We compared red, green, and blue light levels for eight geographical areas all over the world with panchromatic light intensities and characteristics of built-up extent from spatially corresponding pixels and their nearest neighbors. In the meantime, information from more distant neighboring pixels might improve the predictive power of models. In the present study, we explore this neighborhood effect using convolutional neural networks (CNN). The main outcome of our analysis is that the neighborhood effect goes in line with the geographical extent of metropolitan areas under analysis: For smaller areas, optimal input image size is smaller than for bigger ones. At that, for relatively large cities, the optimal input image size tends to differ for different colors, being on average higher for red and lower for blue lights. Compared to other machine learning techniques, CNN models emerged comparable in terms of Pearson’s correlation but showed performed better in terms of WMSE, especially for testing datasets.

Highlights

Artificial night-time lights (NTL), emitted from the residential, industrial, and entertainment areas, and captured by satellites, provides researchers and policy-makers with the information for a wide range of analyses: on the human presence on the Earth [1,2,3,4,5,6,7,8], on NTL adverse effects on human health [9,10,11], on the health of ecosystems [12,13], on night sky observations [14,15,16], etc. This NTL information is currently provided by the day-night band (DNB) sensor, supported by Visible Infrared Imaging Radiometer Suite (VIIRS), and available from the Earth Observation Group site [17]
Information about NTL color is of great importance for a variety of research, since it is known, for instance, that NTL emissions of different diapasons are associated with different economic activities and land-use types [19,20,21], or that NTL in blue diapason is especially effective in melatonin suppression [22] and inducing hormone-dependent cancers [23] and obesity [24]
The analysis showed that model-estimated RGB NTL levels demonstrated a sufficient consistency with the original International Space Station (ISS)-provided RGB NTL data: Pearson’s correlation coefficients, both for training and testing sets, ranging between 0.719 and 0.963, and weighted mean squared error (WMSE) varying from 0.029 to 4.223

Summary

Introduction

Artificial night-time lights (NTL), emitted from the residential, industrial, and entertainment areas, and captured by satellites, provides researchers and policy-makers with the information for a wide range of analyses: on the human presence on the Earth [1,2,3,4,5,6,7,8], on NTL adverse effects on human health [9,10,11], on the health of ecosystems [12,13], on night sky observations [14,15,16], etc. Using CNN as a tool, we settle our task as a regression one: we form multi-layer small-size images from panchromatic VIIRS-DNB NTL and builtup area data, which are further used as CNN input images, and match them with the levels of either red, green, or blue light pixels, located in the center of the corresponding small-size image, and used as a dependent variable.

Results

Conclusion