38, 2012
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XXXIX-B8, 2012
XXII ISPRS Congress, 25 August — 01 September 2012, Melbourne, Australia
NIGHT-TIME LIGHTS AND LEVELS OF DEVELOPMENT: A STUDY USING DMSP-
OLS NIGHT TIME IMAGES AT THE SUB-NATIONAL LEVEL
Koel Roychowdhury, Simon J ones”, Karin Reinke, Colin Arrowsmith
School of Mathematical and Geospatial Sciences
RMIT University
Melbourne, Australia
simon.jones € rmit.edu.au
Commission VIII, WG VIIU2
KEY WORDS: DMSP-OLS, night-time lights, census, scale effects, socio-economic metrics, development
ABSTRACT:
This paper investigates the association between night-time lights and socio-economic metrics at the regional level. This regional
level of understanding is critical as it underpins much economic monitoring and policy-making for sustainable development. Stable
light data obtained from night time images of 2001, captured by Defense Meteorological Satellite Program — Operational Linescan
System (DMSP-OLS) satellite, are used in the study. The data records artificial lights from human habitations from the earth surface
and is a surrogate of the level of development of an area. Data on socio economic metrics at the sub-national level for the year 2001
for the state of Maharashtra in India have been sourced from Primary Census Abstract of India, 2001. However, most of the socio
economic variables are not available at the village level. This paper describes the process of deriving maps of census metrics not
collected by Indian census for small regions (such as villages) using DMSP-OLS images that are otherwise unavailable. Linear
regression models with correlation coefficients ranging from 0.75 to 0.90 (p « 0.05) at the district and the taluk level from the night-
time satellite images were used to predict these census metrics for villages. Maps are produced for villages. Errors associated with
the Modifiable Areal Unit Problem (MAUP) and ecological fallacies are also discussed. The paper concludes with an overall
assessment of the results at these various spatial scales.
1. INTRODUCTION
Night-time satellite images of the earth are recorded by the
Operational Linescan System onboard the Defense
Meteorological Satellite Programme (DMSP-OLS) group of
satellites. This is a passive sun synchronous satellite capable of
capturing artificial lights at nights from the earth surface in the
visible and near-infrared regions of the electromagnetic
spectrum. Initially the satellite was launched with the purpose
of recording cloud covers and cloud — top temperatures over the
earth surface by the US Department of Defense (DoD).
However, in late 1970s it was discovered that the data obtained
from the OLS sensor can record lights from cities at night (Croft
1978; Welch & Zupko 1980). The digital archives of the
datasets are available since 1992. From that time, this dataset
had been used for a wide range of applications. The data was
used to estimate the population of the countries of the world
(Elvidge, CD et al. 1997) and estimating population without
access to electricity (Doll, C. N. H. & Pachauri 2010). The first
global map of the GDP purchasing Power Parity (GDP-PPP)
(Doll, C. N. H., Muller & Elvidge 2000) and economies (Ghosh
et al. 2009) were mapped using DMSP-OLS night-time images.
Gas flares on the earth surface (Elvidge, C et al. 2009) were also
mapped using this data product. Radiance calibrated DMSP-
OLS data were used to map urban and suburban extent around
cities (Roychowdhury, K, Jones, SD & Arrowsmith, C 2009;
Roychowdhury, Taubenboeck & Jones 2011) and for modelling
socio-economic parameters as well as urban population data
(Doll, C. N. H. 2008; Doll, C. N. H, Muller &
Morley 2006; Doll, C. N. H., Muller & Elvidge 2000; Sutton
2003). However, most of these works were carried out at the
global and national level. In the past few years, the utility of the
night-time datasets were assessed at the sub-national level. At
the smaller spatial scale, radiance calibrated DMSP-OLS
dataset was used to estimate population and other socio
economic metrics for the state of Maharashtra in India
(Roychowdhury, K, Jones, S & Arrowsmith, C 2009). High and
low gain radiance calibrated data was used separately for
predicting census metrics at different spatial scales
(Roychowdhury et al. 2011a). The association between night
lights and GDP estimates for India at the district level was
carried out using DMSP-OLS night-time images from 2008
(Bhandari & Roychowdhury 2012).
This paper builds on the previous works of the authors. Satellite
images captured by DMSP-OLS were used by the authors to
propose a surrogate census method for small administrative
regions (that is, districts and taluks) (Roychowdhury et al.
2011b; Roychowdhury et al. 2010). The satellite images are
used in association with the data from last completed Indian
census (also conducted in 2001). The study focuses on the state
of Maharashtra in western India (fig 1). This paper examines the
application of the models proposed to predict and map the
metrics for the villages that are unavailable from traditional
census statistics. These metrics include: number of female
literates per square kilometre;
*
Corresponding author. This is useful to know for communication with the appropriate person in cases with more than one author.
93
