Who's in the dark__satellite based estimates of electrification rates - Urban Remote Sensing: Monitoring, Synthesis and Modeling in the Urban Environment - page 201

15.1 Introduction

the DMSP estimates of electrification rates with reported rates for
86 countries published for year 2005 by the International Energy
Agency (IEA, 2006). Finally we discuss possible sources of error
and ideas for improvements.

The wide distribution of over 6 billion people across more than
200 countries has made it difficult to collect and synthesize con-
sistent data on the human condition at anything more that broad
national and sub-national units. The primary reporting is for pop-
ulation and economic variables such as Gross Domestic Product
(GDP). There is a paucity of data on quality-of-life variables and
where such data are collected variations in the methods, survey
questions used and timetables make the reports difficult to assim-
ilate into a global assessment. Satellite sensors provide one of the
few globally consistent and repeatable sources of observations.
Clearly it would be useful to have one or more satellite derived
indices that could used to estimate socioeconomic parameters,
such as the distribution of economic activity, population, and
living conditions. Historically, earth observing systems that aim
for global coverage have been designed to observe environment
and weather, not human activities. It would be sheer luck to
find data from one of these global earth observing systems that
also made a direct observation of a human activity. But there
are several examples that can be pointed to. Satellite sensors
such as NOAA's AVHRR and NASA's MODIS detect fires, many
of which are anthropogenic in origin, using a combination of
thermal bands. These same sensors detect urban heat islands and
paucity of green vegetation in heavily built up urban cores. But
the most remarkable example of a global earth observing satel-
lite sensor detection of human activity are the night-time lights
collected by the US Air Force Defense Meteorological Satellite
Program (DMSP) Operational Linescan System (OLS).
Human beings around the world use lights at night to enable
the extension of activity past sundown. The brightness of lights
is affected by multiple factors, such as population density, eco-
nomic activity, infrastructure investment, lighting type, lighting
fixtures, and even cultural preferences in lighting. Despite these
complexities, a number of studies have used night-time lights to
map phenomena which would be cost prohibitive to map based
on ground surveys. This includes the distribution of economic
activity (Doll, Muller and Elvidge, 2000; Ebener et al ., 2005;
Ghosh et al ., 2009), the density of constructed surfaces (Elvidge
et al ., 2007a), poverty levels (Elvidge et al ., 2009a), and resource
consumption (Sutton et al ., 2009).
By overlaying lights and population (Fig. 15.1) it is possible
to observe clear differences in the quantity of lighting per person
around the world. Populations in the developed world generally
have a surplus of lighting, yielding the blue-green and white
areas on Fig. 15.1. Areas with high population count and modest
lighting levels show up as pink. (in portions of India and China).
The red colors on Fig. 15.1 indicate populations where no lighting
was detected by the DMSP sensor.
In this study we develop a new application for the night-
ime lights, the estimation of electrification rates. For year 2005
the International Energy Agency (IEA) World Energy Outlook
(IEA 2006) estimated the global electrification rate at 75.6%
with 1.58 billion people living without electricity. Lack of elec-
tric power is a poverty indicator with links to conditions that
are detrimental to health and wellbeing such as lack of refrig-
eration for food, poor water quality, lack of sanitary facilities,
and limited access to health care services. We map the spatial
extent of electrification in 2006 based on the presence of DMSP
detected lighting. Combining the spatial extent of lighting with
population count we estimate electrification rates. We compare

15.2 Methods

15.2.1 Data sources

The two primary data sources for this study are DMSP night-time
lights and gridded population count. Both the night-time lights
and population gridwere fromyear 2006. National level reference
data on the extent of electrification were drawn from the Inter-
national Energy Agency's World Energy Outlook (WEO) 2006.
The DMSP-OLS visible band was designed to enable the
detection of moonlit clouds at night in the visible band. A
photomultiplier tube is used to intensify the visible band signal by
about a million fold. This enables the detection of moonlit clouds
and lighting present at the Earth's surface. NGDChas developed a
capability tomake cloud-free composites of the night-time visible
band OLS data (Elvidge et al ., 2001). Additional procedures are
used to remove ephemeral lights (mostly fires) and background
noise to produce gridded stable lights products.
There are several gridded population products available
(Fig. 15.2). We have found the US Department of Energy Land-
scan data (Dobson et al ., 2000; Bhaduri et al ., 2002) to be the
most compatible with the DMSP night-time lights. Both are
produced in a geographic projection with the same 30 arcsec-
ond grid resolution. Also, the recent Landscan products have
not used night-time lights as an input, thus there is not circu-
larity in using the two datasets. The Landscan data are spatial
allocationsofcensusreportedpopulationnumbersbasedon
models developed using three satellite derived data sources: (1)
NASA MODIS land cover, (2) the topographic data from the
Shuttle Radar Topography Mission (SRTM), and (3) high reso-
lution outlines of human settlements derived fromthe Controlled
Image Base (CIB) from the US National Geospatial Intelligence
Agency (NGA). Landscan data are referred to as population
count instead of population density, which is based on residence.
On a population density grid commercial centers and airports
have very low numbers, despite the fact that there are substan-
tial numbers of people present during certain hours. Landscan
attempts to represent the spatial distribution of population based
on person hours. Thus population is distributed across residen-
tial, commercial, industrial and public areas such as airports
and schools.
The IEA has been compiling and reporting on electrification
rates since 2002 in a publication series titled ''World Energy
Outlook''. They admit there is no internationally accepted def-
inition for electric power access and no standard method for
collecting such data. Their objective has been to report the
percentage of the population has access to electricity in their
home. Data are collected from various sources, ranging from
government agencies, international development programs and
energy research organizations. Where the country reported data
appeared contradictory, out of date, or unreliable the IEA reports
estimates based on consideration of data from similar countries,
earlier surveys, data from the international organizations, and
journal articles.