Better Night Lights Data, For Longer*
| Published date | 01 June 2021 |
| Author | John Gibson |
| Date | 01 June 2021 |
| DOI | http://doi.org/10.1111/obes.12417 |
Better Night Lights Data, For Longer*
JOHN GIBSON
Department of Economics, University of Waikato, Hamilton, New Zealand
(e-mail: jkgibson@waikato.ac.nz)
Abstract
Night lights data are increasingly used in applied economics, almost always from the
Defense Meteorological Satellite Program (DMSP). These data are old, with production
ending in 2013, and are flawed by blurring, lack of calibration and top-coding. These
inaccuracies in DMSP data cause mean-reverting errors. This paper shows newer and
better VIIRS night lights data have 80% higher predictive power for real GDP in a
cross-section of 269 European NUTS2 regions. Spatial inequality is greatly understated
with DMSP data, especially for the most densely populated regions. A Pareto
correction for top-coding of DMSP data has a modest effect.
I Introduction
Satellite-detected night time lights data are increasingly used in applied economics.
The first article in an economics journal using these data was in 2002 (Sutton and
Costanza, 2002) but it was only once Henderson et al. (2011, 2012) published in
the American Economic Review using night lights that many economists became
professionally interested in such data. One indicator of the growing use of these
data comes from a search of the economics literature in IDEAS/RePEc, which
shows 175 records (articles and working papers) since 2010 have either ‘night
lights’or ‘night time lights’or ‘luminosity’in their details.
1
The production of
papers using night lights data is increasing, as 41 of these 175 records date from
either 2019 or 2020.
There is a problem with much of the economics research using night time lights
data. Most studies use the Defense Meteorological Satellite Program (DMSP) data,
which are old and not very accurate. For example, of the 41 IDEAS/RePEc records
from 2019 or 2020, all but four use DMSP data. The inaccuracies in DMSP data
JEL Classification numbers:E20, R12.
*This paper was begun while visiting the Centre for the Study of African Economies, Department of
Economics, University of Oxford and I am grateful for their hospitality. Helpful comments from the editor, an
anonymous referee and seminar audiences at CERDI, Oxford, and the Tinbergen Institute and assistance with
the GIS analysis from Geua Boe-Gibson are gratefully acknowledged. Financial support from Marsden Fund
project UOW1901 is acknowledged. These are the views of the author.
1
Search made on 2 April 2020. The count of 175 records excludes 60 records where the search terms capture
papers that do not use satellite-detected night lights data.
770
©2020 The Department of Economics, University of Oxford and John Wiley & Sons Ltd.
OXFORD BULLETIN OF ECONOMICS AND STATISTICS, 83, 3 (2021) 0305-9049
doi: 10.1111/obes.12417
include: blurred images (Abrahams et al., 2018) and geo-location errors (Tuttle
et al., 2013), so light is attributed to places other than where it is emitted; top-
coding,wherebrightly lit city centres get the same data values as low-density, dimmer
suburbs (Bluhm and Krause, 2018); and uncalibrated variation in DMSP sensor
amplification and inter-satellite differences that impair comparability over time and space
(Gibson et al., 2020). Also, DMSP data are increasingly out of date, as production of these
data ended in 2013. Newer and better night lights data are available from April 2012 from
the Visible Infrared Imaging Radiometer Suite (VIIRS) of instruments on the Suomi
National Polar-orbiting Partnership (NPP) satellite platform. The VIIRS data have monthly
frequency, with only a short lag (e.g. data through to February 2020 were made available in
June 2020), giving an almost real-time measure of night-lit economic activity.
The VIIRS data are far more precise than DMSP data, with 45 times greater
spatial resolution (Elvidge et al., 2013) and have no blurring or geo-location
errors.
2
The VIIRS data accurately measure the radiance of lights on earth, in a
range of lighting conditions (covering almost seven orders of magnitude while
DMSP covers less than two), while DMSP was designed to measure clouds for
short-term weather forecasts. Thus, DMSP data show effects of unrecorded changes
in sensor amplification (to keep brightness of cloud-tops constant over the lunar
cycle), in terms of temporal inconsistency and top-coding. Inter-satellite differences
also create inconsistency. The superiority of VIIRS has resulted in a rapid switch
in the scientific literature; now almost twice as many articles per year publish using
the VIIRS night lights data compared to those using the older and less accurate
DMSP data, yet economists have continued to persist with DMSP data and largely
ignore VIIRS (Gibson et al., 2020).
There are several barriers to the wider use of VIIRS night lights data by
economists. The most widely used DMSP data are annual composites that were
cleaned by scientists at the National Oceanic and Atmospheric Administration
(NOAA) to remove outliers created by ephemeral sources of light like aurora, fires,
lightning and boats. The equivalently cleaned VIIRS annual composites are
currently only available for 2015 and 2016. The monthly VIIRS data reported since
April 2012 have not had the same cleaning and outlier removal, so there is no
overlap with similarly processed DMSP data to allow a like-with-like comparison
(Gibson et al., 2019). Yet without such a comparison, it is harder to highlight
flaws in the DMSP data. While the remote sensing literature has studies showing
the superiority of VIIRS data (e.g. Chen and Nordhaus, 2015), there are no similar
studies in economics journals.
To enable comparisons that illustrate measurement error properties of the DMSP
data, this paper reports on a procedure to create cleaned annual estimates of night
lights from the monthly VIIRS data. This procedure provides overlapping annual
2
VIIRS data are allocated to grids, of about 0.45 ×0.30 km for typicalEuropean latitudes.For DMSP,the
grids are 0.93 ×0.60 km but the underlying spatial resolution of the DMSP sensor is far coarser than is implied
by this resampled grid.
©2020 The Department of Economics, University of Oxford and John Wiley & Sons Ltd
Better night lights data, for longer771
Get this document and AI-powered insights with a free trial of vLex and Vincent AI
Get Started for FreeStart Your 3-day Free Trial of vLex and Vincent AI, Your Precision-Engineered Legal Assistant
-
Access comprehensive legal content with no limitations across vLex's unparalleled global legal database
-
Build stronger arguments with verified citations and CERT citator that tracks case history and precedential strength
-
Transform your legal research from hours to minutes with Vincent AI's intelligent search and analysis capabilities
-
Elevate your practice by focusing your expertise where it matters most while Vincent handles the heavy lifting
Start Your 3-day Free Trial of vLex and Vincent AI, Your Precision-Engineered Legal Assistant
-
Access comprehensive legal content with no limitations across vLex's unparalleled global legal database
-
Build stronger arguments with verified citations and CERT citator that tracks case history and precedential strength
-
Transform your legal research from hours to minutes with Vincent AI's intelligent search and analysis capabilities
-
Elevate your practice by focusing your expertise where it matters most while Vincent handles the heavy lifting
Start Your 3-day Free Trial of vLex and Vincent AI, Your Precision-Engineered Legal Assistant
-
Access comprehensive legal content with no limitations across vLex's unparalleled global legal database
-
Build stronger arguments with verified citations and CERT citator that tracks case history and precedential strength
-
Transform your legal research from hours to minutes with Vincent AI's intelligent search and analysis capabilities
-
Elevate your practice by focusing your expertise where it matters most while Vincent handles the heavy lifting
Start Your 3-day Free Trial of vLex and Vincent AI, Your Precision-Engineered Legal Assistant
-
Access comprehensive legal content with no limitations across vLex's unparalleled global legal database
-
Build stronger arguments with verified citations and CERT citator that tracks case history and precedential strength
-
Transform your legal research from hours to minutes with Vincent AI's intelligent search and analysis capabilities
-
Elevate your practice by focusing your expertise where it matters most while Vincent handles the heavy lifting
Start Your 3-day Free Trial of vLex and Vincent AI, Your Precision-Engineered Legal Assistant
-
Access comprehensive legal content with no limitations across vLex's unparalleled global legal database
-
Build stronger arguments with verified citations and CERT citator that tracks case history and precedential strength
-
Transform your legal research from hours to minutes with Vincent AI's intelligent search and analysis capabilities
-
Elevate your practice by focusing your expertise where it matters most while Vincent handles the heavy lifting