Can night-time light images play a role in evaluating the Syrian Crisis?

Abstract

This study investigates whether night-time light images acquired from the Defense Meteorological Satellite Program’s Operational Linescan System provide spatial and temporal insight with regard to the humanitarian aspects of the Syrian crisis. Evaluating the ongoing crisis in Syria is challenging since reliable witness reports are hard to gather in a war zone. Therefore satellite images, as one of the few sources of objective information, are potentially of great importance. We used 38 monthly Defense Meteorological Satellite Program’s Operational Linescan System composites covering the period between January 2008 and February 2014. The results indicate that night-time light and lit area in Syria declined by about 74% and 73%, respectively, between March 2011 and February 2014. In 12 of 14 provinces, night-time light declined by >60%. Damascus and Quneitra are the exceptions, with night-time light having declined only by about 35%. Notably, the number of internally displaced persons (IDPs) of each province shows a linear correlation with night-time light loss, with an R 2 value of 0.52. Through clustering the time series images, we found that the international border of Syria represents a distinct boundary between regions of differing night-time light spatiotemporal patterns. The contrast across the border increases as the region studied encompasses a wider zone on either side of the border. These findings lend support to the hypothesis that night-time light can be a useful source for monitoring humanitarian crises such as that unfolding in Syria.

Full text

This article was downloaded by: [Wuhan University]
On: 17 October 2014, At: 18:13
Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered
office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
International Journal of Remote
Sensing
Publication details, including instructions for authors and
subscription information:
http://www.tandfonline.com/loi/tres20
Can night-time light images play a role
in evaluating the Syrian Crisis?
Xi Lia & Deren Lia
a State Key Laboratory of Information Engineering in Surveying,
Mapping and Remote Sensing, Wuhan University, Wuhan 430079,
China
Published online: 17 Oct 2014.
To cite this article: Xi Li & Deren Li (2014) Can night-time light images play a role in
evaluating the Syrian Crisis?, International Journal of Remote Sensing, 35:18, 6648-6661, DOI:
10.1080/01431161.2014.971469
To link to this article: http://dx.doi.org/10.1080/01431161.2014.971469
PLEASE SCROLL DOWN FOR ARTICLE
Taylor & Francis makes every effort to ensure the accuracy of all the information (the
“Content”) contained in the publications on our platform. However, Taylor & Francis,
our agents, and our licensors make no representations or warranties whatsoever as to
the accuracy, completeness, or suitability for any purpose of the Content. Any opinions
and views expressed in this publication are the opinions and views of the authors,
and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content
should not be relied upon and should be independently verified with primary sources
of information. Taylor and Francis shall not be liable for any losses, actions, claims,
proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or
howsoever caused arising directly or indirectly in connection with, in relation to or arising
out of the use of the Content.
This article may be used for research, teaching, and private study purposes. Any
substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,
systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &
Conditions of access and use can be found at http://www.tandfonline.com/page/terms-
and-conditions

Can night-time light images play a role in evaluating the
Syrian Crisis?
Xi Li*and Deren Li
State Key Laboratory of Information Engineering in Surveying, Mapping and Remote Sensing,
Wuhan University, Wuhan 430079, China
(Received 4 September 2014; accepted 27 September 2014)
This study investigates whether night-time light images acquired from the Defense
Meteorological Satellite Program’s Operational Linescan System provide spatial and
temporal insight with regard to the humanitarian aspects of the Syrian crisis.
Evaluating the ongoing crisis in Syria is challenging since reliable witness reports
are hard to gather in a war zone. Therefore satellite images, as one of the few sources
of objective information, are potentially of great importance. We used 38 monthly
Defense Meteorological Satellite Program’s Operational Linescan System composites
covering the period between January 2008 and February 2014. The results indicate that
night-time light and lit area in Syria declined by about 74% and 73%, respectively,
between March 2011 and February 2014. In 12 of 14 provinces, night-time light
declined by >60%. Damascus and Quneitra are the exceptions, with night-time light
having declined only by about 35%. Notably, the number of internally displaced
persons (IDPs) of each province shows a linear correlation with night-time light loss,
with an R
2
value of 0.52. Through clustering the time series images, we found that the
international border of Syria represents a distinct boundary between regions of differ-
ing night-time light spatiotemporal patterns. The contrast across the border increases as
the region studied encompasses a wider zone on either side of the border. These
findings lend support to the hypothesis that night-time light can be a useful source
for monitoring humanitarian crises such as that unfolding in Syria.
1. Introduction
The ongoing Syrian Crisis, which broke out in April 2011, has caused severe humanitarian
disasters with more than 190,000 deaths (Cumming-Bruce 2014). International communities
and human rights groups are attempting to evaluate the crisis associated with humanitarian
disasters. The Humanitarian Information Unit in the US Department of State has mapped
internally displaced person (IDP) camps during Syrian Crisis from multi-source data (https://
hiu.state.gov/data/data.aspx). The Syrian Revolution Martyr Database, which documents the
number of death from opposition groups and civilians, provides both temporal and regional
information (http://syrianshuhada.com/?lang=en&). The Syria Needs Analysis Project eval-
uates the human rights violations with witness reports and geographical locations in a
geographic information system (http://www.acaps.org/en/pages/syria-snap-project). The
Syria Tracker provides geographic information system maps to record locations of deaths
during the crisis (https://syriatracker.crowdmap.com/).
Currently, witness reports are the main source for Syrian crisis evaluation. However, it
is difficult to evaluate the neutrality and comprehensiveness of witness reports during
*Corresponding author. Email: li_rs@163.com
International Journal of Remote Sensing, 2014
Vol. 35, No. 18, 6648–6661, http://dx.doi.org/10.1080/01431161.2014.971469
© 2014 Taylor & Francis
Downloaded by [Wuhan University] at 18:13 17 October 2014

violent periods. Consequently, satellite-observed remote-sensing images have been used
as a supplement to ground observations since remote-sensing data are an objective data
source which have been widely used in conflict evaluation. The United Nations Institute
for Training and Research has mapped the refugee camps and IDP camps in Syria using
high-resolution satellite imagery (http://www.unitar.org/unosat/maps/108). The Geospatial
Technologies and Human Rights Project of the American Association for the Advancement
of Science (AAAS) has mapped the battle in the Aleppo Province of Syria using high-
resolution satellite imagery (AAAS 2013).
Undoubtedly, high-resolution images are efficient tools to acquire information from
the conflict areas (AAAS 2013; Sulik and Edwards 2010), but they are expensive when
applied across broad regions. In contrast, coarse-resolution night-time light images are
newly emerging but cheap data sources for regional conflict evaluation. Therefore, they
have been used in regional and global applications (Li, Chen, and Chen 2013; Witmer and
O’Loughlin 2011; Agnew et al. 2008). This study aims to analyse the responses of night-
time light to the Syrian Crisis and evaluate its potential to monitor the conflict.
2. Study data
This study is mainly derived from three data sources: administrative boundaries, satellite-
observed night-time light images, and statistical data from human rights groups. The
international and provincial boundaries of Syria were derived from Global Administrative
Areas (http://www.gadm.org/) and are illustrated in Figure 1. As shown in Figure 1, Syria
Figure 1. The administrative map of Syria, including its provinces and neighbouring countries.
International Journal of Remote Sensing 6649
Downloaded by [Wuhan University] at 18:13 17 October 2014

adjoins Turkey in the north, Iraq in the east, Jordan in the south, and Israel and Lebanon in
the west. Syria is divided into 14 provincial regions, with Damascus, Aleppo, and Homs
as its major economic centres.
The Defense Meteorological Satellite Program’s Operational Linescan System
(DMSP/OLS) images were used as the night-time light data source in this study. The
DMSP/OLS images have been widely applied in socioeconomic studies because of their
unique capacity to reflect human activities (Elvidge et al. 1997; Chen and Nordhaus 2011;
Henderson, Storeygard, and Weil 2011; Zhang and Seto 2011; Li, Ge, and Chen 2013).
DMSP/OLS monthly composites between January 2008 and February 2014 were selected
as the images for the analysis. Due to Syria’s latitude, the DMSP/OLS images from April
to August of each year were over-saturated in all pixels. As a result, the images of 7
months in each year were used as candidate images. However, some of the images were
discarded because of data abnormalities. As a result, there were totally 38 monthly
composites for this analysis, as listed in Table 1.
Each DMSP/OLS image was registered using the image of March 2011 as a base
image. After image registration, the images have geometric errors of less than 0.5 pixel.
Then, we denoised the time series DMSP/OLS images using an algorithm based on
principal component analysis (Small 2012; Small and Elvidge 2013).
Since there is no onboard radiometric calibration for the DMSP/OLS images, an
intercalibration process is necessary to make the time series DMSP/OLS data comparable.
Invariant region-based intercalibration has been widely applied, which selects invariant
pixels as training data to construct the relationship between a standard image and a raw
image which needs calibration, and then the raw image can be calibrated to the same
radiometric level of the base image (Elvidge et al. 2009; Wu et al. 2013; Li et al. 2013). In
this research, we selected a rectangular region located in Turkey with latitude between 37°
1′51″N and 37° 9′51″N and longitude between 35° 6′55″E and 42° 5′55″Easa
potential invariant region, as the night-time light in this region looks stable based on a
visual check. We removed the variant pixels with the following process. The image of
March 2011 was selected as the base image for the intercalibration, and the background
value in this image was removed to ensure that the pixel values in the background (e.g.
sea and desert) were equal to zero. Each raw image was calibrated using the following
steps: (1) in the selected invariant region, pixels were removed that had changed between
the base image and the raw image using an outlier-removing algorithm (Li et al. 2013); (2)
the rest of the pixels, viewed as training data, were used to build a linear equation between
the base image and the raw image (Li et al. 2013); and (3) the raw image was calibrated
using the linear equation. In summary, the intercalibration process employed the automatic
algorithm (Li et al. 2013), and the only difference is that we selected a potential invariant
region manually in this analysis, whereas the traditional algorithm views the full extent as
a potential invariant region. At this stage, all the monthly composites have consistent
geometric and radiometric attributes. We selected images acquired in March 2011 and
February 2014, shown in Figure 2.
3. Analysis by administrative regions
From Figure 2, night-time light in Syria has experienced a very sharp decline since the
crisis, and many small lit patches have gone dark. The city of Aleppo, the site of fierce
battles (AAAS 2013), has lost most of its lit area. Although cities of Homs and Damascus
also lost a lot of night-time lights, the loss intensities seem less than Aleppo. Quantitative
analysis was carried out as follows.
6650 X. Li and D. Li
Downloaded by [Wuhan University] at 18:13 17 October 2014

Table 1. DMSP/OLS monthly composites used in this research.
Year
Month
January February March April May June July August September October November December
2008 F16 F16 –––––– – F16 F16 F16
2009 F16 F16 –––––– – – F16 F16
2010 F18 F18 F18 –––– – – F18 F18 F18
2011 F18 F18 F18 –––– – F18 F18 F18 F18
2012 F18 F18 F18 –––– – F18 F18 F18 F18
2013 F18 F18 F18 –––– – F18 F18 F18 F18
2014 F18 F18 –––––– – – – –
Note: ‘F16’or ‘F18’denotes the month when Syria is covered by DMSP satellite F16 or F18, respectively; ‘–’ means no data are available for that month.
International Journal of Remote Sensing 6651
Downloaded by [Wuhan University] at 18:13 17 October 2014

A direct method of evaluation is the amount of night-time light for the country and
each provincial region. First, the sum of night-time lights (SNLs) of Syria is calculated for
each month and illustrated in Figure 3. Although the SNL has shown some fluctuation
before March 2011, the night-time light in Syria has continuously declined since March
2011. The SNL of each province is also calculated between January 2008 and February
2014. The Golan Heights, as a part of Quneitra, has been controlled by Israel since 1967.
Therefore, we excluded night-time lights of the Golan Heights when calculating the SNL
in Quneitra. Figure 4 shows a sharp decline trend of the night-time lights of all the
provinces since March 2011.
And then, we calculated the change in proportions of SNL in different regions
between March 2011 and February 2014. In addition, the size of lit area, defined as the
area where the light value is greater than 3, was also retrieved for each region, and its
Month
January 2008
0
200,000
400,000
600,000
800,000
1,000,000
1,200,000
1,400,000
February 2008
October 2008
October 2010
October 2011
November 2008
November 2010
November 2011
December 2008
January 2009
January 2010
January 2011
January 2012
February 2009
March 2010
March 2011
September 2011
February 2010
February 2011
February 2012
November 2009
December 2009
December 2010
December 2011
October 2012
November 2012
March 2012
September 2012
December 2012
January 2013
February 2013
January 2014
February 2014
October 2013
November 2013
March 2013
September 2013
December 2013
Sum of night-time light values
Figure 3. SNLs between January 2008 and February 2014 in Syria.
Note: SNL is calculated based on the digital value of all pixels in a region so that SNL has no
physical unit.
(a)(b)
100 Scale
km
N
100 Scale
km
N
Figure 2. The night-time light monthly composites: (a) March 2011; and (b) February 2014.
6652 X. Li and D. Li
Downloaded by [Wuhan University] at 18:13 17 October 2014

10,000
10,000
15,000
20,000
25,000
5000
0
100,000
150,000
15,000
25,000
35,000
40,000
30,000
20,000
200,000
250,000
50,000
5000
0
0
10,000
10,000
0
15,000
20,000
20,000
25,000
30,000
30,000
40,000
50,000
60,000
5000
0
Month
Sum of night-time lights
Sum of night-time lights
Sum of night-time lights
Sum of night-time lights
Sum of night-time lights
Sum of night-time lights
Sum of night-time lights
Sum of night-time lights
Sum of night-time lights
Sum of night-time lights
Sum of night-time lights
Sum of night-time lights
Sum of night-time lights Sum of night-time lights
(a)(b)
00
20,000
20,000
30,000
40,000
40,000
50,000
60,000
60,000
70,000
80,000
80,000
100,000
120,000
140,000
0
20,000
40,000
60,000
80,000
100,000
120,000
140,000
160,000
0
2000
4000
6000
8000
10,000
12,000
14,000
16,000
90,000
10,000
0
20,000
30,000
40,000
50,000
60,000
70,000
80,000
90,000
100,000
20,000
0
40,000
60,000
80,000
100,000
120,000
140,000
160,000
180,000
200,000
10,000
0
20,000
30,000
40,000
50,000
60,000
70,000
80,000
10,000
0
20,000
30,000
40,000
50,000
60,000
70,000
10,000
10,000
0
20,000
30,000
40,000
50,000
60,000
70,000
80,000
January 2008
February 2011
October 2010
November 2011
March 2012
December 2012
September 2013
January 2014
December 2009
December 2008
Month
(d)
January 2008
February 2011
October 2010
November 2011
March 2012
December 2012
September 2013
January 2014
December 2009
December 2008
Month
(e)(f)
January 2008
February 2011
October 2010
November 2011
March 2012
December 2012
September 2013
January 2014
December 2009
December 2008
Month
(g)
January 2008
February 2011
October 2010
November 2011
March 2012
December 2012
September 2013
January 2014
December 2009
December 2008
Month
(h)
January 2008
February 2011
October 2010
November 2011
March 2012
December 2012
September 2013
January 2014
December 2009
December 2008
Month
(i)
January 2008
February 2011
October 2010
November 2011
March 2012
December 2012
September 2013
January 2014
December 2009
December 2008
Month
(j)
January 2008
February 2011
October 2010
November 2011
March 2012
December 2012
September 2013
January 2014
December 2009
December 2008
Month
(k)
January 2008
February 2011
October 2010
November 2011
March 2012
December 2012
September 2013
January 2014
December 2009
December 2008
Month
(l)
January 2008
February 2011
October 2010
November 2011
March 2012
December 2012
September 2013
January 2014
December 2009
December 2008
Month
(m)
January 2008
February 2011
October 2010
November 2011
March 2012
December 2012
September 2013
January 2014
December 2009
December 2008
Month
(n)
January 2008
February 2011
October 2010
November 2011
March 2012
December 2012
September 2013
January 2014
December 2009
December 2008
Month
January 2008
February 2011
October 2010
November 2011
March 2012
December 2012
September 2013
January 2014
December 2009
December 2008
Month
January 2008
February 2011
October 2010
Novovember 2011
March 2012
December 2012
September 2013
January 2014
December 2009
December 2008
(c)
Month
January 2008
February 2011
October 2010
November 2011
Machr 2012
December 2012
September 2013
January 2014
December 2009
December 2008
Figure 4. SNLs between January 2008 and February 2014 in different provinces of Syria: (a) Idlib;
(b) Al-Hasakah; (c) Al-Raqqah; (d) Al-Suwayda; (e) Quneitra; (f) Latakia; (g) Aleppo; (h) Hama; (i)
Homs; (j) Daraa; (k) Deirez-Zor; (l) Rif Dimashq; (m) Tartus; and (n) Damascus.
Note: SNL is calculated based on the digital value of all pixels in a region so that SNL has no
physical unit.
International Journal of Remote Sensing 6653
Downloaded by [Wuhan University] at 18:13 17 October 2014

change in proportion during the period was also calculated. The two indices are illustrated
in Figure 5. As shown in Figure 5, Syria has lost about 74% and 73% of its night-time
lights and lit areas, respectively. In addition, most of the provinces lost >60% of their
night-time lights and lit area. Damascus, the capital of Syria, is an exception which only
lost about 35% of the night-time lights and no lit area during the period, and the night-
time light has fluctuated notably, which is different from most of the provinces that
experienced continuous decline. But Rif Dimashq, the countryside of Damascus, lost
about 63% of its lights, showing that the security situation is much more severe than in
Damascus. This finding is consistent with the fact that the Assad regime has strongly
controlled the capital although the battles around the capital were intense (Blomfield
2012; Barnard 2013). Quneitra, another exception, also lost only 35% of its lights during
the period. Idlib and Aleppo are the provinces with most severe decline in night-time
light, losing 95% and 88% of the night-time lights and losing 96% and 91% of the lit
areas, respectively. In fact, the battles in these two provinces are particularly fierce (Al-
Hazzaa 2014; AAAS 2013). We also found that Deirez-Zor and Rif Dimashq lost 63%
and 60% of their night-time lights, respectively, but only 40% and 42% of their lit area.
We can infer that basic power supplies in these two provinces still did work in most of the
areas although the battles were intense.
It is valuable to test whether the night-time light variation is correlated with statistics
summarizing the humanitarian scale of this crisis. For example, the number of displaced
persons is an important indicator of the magnitude of disasters. Displaced persons include
IDPs and refugees. These are the main summary statistics available with regard to the
conditions in Syria as it is difficult to obtain the number of Syrian refugees from each
province. Even for the IDP data of each province, the only accessible data are from
December 2013 provided by the Syria Needs Analysis Project (SNAP 2014). Thus, we
calculated the difference of SNL of each Syrian province between March 2011 and
December 2013. We should note that the IDP data for Rif Dimashq and Damascus
were combined, so that the SNL of these two regions were also combined for analysis.
The number of IDPs and SNL loss of all the Syrian regions between March 2011 and
December 2013 is presented in Table 2. Based on these two data groups and linear
–1.00
SNL change in proportion
LA change in proportion
(a)(b)
Idlib
AI-Hasakah
AI-Raqqah
AI-Suwayda
Quneitra
Latakia
Aleppo
Hama
Homs
Daraa
Deir ez-Zor
Rif Dimashq
Tartus
Damasous
Syria
Idlib
AI-Hasakah
AI-Raqqah
AI-Suwayda
Quneitra
Latakia
Aleppo
Hama
Homs
Daraa
Deir ez-Zor
Rif Dimashq
Tartus
Damasous
Syria
0.90
0.80
0.70
0.60
0.50
0.40
0.30
0.20
0.00
0.10
–1.00
–0.90
–0.80
–0.70
–0.60
–0.50
–0.40
–0.30
–0.20
–0.00
Region Region
–0.10
Figure 5. The night-time light change in proportions for Syria and all the provinces between
March 2011 and February 2014: (a) SNL change in proportion; and (b) lit area (LA) change in
proportion.
6654 X. Li and D. Li
Downloaded by [Wuhan University] at 18:13 17 October 2014

regression analysis (see Figure 6), we found R
2
= 0.52, showing that night-time light
decline is correlated to the number of displaced persons during the Syrian Crisis. This
finding supports the assumption that multi-temporal night-time light brightness is a proxy
for population dynamics (Bharti et al. 2011). In brief, night-time light variation can reflect
the humanitarian disasters in the Syrian Crisis.
4. Spatial analysis
We have analysed the night-time light in different administrative regions of Syria, but the
night-time light was aggregated in each region and the spatial details were invisible from
the analysis. To show the night-time light trend in a continuous spatial dimension, we
develop a data clustering method for time series night-time light images. The method is
50,0000
0
200,000
Number of IDPs
SNL loss
y = 10.035x + 1836.7
R2 = 0.52
400,000
600,000
800,000
1,000,000
1,200,000
1,400,000
1,600,000
1,800,000
2,000,000
100,000 150,000
Figure 6. Scatter diagram showing the relationship between SNL loss and number of IDPs of all
the provincial regions of Syria between March 2011 and December 2013.
Table 2. The number of IDPs and SNL loss of all the Syrian regions
between March 2011 and December 2013.
Province Number of IDPs SNL loss
Idlib 569,000 33,198
Al-Hasakah 230,000 63,787
Al-Raqqah 251,000 40,176
Al-Suwayda 52,000 13,075
Quneitra 78,000 19,912
Latakia 222,000 29,991
Aleppo 1,735,000 92,041
Hama 423,000 39,722
Homs 588,000 56,131
Daraa 372,000 32,075
Deirez-Zor 420,000 79,306
Rif Dimashq and Damascus 1,080,000 101,067
Tartus 500,000 126,199
International Journal of Remote Sensing 6655
Downloaded by [Wuhan University] at 18:13 17 October 2014

based on the basic hypothesis that similar trends of night-time light should be grouped
into one class, while the trend is irrelevant to the overall magnitude of the night-time
lights. The normalization was carried out using the following form:
l0
i¼li
P
n
j¼1
li=n
;(1)
where liis the night-time light value in a pixel for the ith month, l0
iis the normalized lifor
the same pixel, and nis the number of months. Therefore, the method has three steps:
(1) Extracting lit pixels. Since we only focus on the lit area at night, the dark areas are
excluded at first. For each pixel, if its value is smaller than the threshold of every
month, then it is labelled as a dark pixel else as a lit pixel. The threshold is set to
three in this analysis.
(2) Normalizing time series night-time light. Each pixel in the lit area is normalized
using Equation (1).
(3) Clustering time series night-time light images. We use a K-means algorithm to
cluster the normalized night-time light data into nclasses in the lit areas, and the
dark areas are labelled as the ‘dark region’class.
In this research, the number of classes in the clustering is set to two and three, respec-
tively, thus there are two groups of outcomes. Figure 7 shows the results including the
class maps and clustering centres. The clustering centres illustrate the different types of
night-time light trend in a discrete way, and the maps provide the spatial distribution of
the trends. For the two-class clustering, the green class represents the stable trend, and the
red class represents the declining trend. For the three-class clustering, the green class
represents the stable trend, and the red class represents the moderately declining trend, and
the blue class represents sharply declining trend. It is worth noting that the same colour
has different physical meanings for the two maps. Specifically, for the red class, it
represents the declining trend in the two-class clustering, while it represents the moder-
ately declining trend in the three-class clustering.
From Figure 7, most of the lit areas in Syria have experienced decline of night-time
light, which is very different from the neighbouring lands. Proportions of different classes
were statistically analysed for both Syria and its neighbouring lands, with results listed in
Table 3. In this and the following analysis, we use Syria’s international border, except for
the Golan Heights, which are occupied by Israel and therefore where we use the line of
actual control between Israel and Syria. For the two-class map, 78% of Syria’s lit areas are
covered by the red class, which represents declining lights, and 96.2% of Syria’s neigh-
bouring lit land is covered by the green class, which represents stable lights. For the three-
class map, the situation is similar, as 47.4% and 44.6% of Syria’s lit land has experienced
moderate and sharp declines of night-time light, respectively, whereas these two indicators
for the neighbouring lit land are only 11.4% and 1.0%, respectively. These results accord
with the fact that the international border of Syria distinguished the declining trend on the
Syrian side and the stable trend on the neighbouring countries by visual observation.
The above analysis demonstrates that the international border of Syria is also a border
for the pattern of night-time light variations. This finding supports previous study that the
night-time light and its growth have geographical discontinuity across international
6656 X. Li and D. Li
Downloaded by [Wuhan University] at 18:13 17 October 2014

borders (Pinkovskiy 2013). To further investigate this phenomenon, we made a local
analysis around the international border of Syria based on the clustering results.
Let fc1;...;cngdenote the nclasses of night-time light variation pattern,
frðxÞ
1;...;rðxÞ
ngdenote the proportions of these patterns in lit land of region x, and
0.0
Red class
(b)
Green class
Red class
Blue class
Green class
October 2011
December 2011
February 2012
September 2012
November 2012
October 2013
December 2013
March 2011
February 2014
Month
(d)
(c)
(a)
Scale N
N
km
100
Scale
km
100
Month
Normalized night-time light
Normalized night-time light
March 2013
January 2013
October 2011
December 2011
February 2012
September 2012
November 2012
October 2013
December 2013
March 2011
February 2014
March 2013
January 2013
0.5
1.0
1.5
2.0
2.5
0.0
0.5
1.0
1.5
2.0
2.5
3.0
Figure 7. Class maps and class centres derived from the normalized time series night-time light
images: (a) the two-class map; (b) class centres for the two-class map; (c) the three-class map; and
(d) class centres for the three-class map.
Table 3. Proportions of different classes for lit lands of Syria and the neighbouring regions.
Two-class clustering Three-class clustering
Syrian lit land
Neighbouring lit
land Syrian lit land Neighbouring lit land
Stable Decline Stable Decline Stable
Moderate
decline
Sharp
decline Stable
Moderate
decline
Sharp
decline
22.0% 78.0% 96.2% 3.8% 8.0% 47.4% 44.6% 87.6% 11.4% 1.0%
International Journal of Remote Sensing 6657
Downloaded by [Wuhan University] at 18:13 17 October 2014

frðyÞ
1;...;rðyÞ
ngdenote the proportions of these patterns in lit land of region y. Assuming
that xand yare adjacent regions which are divided by a border, then a border effect index
(BEI) is constructed as:
BEI ¼X
n
i¼1
jrðxÞ
irðyÞ
ij:(2)
If BEI is equal to 0, the border has no effect on the night-time light variation pattern,
because both regions have the same proportions of night-time light variation patterns. At
the other extreme, the border has the strongest effect on the night-time light variation
when BEI is equal to 2. Generally, a larger BEI value represents a border with a stronger
contrast in night-time light variation.
We generate a spatial buffer with a distance xfrom Syria’s international border, and
the buffer zone is divided by the border into two zones, the internal buffer zone and the
external buffer zone (Syria’s neighbouring land). For the external buffer zone, we only
retain the region on the continent by discarding the sea. The buffer zones are illustrated in
Figure 8. Consequently, based on the two zones and the clustering maps, we use Equation
(2) to calculate the border effect. In the analysis, the buffer distance xis ranged from 5 to
50 km with 5 km as the interval. Thus, for every distance x, the BEI is derived, as shown
in Figure 9.
Based on the BEI curves from the two clustering maps, it is interesting to find that the
BEI increases with the buffer zone width. Take the BEI from the two-class map as an
example, when the buffer width is 5 km, the BEI is only 0.5615, but the BEI increases to
0.9940 when the width is 10 km and continues to increase with the buffer width. The BEI
Figure 8. Buffer zones to calculate the border effect index (BEI). xdenotes the buffer zone width.
The external buffer zone is modified by discarding the regions outside the coastline. For the border
between Israel and Syria, we use the line of actual control between Israel and Syria instead of their
international border.
6658 X. Li and D. Li
Downloaded by [Wuhan University] at 18:13 17 October 2014

reaches 1.5181 when the buffer width is 50 km. The three-class map produced the same
trend as Figure 9 shows. Therefore, the international border is an obvious boundary for
differentiating the night-time light variation during the Syrian Crisis, and the differentia-
tion capacity increases with the distance to the border.
5. Conclusion
This study provides a primary analysis on the response of night-time light to the Syrian
Crisis. For the country and all provinces, the night-time light experienced a sharp decline
as the crisis broke out. We found that most of the provinces lost >60% of the night-time
lights and the lit areas because of the war, and the amount of the night-time light loss is
correlated to the number of IDPs. We also find that the international border of Syria is a
boundary to the night-time light variation patterns, reproving that the administrative
border has the effect of socioeconomic discontinuity.
As this research only provides a primary evaluation of the night-time light data for the
Syrian crisis, more information can be discovered by the use of night-time light images in
future studies. For example, night-time light variations in control zones of different
groups, including the Assad regime, Free Syrian Army, Kurds, and the Islamic State of
Iraq and al Shams, can be investigated to evaluate humanitarian situations in these
regions. Additionally, by the use of night-time light images, we can also study how the
Syrian Civil War has spread to Iraq, where the Islamic State of Iraq and al Shams is now
the global focus.
The limit of DMSP/OLS images is that they only capture efficient data for some
months in high-latitude areas. And the new emerging Visible Infrared Imaging
Radiometer Suite (VIIRS) sensor on the Suomi National Polar-orbiting Partnership
(NPP) can acquire efficient data in more months. In addition, NPP/VIIRS night-time
light images have higher spatial resolution and wider radiometric range (Elvidge et al.
2013). Nevertheless, the NPP was launched in the end of 2011, so it did not record the
pre-crisis images, and that is why we did not employ NPP/VIIRS images in this study.
However, considering its superiority to the DMSP/OLS images, it will be employed for
humanitarian disaster and human rights evaluation in future studies.
5
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
10 15
Width of the buffer zone (km)
Border effect index
Border effect index
20 25 30
(a)
35 40 45 50 510 15
Width of the buffer zone (km)
20 25 30
(b)
35 40 45 50
Figure 9. BEI with different widths of buffer zones for the night-time light variation across the
international border of Syria: (a) BEI derived from the two-class clustering map; and (b) BEI derived
from the three-class clustering map.
International Journal of Remote Sensing 6659
Downloaded by [Wuhan University] at 18:13 17 October 2014

Acknowledgements
The DMSP/OLS images were provided by National Geophysical Data Center of the USA, and the
authors appreciate the help provided by Christopher Elvidge and Michael Von Hendy from the
National Geophysical Data Center. The authors would like to thank Timothy Warner and Arthur
Cracknell, the Editor-in-Chief and Co-Editor-in-Chief of IJRS, for their helpful comments.
Funding
This research was supported by the National Natural Science Foundation of China [grant number
41101413], Doctoral Fund of Ministry of Education of China [grant number 20110141120073], and
China Scholarship Council.
References
Agnew, J., T. W. Gillespie, J. Gonzalez, and B. Min. 2008. “Baghdad Nights: Evaluating the
US Military ‘Surge’Using Nighttime Light Signatures.”Environment and Planning A 40:
2285–2295. doi:10.1068/a41200.
Al-Hazzaa, H. A. 2014. Opposition and Regime Forces Split Idlib Province after ISIS Withdrawal.
http://www.damascusbureau.org/?p=6622.
AAAS (American Association for the Advancement of Science). 2013. Conflict in Aleppo, Syria: A
Retrospective Analysis.http://www.aaas.org/aleppo_retrospective.
Barnard, A. 2013. Syrian Forces Recapture Damascus Suburb from Rebels.http://www.nytimes.
com/2013/11/14/world/middleeast/syrian-forces-recapture-damascus-suburb-from-rebels.html?_
r=0.
Bharti, N., A. J. Tatem, M. J. Ferrari, R. F. Grais, A. Djibo, and B. T. Grenfell. 2011. “Explaining
Seasonal Fluctuations of Measles in Niger Using Nighttime Lights Imagery.”Science 334:
1424–1427. doi:10.1126/science.1210554.
Blomfield, A. 2012. Syria: Assad Regime Retakes Control of Damascus Suburbs.http://www.
telegraph.co.uk/news/worldnews/middleeast/syria/9418993/Syria-Assad-regime-retakes-control-
of-Damascus-suburbs.html.
Chen, X., and W. D. Nordhaus. 2011. “Using Luminosity Data as a Proxy for Economic Statistics.”
Proceedings of the National Academy of Sciences of the United States of America 108: 8589–
8594. doi:10.1073/pnas.1017031108.
Cumming-Bruce, N. 2014. Death Toll in Syria Estimated at 191,000.http://www.nytimes.com/2014/
08/23/world/middleeast/un-raises-estimate-of-dead-in-syrian-conflict-to-191000.html?_r=0.
Elvidge, C. D., K. E. Baugh, E. A. Kihn, H. W. Kroehl, E. R. Davis, and C. W. Davis. 1997.
“Relation between Satellite Observed Visible-Near Infrared Emissions, Population, Economic
Activity and Electric Power Consumption.”International Journal of Remote Sensing 18: 1373–
1379. doi:10.1080/014311697218485.
Elvidge, C. D., K. E. Baugh, M. Zhizhin, and F.-C. Hsu. 2013. “Why VIIRS Data are Superior to
DMSP for Mapping Nighttime Lights.”Proceedings of the Asia-Pacific Advanced Network 35:
62–69. doi:10.7125/APAN.35.7.
Elvidge, C. D., D. Ziskin, K. E. Baugh, B. T. Tuttle, T. Ghosh, D. W. Pack, E. H. Erwin, and M.
Zhizhin. 2009. “A Fifteen Year Record of Global Natural Gas Flaring Derived from Satellite
Data.”Energies 2: 595–622. doi:10.3390/en20300595.
Henderson, V., A. Storeygard, and D. N. Weil. 2011. “A Bright Idea for Measuring Economic
Growth.”American Economic Review 101: 194–199. doi:10.1257/aer.101.3.194.
Li, X., F. Chen, and X. Chen. 2013. “Satellite-Observed Nighttime Light Variation as Evidence for
Global Armed Conflicts.”IEEE Journal of Selected Topics in Applied Earth Observations and
Remote Sensing 6: 2302–2315. doi:10.1109/JSTARS.2013.2241021.
Li, X., X. Chen, Y. Zhao, J. Xu, F. Chen, and H. Li. 2013. “Automatic Intercalibration of Night-
Time Light Imagery Using Robust Regression.”Remote Sensing Letters 4: 45–54. doi:10.1080/
2150704X.2012.687471.
Li, X., L. Ge, and X. Chen. 2013. “Detecting Zimbabwe’s Decadal Economic Decline Using
Nighttime Light Imagery.”Remote Sensing 5: 4551–4570. doi:10.3390/rs5094551.
Pinkovskiy, M. 2013. Economic Discontinuities at Borders: Evidence from Satellite Data on Lights
at Night.http://economics.mit.edu/files/7271.
6660 X. Li and D. Li
Downloaded by [Wuhan University] at 18:13 17 October 2014

Small, C. 2012. “Spatiotemporal Dimensionality and Time-Space Characterization of Multitemporal
Imagery.”Remote Sensing of Environment 124: 793–809. doi:10.1016/j.rse.2012.05.031.
Small, C., and C. D. Elvidge. 2013. “Night on Earth: Mapping Decadal Changes of Anthropogenic
Night Light in Asia.”International Journal of Applied Earth Observation and Geoinformation
22: 40–52. doi:10.1016/j.jag.2012.02.009.
Sulik, J. J., and S. Edwards. 2010. “Feature Extraction for Darfur: Geospatial Applications in
the Documentation of Human Rights Abuses.”International Journal of Remote Sensing 31:
2521–2533. doi:10.1080/01431161003698369.
SNAP (Syria Needs Analysis Project). 2014. Regional Analysis for Syria: Brief, May 2014.http://
reliefweb.int/sites/reliefweb.int/files/resources/brief_may_2014.pdf.
Witmer, F. D. W., and J. O’Loughlin. 2011. “Detecting the Effects of Wars in the Caucasus Regions
of Russia and Georgia Using Radiometrically Normalized DMSP-OLS Nighttime Lights
Imagery.”Giscience & Remote Sensing 48: 478–500. doi:10.2747/1548-1603.48.4.478.
Wu, J., S. He, J. Peng, W. Li, and X. Zhong. 2013. “Intercalibration of DMSP-OLS Night-Time
Light Data by the Invariant Region Method.”International Journal of Remote Sensing 34:
7356–7368. doi:10.1080/01431161.2013.820365.
Zhang, Q. L., and K. C. Seto. 2011. “Mapping Urbanization Dynamics at Regional and Global
Scales Using Multi-Temporal DMSP/OLS Nighttime Light Data.”Remote Sensing of
Environment 115: 2320–2329. doi:10.1016/j.rse.2011.04.032.
International Journal of Remote Sensing 6661
Downloaded by [Wuhan University] at 18:13 17 October 2014