Sustainable Cities and Society 57 (2020) 102115

Contents lists available at ScienceDirect

Sustainable Cities and Society

journal homepage: www.elsevier.com/locate/scs

Observing community resilience from space: Using nighttime lights to model T

economic disturbance and recovery pattern in natural disaster
Yi Qianga,*, Qingxu Huangb, Jinwen Xua
a
Department of Geography and Environment, University of Hawaii – Manoa, USA
b
Center for Human-Environment System Sustainability (CHESS), State Key Laboratory of Earth Surface Processes and Resource Ecology (ESPRE), Beijing Normal
University, China

A R T I C LE I N FO A B S T R A C T

Keywords: A major challenge for measuring community resilience is the lack of empirical observations in disasters. As an
Natural disaster effective tool to observe human activities on the earth surface, night-time light (NTL) remote sensing images can
Resilience fill the gap of empirical data for measuring community resilience in natural disasters. This study introduces a
Economic recovery quantitative framework to model recovery patterns of economic activity in a natural disaster using the Defense
Nighttime lights
Meteorological Satellite Program-Operational Linescan System (DMSP-OLS) images. The utility of the framework
Empirical assessment
DMSP/OLS
is demonstrated in a retrospective study of Hurricane Katrina, which uncovered the great economic impact of
Katrina and spatial variation of the disturbance and recovery pattern of economic activity. Environmental and
socio-economic factors that potentially influence economic recovery were explored in statistical analyses.
Instead of a static and holistic index, the framework measures resilience as a dynamic process. The analysis
results provide actionable information for prompting resilience in diverse communities and in different phases of
a disaster. In addition to Hurricane Katrina, the resilience modeling framework is applicable for other disaster
types. The introduced approaches and findings increase our understanding about the complexity of community
resilience and provide support for developing resilient and sustainable communities.

1. Introduction Fussell, Sastry, & VanLandingham, 2010), health (Burton, 2006; Sastry
& VanLandingham, 2009), and psychological conditions (Adeola,
Due to climate change and rapid population growth, human society 2009). These observed disparities can be attributed to various resilience
is faced with increasing threats from natural disasters that can cause of the communities.
significant socio-economic consequences. Coastal communities around Resilience describes the ability of an individual or a system to adapt
the world are particularly vulnerable to natural disasters including both to and recover from external shocks or stresses (Adger, 2000). Although
large-scale rapid-moving disturbances such as hurricane and storm substantial knowledge has been gained on ecological resilience (Perz,
surges (Tebaldi, Strauss, & Zervas, 2012), and the slow-moving pro- Muñoz-Carpena, Kiker, & Holt, 2013) and engineering resilience (Yodo
cesses such as coastal erosion, sea level rise (Nicholls, Hoozemans, & & Wang, 2016), there is yet a consensus on how to measure resilience of
Marchand, 1999) and reduction of ecosystem services (Spalding et al., human communities due to their complexity. In general, quantitative
2014). According to the data from U.S. Census Bureau (2011), 39 % of assessment of community resilience is challenged by two issues. First,
the total population in the United States are living in counties directly the definition of community resilience various in different domains,
on the shorelines and the population density in coastal counties is more which will be discussed in Section 2.1. Moreover, resilience is often
than four times the average density of the whole United States. Since used interchangeably with other relevant concepts such as vulnerability
2005 when Hurricane Katrina and Rita caused catastrophic damage in and adaptive capacity. The various definitions and conceptual frame-
Central Gulf Coast, much attention has been paid to the resilience and works of community resilience influence how researchers measure re-
long-term sustainability of coastal communities. Empirical observations silience (Cutter et al., 2008; Lam, Reams, Li, Li, & Mata, 2016; Sherrieb,
suggest that, under the same strength of disasters, different commu- Norris, & Galea, 2010). The definition disagreement hampers the de-
nities endured different levels of disturbance and presented different velopment of standard metrics to measure resilience. Second, there is
recovery patterns in socio-economic (Finch, Emrich, & Cutter, 2010; lack of empirical data and approaches to quantify community

⁎
Corresponding author.
E-mail addresses: [email protected] (Y. Qiang), [email protected] (Q. Huang), [email protected] (J. Xu).

https://doi.org/10.1016/j.scs.2020.102115
Received 5 September 2019; Received in revised form 9 January 2020; Accepted 24 February 2020
Available online 26 February 2020
2210-6707/ © 2020 Elsevier Ltd. All rights reserved.
Y. Qiang, et al. Sustainable Cities and Society 57 (2020) 102115

resilience. Most of the existing assessments are based on an index ap- Hooks, 2015). Extensive reviews about the definitions of resilience and
proach, which integrates a set of presumed indicators into a composite the related terms can be found in (Cutter et al., 2008; Lam et al., 2016;
score to measure resilience (Cutter, Burton, & Emrich, 2010; Hung, Liao, 2012; Peacock, 2010). Despite the various definitions in the lit-
Yang, Chien, & Liu, 2016; Sempier, Swann, Emmer, Sempier, & erature, community resilience is often associated with two abilities: 1)
Schneider, 2010; Sherrieb et al., 2010). The model specification, in- the ability to absorb/resist/withstand disturbance, and 2) the ability to
dicator selection and weighting are based on prior knowledge or expert respond/recover/restore the acceptable level of functioning and struc-
opinions. Although these indices provide general guidance for pre- ture.
dicting community resilience, their accuracies have not been validated In addition to the qualitative descriptions, a number of theoretical
against empirical observations in disasters (Bakkensen, Fox‐Lent, Read, frameworks have been developed to quantify community resilience. For
& Linkov, 2017; Beccari, 2016). instance, Cutter et al. (2010), Cutter, Ash, & Emrich (2014) define that
Empirical data about human activities and states are difficult to resilience consists of six components including social, economic, in-
obtain in a disaster condition when many social systems fail to function. frastructural, institutional, community, and environmental, which is
Traditional data sources for resilience assessment (e.g. surveys and used as a guidance to select indicators for resilience indices. Lam et al.
census data) have limitations in various aspects, which will be elabo- (2016) measure resilience from the relationships among exposure, da-
rated in the next section. Recently, remote sensing imageries become mage and recovery. Additionally, resilience can be conceptualized as a
popular instruments to monitor human dynamics on the earth surface dynamic process such as the recovery trajectory (also known as re-
such as urban growth (Shahtahmassebi et al., 2016), land cover change covery curve), which describes the continuous change of a functional
(Joshi et al., 2016), and socio-economic conditions (Kuffer, Pfeffer, & capacity of a system affected by a disturbance (e.g. natural disaster).
Sliuzas, 2016). Among the various remote sensing products, night-time The functional capacity could be the social and economic capacity of a
light (NTL) remote sensing has unique ability to capture fluctuations of human community (White, Edwards, Farrar, & Plodinec, 2015), bio-
human activities, which can provide empirical data for resilience as- mass or population of an ecological community (Qiang & Xu, 2019;
sessment. This study introduces a quantitative framework to assess Vercelloni, Kayal, Chancerelle, & Planes, 2019), or the functionality or
community resilience using the DMSP-OLS NTL annual composite serviceability of an infrastructure system (Koliou et al., 2018). The
images as the data source. Specifically, stable lights in the time series of Curve A, B and C in Fig. 1 illustrates common scenarios of resilience
DMSP-OLS annual images are used as a proxy to model recovery pat- from high to low, where the functional capacity suddenly declines after
terns of economic activity after Hurricane Katrina in 2005. Spatial and a disturbance, gradually recovers afterwards, and finally restores to the
statistical analyses are conducted to explore the geographical dis- pre-disaster condition (e.g. Trajectory B) or a new equilibrium (Tra-
parities of the recovery patterns and their relationships with the se- jectory A and C). Given the same strength of disaster, the variation of
lected resilience indicators. Specific questions answered in the case the recovery trajectory is indicative of community resilience. The
study are: 1) which communities appeared to be more or less resilient in maximum deviation from the pre-disaster condition (i.e. maximum
the disaster; 2) how the observed resilience levels are associated with disturbance) reflects the ability of a system to absorb/resist/withstand
the environmental and socio-economic conditions? The introduced disturbance from the disaster. The recovery speed or time indicates the
framework aims to fill the critical gap of empirical data and assessment ability to respond/recover/restore the functional capacity.
methods for community resilience. The analysis results from the case
study increase our understanding about community resilience and
2.2. Resilience assessment – index approaches
provide actionable information to predict and prompt community re-
silience.
Previous work of resilience assessment is mostly based on an index
The rest of the article is organized as follows. Section 2 briefly re-
approach, which aggregates a number of socio-economic and environ-
views the related work about the definitions, conceptual frameworks
mental indicators into an overall resilience index. The Baseline
and assessment methods of community resilience. Section 3 introduces
Resilience Index for Communities (BRIC) developed by Cutter et al.
the data sources, assessment framework of community resilience based
(2010) is one of the first and most cited resilience index. Analogous to
on NTL data and statistical analyses. Section 4 presents the analysis
the previous work of the Social Vulnerability Index (SoVI) (Cutter,
results in the case study of Hurricane Katrina, followed by the discus-
Boruff, & Shirley, 2003), BRIC is aggregated from 36 indicators of the
sions in Section 5 and conclusions in Section 6.
baseline characteristics of community resilience. The indicators are
rescaled into [0,1] and then aggregated in five categories including
2. Related work
social, economic, institutional, infrastructure, and community. The
overall BRIC index is the summation of the aggregated indices in the
2.1. Definition and conceptual framework

The concept of resilience was first introduced by Holling (1973),

who views resilience as the ability of an ecological system to absorb
change in the face of extreme perturbation and yet continue to persist.
Later, Timmerman (1981) applied the concept of resilience to social
systems and defined resilience as the measure of a system’s capacity to
absorb and recover from disastrous events. The resilience of a social
system is also known as community resilience. Extending Timmerman’s
definition, Cutter et al. (2008) further elaborated that community re-
silience includes both the inherent conditions of a system to absorb
impacts and cope with an event and post-event, adaptive processes that
facilitate the ability of the social system to re-organize, change, and
learn in response to a threat. Norris, Stevens, Pfefferbaum, Wyche, and
Pfefferbaum (2008) considered resilience as a process linking commu-
nities’ capacities in response to the disturbance. In the field of en-
gineering and infrastructure systems, resilience describes the ability of Fig. 1. Recovery trajectory of a system due to an external disturbance. Curve A,
resisting and absorbing disturbances and the ability of adapting to B and C represent three possible scenarios of resilience from high to low.
disruptions, and returning to normal functionalities (Faturechi & Miller- (Modified from White et al., 2015)

2
Y. Qiang, et al. Sustainable Cities and Society 57 (2020) 102115

five categories. Analogously, the Community Disaster Resilience Index comprehensive review of applications of NTL data can be found in
(CDRI) by (Peacock, 2010) categorizes resilience indicators into a 4 × 4 (Huang, Yang, Gao, Yang, & Zhao, 2014). For disaster management, the
matrix with capital domains (i.e. the social, economic, physical, and NTL data have been applied to identify damage (Gillespie, Frankenberg,
human capital) and disaster phases (i.e. the mitigation, preparedness, Chum, & Thomas, 2014; Kohiyama et al., 2004), power outage
response and recovery phase). The selected indicators are aggregated in (Hultquist, Simpson, Cervone, & Huang, 2015; Zhao et al., 2018), and
16 categories in the matrix, which are then aggregated into the CDRI analyze the change of human activities (Li, Zhan, Tao, & Li, 2018) and
index. Other work on resilience indices include (Foster, 2012; Hung urbanization (Huang, Wang, & Lu, 2019) affected by disasters. Despite
et al., 2016; Sherrieb et al., 2010). Although these indices provide these applications, the utility of NTL data in modeling community re-
general predictions of resilience by integrating prior knowledge and silience has not been fully exploited in a theoretical framework. Ex-
expert opinions, they do not inform specific disaster outcomes. For tending the conceptual framework of recovery trajectory, this study
instance, it is unclear whether a high resilience index implies low introduces a quantitative approach to model resilience using DMSP/
property loss, low casualty and injury, or fast economic recovery. The OLS NTL images as the data source.
unspecified outcomes diminish the value of the indices for specific
decision-making. Moreover, most of the resilience indices have not 3. Method
been calibrated or validated against empirical observations. Bakkensen
et al. (2017) validated the BRIC and CDRI with observed losses, fatal- 3.1. Inter-calibration of NTL images
ities, and disaster declarations, and found low or even contradictory
correlations between the indices and disaster outcomes. The Stable Lights images collected by the Defense Meteorological
Satellite Program Operational Line Scanner (DMSP/OLS) were used for
2.3. Resilience assessment – empirical approaches this study. The images are cloud-free composites created using all the
available archived DMSP-OLS smooth resolution data from the year
In addition to the index approach, efforts have been made to assess 1992 to 2013. The DMSP/OLS images include 34 annual composites at
community resilience using empirical observations in disasters. For a 30 arc second resolution collected by six different satellites (F10, F12,
instance, Lam, Pace, Campanella, LeSage, and Arenas (2009) and F14, F15, F16, and F18). Due to the absence of inter-satellite calibration
LeSage, Pace, Lam, Campanella, and Liu (2011) conducted a series of and onboard calibration, the digital numbers (DN) in the DMSP-OLS
telephone and street surveys to continuously monitor business re- images cannot be converted to exact radiance. To analyze continuous
opening in New Orleans after Hurricane Katrina. The time series of recovery trajectories over time, the DMSP-OLS images need to be ca-
open businesses resemble the recovery trajectory illustrated in Fig. 1, librated to make the images in different years and satellites comparable.
which declines sharply after Katrina and gradually recovers afterwards. A widely applied inter-calibration procedure was proposed by (Elvidge
The various recovery patterns in different communities were associated et al., 2009), which uses a quadratic polynomial regression to adjust the
with environmental and socio-economic variables to explain why some DNs against a reference image (see Eq. (1)).
communities restored businesses more quickly than others. However,
DNcalibrated = a1 + a2 ∙DN + a3 ∙DN 2 (1)
the surveys were costly, labor-intensive and time consuming, which is
not widely applicable. Later, (Lam, Qiang, Arenas, Brito, & Liu, 2015; The inter-calibration process follows the same procedure as in-
Lam, Reams, Li, Li, & Mata, 2016) developed the Resilience Inference troduced in (Elvidge et al., 2009). By reviewing the data, it was found
Model (RIM) which uses the number of disasters, damage and popula- that the image F121999 has the highest average DN in the United
tion growth as proxies to measure community resilience. Based on ag- States. Due to the saturation of the DMSP/OLS in bright areas (e.g. city
gregated data in a 10-year period, the RIM model considers that a re- centers), the F121999 was used as the reference image and all other
silient community can resist damage and maintain high population images were calibrated to match the DNs in F121999. Los Angeles was
growth while endured a high number of disasters. Recently, with the chosen as the reference site, as it has been a mature metropolis where
advent of the Big Data era, data crowdsourcing and social media plat- the light change is negligible (Hsu, Baugh, Ghosh, Zhizhin, & Elvidge,
forms provide new opportunities to observe individuals’ activities and 2015). Due to the uneven distribution of DNs in the images, a random
narratives at finer spatial and temporal resolutions. For instance, Zou, sampling will lead to overfit near the two extremes of DNs (i.e. 1 and
Lam, Cai, & Qiang (2018) used the frequency and sentiment of geo- 63) where pixels are concentrated. To ensure the regression equations
tagged Twitter messages (tweets) to monitor dynamic conditions of evenly fit the entire value range, a stratified sample of lit pixels (200
communities in Hurricane Sandy. The recovery trajectories of com- pixels in each DN value) were extracted in the reference site for the
munities can be reflected from time series of ratios and average senti- calibration. The 2nd order regression equation was calibrated for each
ment of tweets during the disaster. Timeliness, low cost and scalability image with the reference to F121999. The coefficients of the regression
are the main advantages of social media data. However, the Big Data equations are in Table 1. After the calibration, images in the same year
approaches are criticized for the biased user profile (Zou, Lam, Shams were averaged into one image, leading to a time series of annual images
et al., 2018) and low data quality (much noise and misinformation) (Li from 1992 to 2013.
et al., 2016).
3.2. Estimation of gross domestic product
2.4. Remote sensing for disaster management
The inter-calibrated DMSP-OLS images were clipped in the affected
Remotely sensed imageries have been widely applied in disaster risk area in Hurricane Katrina, which include 179 counties declared as
mapping (Bates, 2004; Hong, Adler, & Huffman, 2007) and damage disaster areas by the Federal Emergency Management Agency (FEMA).
assessment (Cooner, Shao, & Campbell, 2016; Dong & Shan, 2013; Hurricane Katrina made the first landfall in Florida on August 25th,
Vetrivel, Gerke, Kerle, Nex, & Vosselman, 2018). However, most remote 2005 and the second in Louisiana on August 29th, 2005. The disaster
sensing products are not very useful for resilience assessment due to areas include the entire Louisiana (64 parishes) and Mississippi (82
their insensitivity to decline of human activity. For instance, it is dif- counties), 22 counties in Alabama, and 11 counties in Florida (Fig. 2).
ficult to detect a ‘ghost town’ or a damaged city from Landsat images. Zonal operation was applied to aggregate the DNs in the counties. After
As an alternative, nighttime light (NTL) remote sensing images have logarithm transformation, the sum of DNs and number of lit pixels can
been proved an effective means to observe the dynamics (both increase well predict (R2= 0.92) the gross domestic product (GDP) in the 179
and decline) of population (Zhuo et al., 2009), urbanization (Xie & counties in a linear model (Eq. (2)). The goodness of fit (i.e. R2 ) is
Weng, 2017) and economic activities (Li, Ge, & Chen, 2013). A highest for GDP compared to other economic indicators (e.g. personal

3
Y. Qiang, et al. Sustainable Cities and Society 57 (2020) 102115

Table 1 ( fpost (2005)) was calculated to represent instant economic disturbance

Coefficients of the quadratic polynomial regression for inter-calibration. caused by Katrina (Eq. (4)). This metric (denoted as D for simplicity)
Satellite Year measures the ability of a system to absorb/resist/withstand disturbance
a1 a2 a3 R2
in a disaster. Second, the difference between the slope of the post-dis-
F10 1992 0.99490 1.19694 −0.00419 0.84241 aster trajectory (r2 ) and the slope of business-as-usual trajectory (r1) was
F10 1993 −0.53668 1.51570 −0.00845 0.90954 calculated to indicate the recovering rate (Rr ) of GDP after Katrina (Eq.
F10 1994 1.09383 1.46457 −0.00852 0.87570
(5)). A high (positive) Rr would implies a strong ability to respond/
F12 1994 0.68785 1.22501 −0.00428 0.88668
F12 1995 1.29538 1.19446 −0.00428 0.88056 recover/restore the GDP growth to catch up with the business-as-usual
F12 1996 0.52894 1.33022 −0.00633 0.87521 trajectory. Third, the accumulated difference between the post-disaster
F12 1997 1.47600 1.18090 −0.00472 0.83470 GDP and the business-as-usual GDP from 2005 to 2013 was calculated
F12 1998 0.92953 1.08940 −0.00259 0.92079 using Eq. (6). This metrics measures the accumulated economic loss (L)
F12 1999 1 1 1 1
F14 1997 2.04904 1.55982 −0.01079 0.85754
due to the deviation of GDP growth from the business-as-usual trajec-
F14 1998 0.94451 1.55107 −0.00977 0.95204 tory, which is illustrated as the grey area in Fig. 3. L represents the
F14 1999 0.80845 1.52425 −0.00911 0.95928 combined effect of D and RR. A high L can be interpreted as low resi-
F14 2000 2.42372 1.29577 −0.00605 0.90692 lience, which typically consists of a high instant disturbance (D) and
F14 2001 2.51390 1.43262 −0.00829 0.91364
slow recovery (Rr). A low L means the opposite. L differentiates the
F14 2002 3.59977 1.34024 −0.00760 0.87948
F14 2003 3.05182 1.37950 −0.00788 0.90928 intermediate situations such as high D and high Rr or low D and low Rr.
F15 2000 0.55891 1.15751 −0.00315 0.94024 L is calculated from 2005 to 2013 after Katrina when the DMSP-OLS
F15 2001 0.57387 1.23965 −0.00433 0.93629 images are available.
F15 2002 1.30027 1.19630 −0.00416 0.93331
F15 2003 1.86389 1.64457 −0.01154 0.92080 D = fpre (2005) − fpost (2005)
F15 2004 2.74263 1.56383 −0.01043 0.90945
(4)
F15 2005 2.51648 1.55181 −0.01034 0.88839
F15 2006 2.94339 1.38264 −0.00811 0.86845 Rr = r2 − r1 (5)
F15 2007 3.99771 1.41246 −0.00897 0.82033
F16 2004 2.32112 1.37760 −0.00795 0.87267
F16 2005 2.90092 1.56247 −0.01079 0.87283 2013
F16
F16
2006
2007
2.64689
2.83160
1.25093
1.15147
−0.00583
−0.00477
0.85792
0.84669
L= ∫ (fpre (t ) − fpost (t ))
2005 (6)
F16 2008 2.94479 1.17826 −0.00534 0.81480
F16 2009 2.36856 1.24705 −0.00609 0.84942
F18 2010 2.35041 0.94747 −0.00196 0.80366
F18 2011 1.63669 1.20682 −0.00557 0.85396
F18 2012 2.56954 1.07559 −0.00385 0.82972 3.4. Regression analysis
F18 2013 2.25727 1.07572 −0.00365 0.85132
Regression analyses were applied to examine the associations of the
three resilience metrics (D, RR and L) with environmental and socio-
income, number of employees and business establishments). As county- economic variables. The selected variables fall into four categories,
level GPD data in the U.S. is only available in 2012–2015 (Bureau of including impact intensity, environmental, socio-economic, and in-
Economic Analysis, U.S. Department of Commerce, 2018), the GDP data dustrial structure. Max wind gust speed and accumulated rainfall re-
and DMSP-OLS images in 2012 and 2013 (the overlapping years of the present the destructive power of a hurricane (NOAA, 2006). Elevation,
two data sources) were used to derive the regression equation. All the proximity to coast and ratio of urban in flood zones indicate the ex-
GDP data are adjusted to the value of current U.S. dollar. Finally, the posure to Hurricane-induced storm surge and flooding (Dasgupta,
GDP of the counties from 1992 to 2013 was estimated using Eq. (3). Laplante, Murray, & Wheeler, 2011; Jonkman, Maaskant, Boyd, &
log(GDP2003,2004 ) = 3.375∙log(DNsum) − 2.548∙log(litDN ) − 1.761 (2) Levitan, 2009). The socio-economic category includes 10 commonly
used indicators in community resilience measurements (Cai, Lam, Zou,
GDP1992 − 2013 = e (3.375 ∙ log(DNsum) − 2.548 ∙ log(litDN ) − 1.761) (3) & Qiang, 2018; Cutter et al., 2014). Ratios of establishments in different
scales and industrial sectors represent the industry structure, which
potentially influences economic recovery (Martin & Sunley, 2015).
3.3. Measurement framework Most of the variables were acquired from datasets released before 2005
to represent pre-disaster conditions. Variables that are not reported in
The estimated GDP was analyzed in the framework in Fig. 3 to as- counties were preprocessed and averaged into counties (Table 2). In-
sess the resilience of the counties. The GDP of each county is rescaled stead of composing a comprehensive index for resilience, the objective
into [0, 1] to be comparable. Two regression lines were derived for the of the analysis is examining the relationships between the GDP recovery
normalized GDP from 1992 to 2004 and from 2005 to 2013 to represent trajectories and the hypothetical resilience indicators.
the pre- and post-Katrina economic trajectory respectively. Although Two types of regression analysis were carried out in this study. First,
various models are available for GDP projection, the linear model is still univariate regression was applied to examine the relationships between
considered a valid benchmark to project GDP growth in the U.S. D, RR and L and each individual variable. Second, multivariate regres-
(Ferrara, Marcellino, & Mogliani, 2015; Marcellino, 2008), especially in sion was utilized to relate the three metrics with all the variables and
the recent decades when the volatility of US GDP growth tend to decline variables in the four categories. R2 of the multivariate regression in-
(Burren & Neusser, 2010). Despite the slight decline (-1.8 %) in 2009 dicates the proportion of variance explained by the different categories
due to the financial crisis, the annual GDP in the U.S. from 1990 to of variables. Adjusted R2 were compared to evaluate the prediction
2018 generally follows a linear trend (World Bank, 2019). The exten- power of the different categories for the recovery metrics. The regres-
sion of the pre-Katrina trajectory (thin dashed line in Fig. 3) represents sion analyses aim to explore the underlying factors that influence re-
the business-as-usual condition as if the pre-Katrina economic growth silience and proportion of resilience variance that can be explained by
persists. the variables. All variables were re-scaled into z-scores for the regres-
Three metrics were calculated for each county. First, the difference sion analyses. Thus the b coefficient of the regression represents the
between the projected GDP in the business-as-usual trajectory direction and standard deviations of the relationship. The regression
( fpre (2005)) and the actual 2005 GDP in the post-Katrina trajectory analyses were conducted in the linear model (lm) function in R.

4
Y. Qiang, et al. Sustainable Cities and Society 57 (2020) 102115

Fig. 2. The track of Hurricane Katrina and counties that are presidentially declared as disaster area.

pre-Katrina growth trend persists. The accumulated loss of GDP from

2005 to 2013 (gray area in in Fig. 5) is about 2.2 billion current dollar
and the GDP is unlikely to restore to the business-as-usual trajectory in
the following years. Note, this loss is estimated only from the declined
GDP in the FEMA declared disaster area, which does not include other
indirect losses in a broader area. Note, due to the instability of the
DMSP-OLS images, the estimated GDP fluctuates from year to year.
Although the estimation of GDP in a specific year may not be accurate,
the general trend indicates the strong economic impact of the disaster.

4.2. Spatial variation of resilience

Fig. 3. Framework of GDP trajectory in Hurricane Katrina. The GDP recovery trajectory varies in counties. As illustrated in
Fig. 6, the estimated GDP in Orleans Parish (blue) and St. Bernard
4. Result Parish (red) had a substantive decline in 2005 when Katrina stroke.
Both parishes are in the metropolitan area of New Orleans near the
4.1. Overall economic impact landfall location of Katrina. The post-Katrina GDP in St. Bernard Parish
shows a faster recovery rate than Orleans Parish after Katrina. In con-
Fig. 4 demonstrates the change of NTL brightness (DN value) from trast, St. Tammany Parish (green) in the north shore of Lake Pontch-
2004 to 2005 near the landfall location of Katrina, where a decline of artrain had little economic impact and maintained a steady GDP growth
NTL brightness can be observed in New Orleans and the surrounding after Katrina, despite its proximity to the hurricane track.
coastal cities. The annual GDP estimates from the NTL data (see Fig. 5) As shown in Fig. 7 (a), Hurricane Katrina caused large instant dis-
reveal that Hurricane Katrina has fundamentally altered the economic turbance (D ) in counties near the two landfall locations in Florida and
growth in the declared disaster area. In contrast to the steady growth in Louisiana. Note, the high D in southwest Louisiana (around Lake
the Southeast Region (despite the slight drop in 2009 due to the fi- Charles Parish) is possibly due to Hurricane Rita, a Category 3 hurri-
nancial crisis), the GDP in the disaster area declined sharply in 2005 cane landed near the Louisiana-Texas border a month after Katrina.
and grew at a slower rate in the following years. The Southeast Region Fig. 7 (b) shows that Louisiana counties have a higher recovery rate (Rr)
is delineated by Bureau of Economic Analysis, U.S. Department of of GDP after Katrina, except Jefferson and Orleans Parish near the
Commerce (2015) for releasing economic statistics (see Fig. 2). Ac- center of New Orleans which were struggling to recover after Katrina.
cording to our estimation, the GDP in 2005 in the entire disaster area High accumulated GDP loss (L) are distributed in coastal cities, in-
declined 24.2 % compared with the business-as-usual condition if the cluding Gulfport (MS), Mobile (AL), Panama City and Miami (FL). The
inland counties generally have lower accumulated loss.

5
Y. Qiang, et al. Sustainable Cities and Society 57 (2020) 102115

Table 2
Description of variables used in the regression analyses.
Category Variable Description Data source Preprocessing

Impact intensity Max gust Recorded maximum gust speed from Aug. 23–30, 2005 Knobb et al. (2005) Kriging interpolation, Zonal
operation
Rainfall Total rainfall from Aug. 23–30, 2005 National Weather Service Re-sampling, Zonal operation

Environmental Mean elevation Mean elevation U.S. Geological Survey Zonal operation
Distance to Coast Mean distance to coastline NOAA Euclidean distance, Zonal
operation
% of urban in flood Percent of developed land in flood zone FEMA flood map, National Land Method in (Qiang, 2019)
zone Cover Database

Socio-economic % White Percent of population in one race: White U.S. decennial census (2000)
% Black Percent of population in one race: Black or African American U.S. decennial census (2000)
% Asian Percent of population in one race: Asian U.S. decennial census (2000)
% Hispanic & Latino Percent of Hispanic or Latino population U.S. decennial census (2000)
% children Percent of population under 18 years old U.S. decennial census (2000)
% elderly adult Percent of population above 65 years old U.S. decennial census (2000)
% owner occupied Percent of owner occupied housing unit U.S. decennial census (2000)
homes
% bachelor degree Percent of population (> 25 years old) with 4 or more years U.S. decennial census (2000)
of college or bachelor's degree or higher
% poverty Percent of population whose income is below poverty level U.S. decennial census (2000)
Per cap. income Per capita income U.S. decennial census (2000)

Industrial structure % small businesses Percent of establishments with < 20 employees County business patterns (2004)
% large businesses Percent of establishments with > 500 employees County business patterns (2004)
% Agriculture Percent of establishments in agriculture, forestry, fishing and County business patterns (2004)
hunting
% Mining Percent of establishments in mining, quarrying, and oil and County business patterns (2004)
gas extraction
% Manufacture Percent of establishments in manufacturing County business patterns (2004)

4.3. Regression analysis variables can only explain 39.9 %, 29.3 % and 27.4 % variance of D, Rr,
and L respectively (Table 4), which implies that other variables should
Various relationships were found between the three metrics (D, Rr, be considered to model the recovery pattern. Due to the different
L) and the selected variables (Table 3): numbers of variables in the four categories, adjusted R2 was used to
Instant disturbance (D) has the strongest relationship with the compare the variance explained by the different categories. In general,
percent of Asian people (highest R2) among the 20 selected variables. socio-economic variables can best predict (highest adjusted R2) instant
The positive coefficient b indicates that counties with a higher ratio of disturbance (D), followed by environmental condition and industrial
Asian people have endured greater GDP disturbance in Katrina. This structure. Industrial structure is the best indicator of recovery rate (Rr),
result confirms the findings in Vu, VanLandingham, Do, and Bankston followed by environmental and socio-economic variables. Socio-eco-
(2009) that 78 % of Vietnamese (the largest Asian group in Louisiana nomic variables can best predict the accumulated economic loss (L). It
and Mississippi) left their home during the hurricane and gradually is worth-noting that the disaster impacts explain the least variance in all
returns afterwards. D is also highly correlated with the physical impacts the three metrics, which implies that the intrinsic community capacities
and environmental conditions. Specifically, counties with high gust (e.g. socio-economic conditions) are more decisive to economic re-
speed, high accumulated rainfall, low elevation, high ratio of urban in covery than the physical disaster impacts.
flood zone, or close to the coastline tend to have a high D. This reflects
the fact that New Orleans, which has low elevation and high ratio of 5. Discussion
urban flood exposure, was seriously damaged by flood inundation due
to levee breech. Additionally, counties with a higher percentage of Despite the extensive discussions in the literature, quantitative as-
agricultural industry have a lower D. sessment of community resilience is still a challenge due to the lack of
High recovery rates (Rr) are associated with high percentages of empirical data continuously collected in disasters. In the U.S, census
energy industry (e.g. mining, quarrying, and oil and gas extraction), data are released decennially and county-level GDP data is only avail-
implying that the energy industry rebounded more quickly after the able in limited years, not to mention the developing world. Other data
disturbance. Moreover, low elevation, proximity to coast, high ratios of collection methods such as field surveys and interviews are costly and
children and elderly people tend to impede the GDP recovery. time-consuming. As an alternative, NTL remote sensing is an efficient
Accumulated economic loss (L) is negatively correlated (highest R2) means to observe human activities (such as population, GDP, and en-
with the ratio of owner-occupied housing units, indicating communities ergy consumption) from space. Most importantly, the NTL images have
with more home-owner residents managed to prevent the overall eco- the unique ability to detect declines of human activities, which is not
nomic loss. Communities with a higher ratio of Asian, Hispanic and easy for other types of remote sensing imageries. The continuous scan
Latino, children and elderly adults have suffered higher accumulated of NTL images can capture the disturbance and recovery pattern of
loss. High L is also found in counties with a high urban exposure to human activity during natural disasters at low cost and in a timely
flood zone. Additionally, counties reliant on agricultural and mining manner. The introduced framework can increase our understanding
industries tend to have a lower long-term loss. about community resilience and help to improve resilience prediction
R2 of the multivariate regression indicates that the selected models in terms of variable selection and weighting.

6
Y. Qiang, et al. Sustainable Cities and Society 57 (2020) 102115

Fig. 4. Changes of NTL brightness (DN value) in the 2004 and 2005 DMSP-OLS annual composite images.

In addition to Hurricane Katrina, the introduced assessment fra- Not limited to the DMSP/OLS images, the introduced framework can
mework based on NTL data is applicable to other natural disasters (e.g. use other data sources as input. For instance, the Visible Infrared
tsunami and earthquake) that cause economic disturbance. The fra- Imaging Radiometer Suite (VIIRS) images, the new generation NTL
mework is particularly useful in situations where official socio-eco- remote sensing products since 2011, provide many new features (e.g.
nomic data are not available at the desired spatio-temporal resolution. higher spatial and radiometric resolution) that are useful for measuring

Fig. 5. Annual estimated GDP in current dollars in the affected area (refer to the left axis) and the Southeast Region (refer to the right axis) (Data source: Bureau of
Economic Analysis).

7
Y. Qiang, et al. Sustainable Cities and Society 57 (2020) 102115

Fig. 6. Estimated GDP time series of Orleans Parish (blue), St. Bernard Parish (red) and St. Tammany Parish (green).

community resilience in disasters with a smaller impact area and communities with some demographic and socio-economic character-
shorter period. istics may have difficulties to bounce back, pinpointing areas where
As demonstrated in this study, the GDP time series estimated from special assistance or policy levers should be applied. Compared with the
the NTL data can capture the various recovery trajectories at the county traditional resilience indices without specific outputs, these analysis
level. The analysis results uncover the strong economic impact of results can provide more actionable information to support decision-
Katrina in the affected region, where the GDP has given up its rapid making at different phases of a disaster.
growth and shifted to a different trajectory after 2005. This finding Despite the merits demonstrated in this study, the introduced ap-
confirms the tremendous and long-lasting impact of Katrina on popu- proach can be improved in the following aspects. First, despite the good
lation migration (2010, Fussell, Curtis, & DeWaard, 2014), declined model fit at the county level, the DMSP-OLS images cannot perfectly
urban growth (Qiang & Lam, 2016) and economic activity (Baade, predict economic activity due to their inherent limitations (e.g. low
Baumann, & Matheson, 2007; Petterson, Stanley, Glazier, & Philipp, spatial resolution, lack of onboard and inter-satellite calibration, and
2006). According to the recent estimates of the U.S. Census (2019), limited dynamic range). In future studies, the assessment results need to
until 2017 the population and housing price in New Orleans has not yet be validated against additional datasets such as other NTL products
recovered to the pre-Katrina level. The various GDP recovery patterns (e.g. VIIRS images) or socio-economic data continuously collected
reveal geographical disparities of community resilience related to the during the disaster. Second, other factors that influence economic re-
local environmental and socio-economic conditions. covery should be taken into account in future studies. For instance,
The recovery pattern of a social system is dependent on both in- economic cycles (e.g. great recessions) can slow down business re-
tensity of disaster impacts and resilience of the system. The analysis covery after a disaster. In the introduced approach, linear models were
results (see Table 4) suggest that the physical impact variables (in- used to generalize the economic growth in the pre- and post-Katrina
cluding max gust speed and accumulated rainfall) can only explain periods, where the yearly fluctuations (e.g. the financial crisis in 2009)
limited variance (R2 = 0.065) in the recovery pattern. Instead, the are averaged in the trend lines. Ideally, the “business-as-usual” trajec-
inherent conditions of communities (including environmental, socio- tories should be projected with more sophisticated models to eliminate
economic and industrial conditions) play a central role in shaping the the effect of economic cycles. Third, the 20 selected variables can only
recovery pattern. The univariate regression with individual variables explain ∼40 % variation of the measured metrics, which reveal the
provides empirical evidence about the underlying factors that influence complexity of community resilience. The prediction power of the model
the recovery. The results may also inform specific plans to prompt re- can be improved by including more variables and using more sophis-
silience at different phases of a disaster. For instance, the strong cor- ticated model specifications (e.g. non-linear models). Robust models for
relation between instant disturbance (D) and the environmental con- community resilience prediction can be developed by accumulating
ditions (e.g. elevation, proximity to coast and % of urban area in flood empirical evidence in more disaster events.
zone) suggests that reducing exposure is the most effective way to re-
duce the direct impact caused by disasters. In the recovery phase,

Fig. 7. Spatial variation of (a) instant disturbance (D ), (b) recovering rate (Rr) and (c) accumulated GDP loss (L ). Red indicates high instant disturbance (D), slower
recovery (Rr) and high accumulated loss (L).

8
Y. Qiang, et al. Sustainable Cities and Society 57 (2020) 102115

Table 3
Univariate regression analysis of instant disturbance (D), recovery (Rr) and economic loss (L) with environmental and socio-economic variables. Bold font indicates
statistical significance (p < 0.01). Detailed statistics of the regression analysis (e.g. confidence intervals, standard errors of residuals, and residual distributions) can
be found in the supplementary material.
Variable category Variable Instant disturbance (D) Recovery rate (Rr) Economic loss (L)

2 2
b p value R b p value R b p value R2

Impact intensity Max gust 0.272 0.000 0.074 0.000 0.995 0.000 0.046 0.545 0.002
Rainfall 0.257 0.001 0.066 −0.064 0.393 0.004 0.053 0.477 0.003

Environment Mean elevation −0.375 0.000 0.140 −0.336 0.000 0.113 −0.113 0.132 0.013
Distance to Coast −0.358 0.000 0.128 −0.252 0.001 0.063 −0.081 0.279 0.007
% of urban in flood zone 0.346 0.000 0.146 0.117 0.122 0.015 0.239 0.003 0.056

Socio-economic % White 0.026 0.733 0.001 −0.057 0.447 0.003 0.013 0.867 0.000
% Black −0.060 0.425 0.004 0.068 0.369 0.005 −0.041 0.583 0.002
% Asian 0.420 0.000 0.176 −0.083 0.271 0.007 0.333 0.000 0.111
% Hispanic & Latino 0.299 0.000 0.089 −0.139 0.064 0.019 0.213 0.004 0.045
% children 0.086 0.250 0.007 −0.257 0.001 0.066 0.255 0.001 0.065
% elderly adult 0.083 0.267 0.007 −0.267 0.000 0.071 0.248 0.001 0.061
% owner occupied homes −0.274 0.000 0.075 0.126 0.092 0.016 −0.338 0.000 0.114
% bachelor degree 0.161 0.031 0.026 −0.102 0.172 0.011 0.203 0.006 0.041
% poverty −0.136 0.068 0.019 0.110 0.141 0.012 −0.129 0.086 0.017
Per cap. income 0.243 0.001 0.059 −0.140 0.061 0.020 0.243 0.001 0.059

Industrial structure % small businesses −0.184 0.014 0.034 −0.116 0.123 0.013 −0.085 0.259 0.007
% large businesses 0.090 0.231 0.008 −0.110 0.143 0.012 0.142 0.057 0.020
% Agriculture −0.321 0.000 0.103 −0.116 0.121 0.014 −0.243 0.001 0.059
% Mining 0.084 0.263 0.007 0.404 0.000 0.163 −0.214 0.004 0.046
% Manufacture −0.154 0.040 0.024 −0.077 0.303 0.006 −0.088 0.243 0.008

Table 4
Adjusted R2 of the multivariate regressions between the resilience metrics (D, Rr, and L) and variables in different categories.
Variable category Instant disturbance (D) Recovery rate (Rr) Economic loss (L)

2 2 2 2
R Adj. R R Adj. R R2 Adj. R2

All variables 0.399 0.327 0.369 0.293 0.274 0.187

Impact intensity 0.076 0.065 0.016 0.004 0.003 −0.008
Environmental 0.188 0.172 0.089 0.071 0.056 0.038
Socio-economic 0.286 0.243 0.099 0.045 0.246 0.201
Industrial structure 0.138 0.114 0.194 0.170 0.114 0.088

6. Conclusion Funding source

This study introduces a quantitative framework for resilience as- This article is based on work supported by two research grants from
sessment using DMSP-OLS Nighttime Lights images. The framework the U.S. National Science Foundation: one under the Coastlines and
was applied to model the recovery patterns of economic activity in the People (CoPe) Program (Award No. 1940091) and the other under the
affected area in Hurricane Katrina 2005. The analyses show the great Methodology, Measurement & Statistics (MMS) Program (Award No.
economic disturbance caused by Katrina and the slow recovery in the 1853866). Any opinions, findings, and conclusions or recommendations
entire affected area. The county-level analyses indicate strong spatial expressed in this material are those of the authors and do not ne-
variation of the recovery pattern. Statistical analyses were carried out cessarily reflect the views of the funding agencies.
to explore the underlying factors that influence the recovery patterns.
This study demonstrates the utility of NTL images in monitoring human
dynamics in natural disasters, which filled the critical gap of empirical Declaration of Competing Interests
data and assessment methods for resilience research. Based on the
framework of recovery trajectory, resilience is modelled as a dynamic The authors declare that they have no known competing financial
process. Compared with the traditional resilience indices, the modeling interests or personal relationships that could have appeared to influ-
results can provide more specific and actionable measures to promot ence the work reported in this paper.
resilience in diverse communities and in different phases of a disaster.
This study re-visits Hurricane Katrina using DMSP-OLS NTL images.
However, the assessment approach is applicable for other disaster Appendix A. Supplementary data
events using other NTL products (e.g. VIIRS DNB images). The results
increase our understanding about the complexity of community resi- Supplementary material related to this article can be found, in the
lience and provide support for decision-makers to develop resilient and online version, at doi:https://doi.org/10.1016/j.scs.2020.102115.
sustainable communities.

9
Y. Qiang, et al. Sustainable Cities and Society 57 (2020) 102115

References Huang, Q., Yang, X., Gao, B., Yang, Y., & Zhao, Y. (2014). Application of DMSP/OLS
nighttime light images: A meta-analysis and a systematic literature review. Remote
Sensing, 6, 6844–6866. https://doi.org/10.3390/rs6086844.
Adeola, F. O. (2009). Mental health & psychosocial distress sequelae of Katrina: An em- Huang, X., Wang, C., & Lu, J. (2019). Understanding spatiotemporal development of
pirical study of survivors. Human Ecology Review, 16, 195–210. human settlement in hurricane-prone areas on U.S. Atlantic and gulf coasts using
Adger, W. N. (2000). Social and ecological resilience: Are they related? Progress in Human nighttime remote sensing. Natural Hazards and Earth System Sciences Discussions.
Geography, 24, 347–364. https://doi.org/10.1191/030913200701540465. https://doi.org/10.5194/nhess-2019-64 Discuss. 1–22.
Baade, R. A., Baumann, R., & Matheson, V. (2007). Estimating the economic impact of Hultquist, C., Simpson, M., Cervone, G., & Huang, Q. (2015). Using nightlight remote
natural and social disasters, with an application to hurricane katrina. Urban Studies, sensing imagery and twitter data to study power outages, in: Proceedings of the 1st
44, 2061–2076. https://doi.org/10.1080/00420980701518917. ACM SIGSPATIAL international workshop on the use of GIS in emergency manage-
Bakkensen, L. A., Fox‐Lent, C., Read, L. K., & Linkov, I. (2017). Validating resilience and ment, EM-GIS’ 15. ACM, New York, NY, USA. https://doi.org/10.1145/2835596.
vulnerability indices in the context of natural disasters. Risk Analysis, 37, 982–1004. 2835601 pp. 6:1–6:6.
https://doi.org/10.1111/risa.12677. Hung, H.-C., Yang, C.-Y., Chien, C.-Y., & Liu, Y.-C. (2016). Building resilience:
Bates, P. D. (2004). Remote sensing and flood inundation modelling. Hydrological Mainstreaming community participation into integrated assessment of resilience to
Processes, 18, 2593–2597. https://doi.org/10.1002/hyp.5649. climatic hazards in metropolitan land use management. Land Use Policy, 50, 48–58.
Beccari, B. (2016). A comparative analysis of disaster risk, vulnerability and resilience https://doi.org/10.1016/j.landusepol.2015.08.029.
composite indicators. PLOS Currents Disasters. https://doi.org/10.1371/currents.dis. Jonkman, S. N., Maaskant, B., Boyd, E., & Levitan, M. L. (2009). Loss of life caused by the
453df025e34b682e9737f95070f9b970. flooding of New Orleans after hurricane katrina: Analysis of the relationship between
Bureau of Economic Analysis, U.S. Department of Commerce (2018). Prototype gross do- flood characteristics and mortality. Risk Analysis, 29, 676–698. https://doi.org/10.
mestic product by county. 2012–2015. 1111/j.1539-6924.2008.01190.x.
Bureau of Economic Analysis, U.S. Department of Commerce (2015). Gross domestic Joshi, N., Baumann, M., Ehammer, A., Fensholt, R., Grogan, K., Hostert, P., et al. (2016).
product by state. Advance 2014 and Revised 1997-2013. A review of the application of optical and radar remote sensing data fusion to land
Burren, D., & Neusser, K. (2010). The decline in volatility of US GDP growth. Applied use mapping and monitoring. Remote Sensing, 8, 70. https://doi.org/10.3390/
Economics Letters, 17, 1625–1631. https://doi.org/10.1080/13504850903085050. rs8010070.
Burton, A. (2006). Health disparities: Crisis not over for hurricane victims. Environmental Knobb, R. D., Rhome, J. R., & Brown, R. D. (2005). Tropical Cyclone Report: Hurricane
Health Perspectives, 114, A462. KatrinaNOAA National Hurricane Centerhttps://www.nhc.noaa.gov/data/tcr/
Cai, H., Lam, N. S. N., Zou, L., & Qiang, Y. (2018). Modeling the dynamics of community AL122005_Katrina.pdf.
resilience to coastal hazards using a bayesian network. Annals of the American Kohiyama, M., Hayashi, H., Maki, N., Higashida, M., Kroehl, H. W., Elvidge, C. D., et al.
Association of Geographers, 108, 1260–1279. https://doi.org/10.1080/24694452. (2004). Early damaged area estimation system using DMSP-OLS night-time imagery.
2017.1421896. International Journal of Remote Sensing, 25, 2015–2036. https://doi.org/10.1080/
Cooner, A. J., Shao, Y., & Campbell, J. B. (2016). Detection of urban damage using remote 01431160310001595033.
sensing and machine learning algorithms: Revisiting the 2010 Haiti earthquake. Koliou, M., Lindt, J. W., van de, McAllister, T. P., Ellingwood, B. R., Dillard, M., & Cutler,
Remote Sensing, 8, 868. https://doi.org/10.3390/rs8100868. H. (2018). State of the research in community resilience: Progress and challenges.
Cutter, S. L., Ash, K. D., & Emrich, C. T. (2014). The geographies of community disaster Sustain. Resilient Infrastruct., 0, 1–21. https://doi.org/10.1080/23789689.2017.
resilience. Global Environmental Change: Human and Policy Dimensions, 29, 65–77. 1418547.
https://doi.org/10.1016/j.gloenvcha.2014.08.005. Kuffer, M., Pfeffer, K., & Sliuzas, R. (2016). Slums from space—15 years of slum mapping
Cutter, S. L., Barnes, L., Berry, M., Burton, C., Evans, E., Tate, E., et al. (2008). A place- using remote sensing. Remote Sensing, 8, 455. https://doi.org/10.3390/rs8060455.
based model for understanding community resilience to natural disasters. Global Lam, N. S. N., Pace, K., Campanella, R., LeSage, J., & Arenas, H. (2009). Business return in
Environmental Change: Human and Policy Dimensions, 18, 598–606. https://doi.org/ New Orleans: Decision making amid post-katrina uncertainty. PloS One, 4, e6765.
10.1016/j.gloenvcha.2008.07.013. https://doi.org/10.1371/journal.pone.0006765.
Cutter, S. L., Boruff, B. J., & Shirley, W. L. (2003). Social vulnerability to environmental Lam, N. S.-N., Qiang, Y., Arenas, H., Brito, P., & Liu, K.-B. (2015). Mapping and assessing
hazards. Social Science Quarterly, 84, 242–261. https://doi.org/10.1111/1540-6237. coastal resilience in the Caribbean region. Cartography and Geographic Information
8402002. Science, 406, 1–8. https://doi.org/10.1080/15230406.2015.1040999.
Cutter, S. L., Burton, C. G., & Emrich, C. T. (2010). Disaster resilience indicators for Lam, N. S. N., Reams, M., Li, K., Li, C., & Mata, L. P. (2016). Measuring community
benchmarking baseline conditions. Journal of Homeland Security and Emergency resilience to coastal hazards along the Northern Gulf of Mexico. Natural Hazards
Management, 7, 14. https://doi.org/10.2202/1547-7355.1732. Review, 17. https://doi.org/10.1061/(ASCE)NH.1527-6996.0000193 e193–e193.
Dasgupta, S., Laplante, B., Murray, S., & Wheeler, D. (2011). Exposure of developing LeSage, J. P., Pace, R. K., Lam, N., Campanella, R., & Liu, X. (2011). New Orleans business
countries to sea-level rise and storm surges. Climatic Change, 106, 567–579. https:// recovery in the aftermath of Hurricane Katrina. Journal of the Royal Statistical Society
doi.org/10.1007/s10584-010-9959-6. Series A, 174, 1007–1027. https://doi.org/10.1111/j.1467-985X.2011.00712.x.
Dong, L., & Shan, J. (2013). A comprehensive review of earthquake-induced building Li, S., Dragicevic, S., Castro, F. A., Sester, M., Winter, S., Coltekin, A., et al. (2016).
damage detection with remote sensing techniques. ISPRS Journal of Photogrammetry Geospatial big data handling theory and methods: A review and research challenges.
and Remote Sensing, 84, 85–99. https://doi.org/10.1016/j.isprsjprs.2013.06.011. ISPRS Journal of Photogrammetry and Remote Sensing, 115, 119–133. https://doi.org/
Elvidge, C. D., Ziskin, D., Baugh, K. E., Tuttle, B. T., Ghosh, T., Pack, D. W., et al. (2009). 10.1016/j.isprsjprs.2015.10.012 Theme issue “State-of-the-art in photogrammetry,
A fifteen year record of global natural gas flaring derived from satellite data. Energies, remote sensing and spatial information science”.
2, 595–622. https://doi.org/10.3390/en20300595. Li, X., Ge, L., & Chen, X. (2013). Detecting Zimbabwe’s decadal economic decline using
Faturechi, R., & Miller-Hooks, E. (2015). Measuring the performance of transportation nighttime light imagery. Remote Sensing, 5, 4551–4570. https://doi.org/10.3390/
infrastructure systems in disasters: A comprehensive review. Journal of Infrastructure rs5094551.
Systems, 21, 04014025. https://doi.org/10.1061/(ASCE)IS.1943-555X.0000212. Li, X., Zhan, C., Tao, J., & Li, L. (2018). Long-term monitoring of the impacts of disaster
Ferrara, L., Marcellino, M., & Mogliani, M. (2015). Macroeconomic forecasting during the on human activity using DMSP/OLS nighttime light data: A case study of the 2008
Great Recession: The return of non-linearity? International Journal of Forecasting, 31, wenchuan, china earthquake. Remote Sensing, 10, 588. https://doi.org/10.3390/
664–679. https://doi.org/10.1016/j.ijforecast.2014.11.005. rs10040588.
Finch, C., Emrich, C. T., & Cutter, S. L. (2010). Disaster disparities and differential re- Liao, K.-H. (2012). A theory on urban resilience to floods—A basis for alternative plan-
covery in New Orleans. Population and Environment, 31, 179–202. https://doi.org/10. ning practices. Ecology and Society, 17. https://doi.org/10.5751/ES-05231-170448.
1007/s11111-009-0099-8. Marcellino, M. (2008). A linear benchmark for forecasting GDP growth and inflation?
Foster, K. (2012). In search of regional resilience. In M. Weir, H. Wial, & H. Wolman Journal of Forecasting, 27, 305–340. https://doi.org/10.1002/for.1059.
(Eds.). Building regional resilience: Urban and regional policy and its effects. Washington, Martin, R., & Sunley, P. (2015). On the notion of regional economic resilience:
DC: Brookings Institute Press. Conceptualization and explanation. Journal of Economic Geography, 15, 1–42. https://
Fussell, E., Curtis, K. J., & DeWaard, J. (2014). Recovery migration to the City of New doi.org/10.1093/jeg/lbu015.
Orleans after Hurricane Katrina: A migration systems approach. Population and Nicholls, R. J., Hoozemans, F. M. J., & Marchand, M. (1999). Increasing flood risk and
Environment, 35, 305–322. https://doi.org/10.1007/s11111-014-0204-5. wetland losses due to global sea-level rise: Regional and global analyses. Global
Fussell, E., Sastry, N., & VanLandingham, M. (2010). Race, socioeconomic status, and Environmental Change, 9, S69–S87. https://doi.org/10.1016/S0959-3780(99)
return migration to New Orleans after Hurricane Katrina. Population and Environment, 00019-9.
31, 20–42. https://doi.org/10.1007/s11111-009-0092-2. NOAA (2006). Service assessment: Hurricane katrina August. 23-31, 2005.
Gillespie, T. W., Frankenberg, E., Chum, K. F., & Thomas, D. (2014). Nighttime lights time Norris, F. H., Stevens, S. P., Pfefferbaum, B., Wyche, K. F., & Pfefferbaum, R. L. (2008).
series of tsunami damage, recovery, and economic metrics in Sumatra. Indonesia Community resilience as a metaphor, theory, set of capacities, and strategy for dis-
Remote Sensing Letter Print, 5, 286–294. https://doi.org/10.1080/2150704X.2014. aster readiness. American Journal of Community Psychology, 41, 127–150. https://doi.
900205. org/10.1007/s10464-007-9156-6.
Holling, C. S. (1973). Resilience and stability of ecological systems. Annual Review of Peacock, W. G. (2010). Final report advancing the resilience of coastal localities 10-02R.
Ecology and Systematics, 4, 1–23. https://doi.org/10.1146/annurev.es.04.110173. Perz, S. G., Muñoz-Carpena, R., Kiker, G., & Holt, R. D. (2013). Evaluating ecological
000245. resilience with global sensitivity and uncertainty analysis. Ecological Modelling, 263,
Hong, Y., Adler, R., & Huffman, G. (2007). Use of satellite remote sensing data in the 174–186. https://doi.org/10.1016/j.ecolmodel.2013.04.024.
mapping of global landslide susceptibility. Natural Hazards, 43, 245–256. https://doi. Petterson, J. S., Stanley, L. D., Glazier, E., & Philipp, J. (2006). A preliminary assessment
org/10.1007/s11069-006-9104-z. of social and economic impacts associated with hurricane katrina. American
Hsu, F.-C., Baugh, K. E., Ghosh, T., Zhizhin, M., & Elvidge, C. D. (2015). DMSP-OLS ra- Anthropologist, 108, 643–670. https://doi.org/10.1525/aa.2006.108.4.643.
diance calibrated nighttime lights time series with intercalibration. Remote Sensing, 7, Qiang, Y. (2019). Disparities of population exposed to flood hazards in the United States.
1855–1876. https://doi.org/10.3390/rs70201855. Journal of Environmental Management, 232, 295–304. https://doi.org/10.1016/j.

10
Y. Qiang, et al. Sustainable Cities and Society 57 (2020) 102115

jenvman.2018.11.039. resiliency of French Polynesian coral reefs to environmental disturbances. Scientific

Qiang, Y., & Lam, N. S.-N. (2016). The impact of Hurricane Katrina on urban growth in Reports, 9, 1027. https://doi.org/10.1038/s41598-018-38228-5.
Louisiana: An analysis using data mining and simulation approaches. International Vetrivel, A., Gerke, M., Kerle, N., Nex, F., & Vosselman, G. (2018). Disaster damage de-
Journal of Geographical Information Science, 30, 1832–1852. https://doi.org/10.1080/ tection through synergistic use of deep learning and 3D point cloud features derived
13658816.2016.1144886. from very high resolution oblique aerial images, and multiple-kernel-learning. ISPRS
Qiang, Y., & Xu, J. (2019). Empirical assessment of road network resilience in natural Journal of Photogrammetry and Remote Sensing, 140, 45–59. https://doi.org/10.1016/
hazards using crowdsourced traffic data. International Journal of Geographical j.isprsjprs.2017.03.001 Geospatial Computer Vision.
Information Science, 1–17. https://doi.org/10.1080/13658816.2019.1694681. Vu, L., VanLandingham, M. J., Do, M., & Bankston, C. L. (2009). Evacuation and return of
Sastry, N., & VanLandingham, M. (2009). One year later: Mental illness prevalence and vietnamese new orleanians affected by hurricane katrina. Organization & Environment,
disparities among new orleans residents displaced by hurricane katrina. American 22, 422–436. https://doi.org/10.1177/1086026609347187.
Journal of Public Health, 99, S725–S731. https://doi.org/10.2105/AJPH.2009. White, R. K., Edwards, W. C., Farrar, A., & Plodinec, M. J. (2015). A practical approach to
174854. building resilience in America’s communities. The American Behavioral Scientist, 59,
Sempier, T. T., Swann, D. L., Emmer, R., Sempier, S. H., & Schneider, M. (2010). Coastal 200–219. https://doi.org/10.1177/0002764214550296.
community resilience index: A community self-assessment (No. MASGP-08-014). World Bank (2019). World Bank national accounts data.
Shahtahmassebi, A. R., Song, J., Zheng, Q., Blackburn, G. A., Wang, K., Huang, L. Y., et al. Xie, Y., & Weng, Q. (2017). Spatiotemporally enhancing time-series DMSP/OLS nighttime
(2016). Remote sensing of impervious surface growth: A framework for quantifying light imagery for assessing large-scale urban dynamics. ISPRS Journal of
urban expansion and re-densification mechanisms. International Journal of Applied Photogrammetry and Remote Sensing, 128, 1–15. https://doi.org/10.1016/j.isprsjprs.
Earth Observation and Geoinformation, 46, 94–112. https://doi.org/10.1016/j.jag. 2017.03.003.
2015.11.007. Yodo, N., & Wang, P. (2016). Engineering resilience quantification and system design
Sherrieb, K., Norris, F. H., & Galea, S. (2010). Measuring capacities for community re- implications: A literature survey. Journal of Mechanical Design, 138. https://doi.org/
silience. Social Indicators Research, 99, 227–247. https://doi.org/10.1007/s11205- 10.1115/1.4034223.
010-9576-9. Zhao, X., Yu, B., Liu, Y., Yao, S., Lian, T., Chen, L., et al. (2018). NPP-VIIRS DNB daily
Spalding, M. D., McIvor, A. L., Beck, M. W., Koch, E. W., Möller, I., Reed, D. J., et al. data in natural disaster assessment: Evidence from selected case studies. Remote
(2014). Coastal ecosystems: A critical element of risk reduction. Conservation Letters, Sensing, 10, 1526. https://doi.org/10.3390/rs10101526.
7, 293–301. https://doi.org/10.1111/conl.12074. Zhuo, L., Ichinose, T., Zheng, J., Chen, J., Shi, P. J., & Li, X. (2009). Modelling the po-
Tebaldi, C., Strauss, B. H., & Zervas, C. E. (2012). Modelling sea level rise impacts on pulation density of China at the pixel level based on DMSP/OLS non‐-
storm surges along US coasts. Environmental Research Letters, 7, 014032. https://doi. radiance‐calibrated night‐time light images. International Journal of Remote Sensing,
org/10.1088/1748-9326/7/1/014032. 30, 1003–1018. https://doi.org/10.1080/01431160802430693.
Timmerman, P. (1981). Vulnerability, resilience and the collapse of society - a review of Zou, L., Lam, N. S.-N., Cai, H., & Qiang, Y. (2018). Mining twitter data for improved
models and possible climatic applications. University of Toronto Institute For understanding of disaster resilience. Annals of the American Association of Geographers,
Environmental Studies Environmental Monograph. Institute for Environmental 108(5), 1422–1441. https://doi.org/10.1080/24694452.2017.1421897.
Studies, University of Toronto. Zou, L., Lam, N. S. N., Shams, S., Cai, H., Meyer, M. A., Yang, S., et al. (2018). Social and
U.S. Census (2019). American community survey. geographical disparities in Twitter use during Hurricane Harvey. International Journal
U.S. Census Bureau (2011). Census 2010. U.S. Census Bureau. of Digital Earth, 1–19. https://doi.org/10.1080/17538947.2018.1545878.
Vercelloni, J., Kayal, M., Chancerelle, Y., & Planes, S. (2019). Exposure, vulnerability, and

11