ArticlePDF Available

Abstract and Figures

In 2017, Hurricane Harvey caused substantial loss of life and property in the swiftly urbanizing region of Houston, TX. Now in its wake, researchers are tasked with investigating how to plan for and mitigate the impact of similar events in the future, despite expectations of increased storm intensity and frequency as well as accelerating urbanization trends. Critical to this task is the development of automated workflows for producing accurate and consistent land cover maps of sufficiently fine spatio-temporal resolution over large areas and long timespans. In this study, we developed an innovative automated classification algorithm that overcomes some of the traditional trade-offs between fine spatio-temporal resolution and extent – to produce a multi-scene, 30m annual land cover time series characterizing 21 years of land cover dynamics in the 35,000 km2 Greater Houston area. The ensemble algorithm takes advantage of the synergistic value of employing all acceptable Landsat imagery in a given year, using aggregate votes from the posterior predictive distributions of multiple image composites to mitigate against misclassifications in any one image, and fill gaps due to missing and contaminated data, such as those from clouds and cloud shadows. The procedure is fully automated, combining adaptive signature generalization and spatio-temporal stabilization for consistency across sensors and scenes. The land cover time series is validated using independent, multi-temporal fine-resolution imagery, achieving crisp overall accuracies between 78–86% and fuzzy overall accuracies between 91–94%. Validated maps and corresponding areal cover estimates corroborate what census and economic data from the Greater Houston area likewise indicate: rapid growth from 1997–2017, demonstrated by the conversion of 2,040 km² (± 400 km²) to developed land cover, 14% of which resulted from the conversion of wetlands. Beyond its implications for urbanization trends in Greater Houston, this study demonstrates the potential for automated approaches to quantifying large extent, fine resolution land cover change, as well as the added value of temporally-dense time series for characterizing higher-order spatio-temporal dynamics of land cover, including periodicity, abrupt transitions, and time lags from underlying demographic and socio-economic trends.
Content may be subject to copyright.
Characterizing multi-decadal, annual land cover change
dynamics in Houston, TX based on automated classication
of Landsat imagery
C.R. Hakkenberg
a
, M.P. Dannenberg
b,c
, C. Song
d
and K.B. Ensor
a
a
Department of Statistics, Rice University, Houston, TX, USA;
b
Department of Geographical and
Sustainability Sciences, University of Iowa, Iowa City, IA, USA;
c
School of Natural Resources and the
Environment, University of Arizona, Tucson, AZ, USA;
d
Department of Geography, University of North
Carolina at Chapel Hill, Chapel Hill, NC, USA
ABSTRACT
In 2017, Hurricane Harvey caused substantial loss of life and property
in the swiftly urbanizing region of Houston, TX. Now in its wake,
researchers are tasked with investigating how to plan for and miti-
gate the impact of similar events in the future, despite expectations
of increased storm intensity and frequency as well as accelerating
urbanization trends. Critical to this task is the development of auto-
mated workows for producing accurate and consistent land cover
maps of suciently ne spatio-temporal resolution over large areas
and long timespans. In this study, we developed an innovative auto-
mated classication algorithm that overcomes some of the tradi-
tional trade-os between ne spatio-temporal resolution and
extent to produce a multi-scene, 30m annual land cover time series
characterizing 21 years of land cover dynamics in the 35,000 km2
Greater Houston area. The ensemble algorithm takes advantage of
the synergistic value of employing all acceptable Landsat imagery in
a given year, using aggregate votes from the posterior predictive
distributions of multiple image composites to mitigate against mis-
classications in any one image, and ll gaps due to missing and
contaminated data, such as those from clouds and cloud shadows.
The procedure is fully automated, combining adaptive signature
generalization and spatio-temporal stabilization for consistency
across sensors and scenes. The land cover time series is validated
using independent, multi-temporal ne-resolution imagery, achiev-
ing crisp overall accuracies between 7886% and fuzzy overall
accuracies between 9194%. Validated maps and corresponding
areal cover estimates corroborate what census and economic data
from the Greater Houston area likewise indicate: rapid growth from
19972017, demonstrated by the conversion of 2,040 km
2
(± 400 km
2
) to developed land cover, 14% of which resulted from
the conversion of wetlands. Beyond its implications for urbanization
trends in Greater Houston, this study demonstrates the potential for
automated approaches to quantifying large extent, ne resolution
land cover change, as well as the added value of temporally-dense
time series for characterizing higher-order spatio-temporal dynamics
of land cover, including periodicity, abrupt transitions, and time lags
from underlying demographic and socio-economic trends.
ARTICLE HISTORY
Received 15 June 2018
Accepted 18 August 2018
CONTACT C.R. Hakkenberg ch55@rice.edu Department of Statistics, Rice University, Duncan Hall #2077,
Houston, TX 77251, USA
Supplementary data for this article can be accessed here.
INTERNATIONAL JOURNAL OF REMOTE SENSING
https://doi.org/10.1080/01431161.2018.1516318
© 2018 Informa UK Limited, trading as Taylor & Francis Group
1. Introduction
When Hurricane Harvey made landfall in Texas in August 2017, it resulted in the largest
rainfall event on record in the US, producing as much as 1200 mm of rain over a seven-
day period. The hurricane and subsequent ooding resulted in at least 89 deaths, 30,000
displaced people, and $125 billion dollars in damage its impact exacerbated as stalled
over one of the USs largest urban areas: Houston, TX (NOAA 2018). Over the past several
decades of rapid growth and development, Greater Houston has adopted a resistance-
based ood risk reduction strategy, relying on large-scale engineering solutions to
distribute the increased run-oassociated with its large-scale, largely-unzoned urban
development (Brody, Kim, and Gunn 2013). However, despite these infrastructural
improvements, ood vulnerability persists due in part to the vast expansion of low
intensity impervious land cover characteristic of sprawling urbanization (Jaret et al.
2009). Owing to the simultaneous expectation of higher frequency and stronger inten-
sity hurricanes in the region (Knutson et al. 2010; Emanuel 2017), studies are urgently
needed to investigate the independent and interactive aspects of global climate change
and local land cover conversion in contributing to storm damage across vulnerable
urban areas like Houston. Critical to this eort is the development of automated work-
ows for producing accurate and consistent land cover maps capable of characterizing
historical patterns and temporal trajectories of land cover change, as well as their
spatially-variant change rates at a suciently ne spatio-temporal resolution.
In this regard, the Landsat satellite data archive oers researchers an unparalleled
source of historical medium resolution optical imagery, enabling the compilation of
multi-decadal land-cover change trajectories temporal sequences of land-cover classes
derived from satellite images at multiple dates (Loveland and Dwyer 2012; Gómez,
White, and Wulder 2016). However, the production of annual land cover classications
over multiple Landsat scene extents and long timespans is complicated by a number of
factors including radiometric inconsistencies in reectance retrievals through space
(between neighbouring paths) and time (between sensors) (Vogelmann et al. 2016). In
addition, low acquisition frequency may result in irregular dates of usable imagery,
exacerbating dierences in scene conditions due to changes in land surface phenology,
atmospheric conditions, and illumination angles (C. Song et al. 2015; Song and
Woodcock 2003). These considerations have led some researchers to employ multi-
year imagery for classication surrounding a nominal year, resulting in a sparse time
series at a frequency on the order of 710 years (Sexton et al. 2013; Fenta et al. 2017)to
36 years (Dou and Chen 2017; Homer et al. 2015). And while the expanded temporal
window for input imagery often results in high quality map products, they may not be
precise enough to accurately reect land cover conditions for the nominal year and, as a
time series, may be too coarse to capture higher-order temporal dynamics critical to
assessing spatio-temporal complexities of humanenvironment systems (Jensen and
Cowen 1999; Lunetta et al. 2004). In response, recent studies have focused on a range
of data fusion, composite, and interpolation approaches to create land cover time series
at increasingly ne resolutions and large extents in the spatial and temporal domains
(Gong et al. 2013; Song et al. 2016; Li, Gong, and Liang 2015).
In this study, we present a multi-scene, annual land cover time series characterizing
21 years of land cover trends in the 35,000 km
2
Greater Houston area. The methodology
2C. R. HAKKENBERG ET AL.
employed is unique in that it entirely automates the image processing, data fusion, and
classication workow to produce a temporally dense and consistent land cover time
series using adaptive signature generalization, multi-scene compositing and ensemble
classication using all acceptable Landsat imagery in a given year, as well as spatio-
temporal stabilization for consistency across sensors and scenes. A distinct merit of this
study is that the proposed automated classication method overcomes some of the
traditional trade-os between spatio-temporal resolution and extent, with nal maps
possessing a ne resolution (annual, 30m) over a large duration and extent (21 years,
35,000 km
2
). The resulting map time series is compared with concurrent NLCD products,
and validated using multi-year, ne resolution independent reference imagery. Using
results from the probability-based sampling design of the accuracy assessment proce-
dure, we quantify the areal extent of land cover conversions, as well as change rates. As
a case study quantifying the rapid urbanization of Greater Houston, this research
demonstrates the potential for automated remote sensing workows to move beyond
bi-temporal change detection to characterize higher-order, annual spatio-temporal
dynamics of land cover change, including periodicity, abrupt transitions, and time lags
emerging from underlying demographic and socio-economic trends.
2. Materials and methods
2.1. Study area
The 35,000 km
2
study area consists of the 13 counties dening the Houston-Galveston
Area (HGAC 2018), namely: Austin, Brazoria, Chambers, Colorado, Fort Bend, Galveston,
Harris, Liberty, Matagorda, Montgomery, Walker, Waller, and Wharton counties
(Figure 1). Over the 21-year period, Greater Houston added 2.7 million residents, grow-
ing by 59% from a total population of 4.3 million in 1997 to 6.8 million in 2017 (U.S.
Census Bureau 2018). Greater Houston ranks as the fourth largest metropolitan area by
population in the United States (Wilson et al. 2012). Urban centres are primarily
restricted to Houston, Sugarland, and The Woodlands, which together house 88% of
the regions total population (U.S. Census Bureau 2018). Outlying counties in the rural-
urban interface consist largely of a network of interconnected towns and satellite
communities surrounded primarily by agriculture, pasture, forest, and grassland.
2.2. Remotely-sensed data
All classications were derived from Landsat satellite imagery spanning three satellite
missions the Landsat-5 Thematic Mapper (TM) for 19972011, the Landsat 7 Enhanced
Thematic Mapper Plus (ETM+) for 19992012, and the Landsat 8 Operational Land
Imager (OLI) for 20132017 and four Landsat World Reference System 2 (WRS-2)
scenes: path/row 25/39, 25/40, 26/39, and 29/40 (Figure 1). All input imagery consists
of radiometrically-calibrated and orthorectied Landsat Collection 1 Level-1 products
conforming to prescribed criteria for < 10% cloud cover and possessing at least three
phenological states per year: leaf-o(DOY 30160), early growing season (61180), and
late growing season (DOY 181300) (Appendix 1). Imagery was constrained to the
calendar year of interest to ensure temporal precision in time series change detection,
INTERNATIONAL JOURNAL OF REMOTE SENSING 3
and thereby precludes other commonly used predictor layers (e.g. DEMs) unavailable on
an annual basis. Only pixels with high condence in quality, as designated in corre-
sponding Quality Assessment bands, were retained. For those years possessing sparse
cloud-free imagery, lacking an acceptable range of acquisition dates, or otherwise
heavily impacted by ETM+ Scan Line Corrector (SLC-o) data gaps, the cloud-cover
and DOY criteria were relaxed. Given these constraints, a total of 262 Landsat scenes
were used for the 21-year time series (Figure 2).
Training data for all classications come from the U.S. Geological Survey (USGS)
National Land Cover Database (NLCD) from 2001 (Homer et al. 2007), 2006 (Fry et al.
2011), and 2011 (Homer et al. 2015). Owing to trade-os between classication accuracy
and thematic precision, cover types were simplied to focus more acutely on urbaniza-
tion trends (hereafter dened as land cover conversion from a non-Developed to a
Developed class) rather than subtle ecological transitions such as wetland delineation,
otherwise beyond the scope of the current study. Therefore, vegetation classes adopted
from the NLCDs Anderson Level 2 typology were bifurcated into woody and non-woody
vegetation whereby deciduous forest, evergreen forest, mixed forest, shrub/scrub, and
woody wetlands were combined as Forest, while grassland/herbaceous, emergent
herbaceous wetlands, and pasture/hay were merged as Grassland/Pasture. All other
classes occurring in the study area, as dened by the NLCD, were retained (Table 1). All
Landsat images and NLCD classied maps were reprojected from their native coordinate
system to a shared State Plane coordinate system, clipped to the 13-county study area,
and buered outward by 90m (~ 3 pixels) on all sides to mitigate against edge eects in
spatial ltering.
Validation imagery consists of 30 ne-resolution images from the IKONOS, Quickbird,
and Worldview-2 satellite sensors (©2018, DigitalGlobe; NextView License) and two
Figure 1. Greater Houston study area. (a) Study area extent (light grey) and four Landsat scene
footprints with path/row designation (black outlines) superimposed on maps of the US and Texas;
(b) County map with validation imagery extents (coloured by year) and validation samples (points).
4C. R. HAKKENBERG ET AL.
airborne platforms: Andrew Lonnie Sikes and Houston Galveston Area Council aerial
imagery (Kinder Institute 2018)(Figures 1(b) and 2; Appendix 2).
2.3. Class membership probabilities
Preliminary posterior class membership probabilities were derived from Landsat imagery
based a three-step process: (1) image and band compositing using principal compo-
nents analysis (PCA), (2) automatic adaptive signature generalization (AASG) (Gray and
Song 2013; Dannenberg, Hakkenberg, and Song 2016), and (3) random forest (RF)
supervised classication (Breiman 2001)(Figure 3). Prior to classication, all Landsat
bands in a given image (6 bands, excluding thermal and ne-resolution panchromatic
bands) were reduced to their rst three PCA axes (PCA3) for computational eciency.
Concurrently, all images in a given year (i.e. 37 images times 6 bands per image = 18
Figure 2. Distribution of Landsat scenes and ne resolution validation imagery. 262 Landsat scenes
in total. Thirty validation images depicted by sensor abbreviation (ALS Andrew Lonnie Sikes and
HGA Houston Galveston Area Council aerial imagery; as well as IK IKONOS, QB Quickbird, and WV
Worldview-2 satellite imagery.
Table 1. Land cover class NLCD comparison.
Cover class Corresponding NLCD class (code)
Barren/Sand Barren Land Rock/Sand/Clay (31)
Developed-Open Developed, Open Space (21)
Developed-Low Developed, Low Intensity (22)
Developed-Medium Developed, Medium Intensity (23)
Developed-High Developed, High Intensity (24)
Cultivated Crops Cultivated Crops (82)
Grassland/Pasture Grassland/Herbaceous (71)
Emergent herbaceous wetlands (95)
Pasture/Hay (81)
Forest Deciduous Forest (41)
Evergreen Forest (42)
Mixed Forest (43)
Shrub/Scrub (52)
Woody Wetlands (90)
Water Open Water (11)
INTERNATIONAL JOURNAL OF REMOTE SENSING 5
42 raw bands) were reduced to their rst 10 PCA axes (PCA10), which represent > 99% of
total variation in each annual image stack.
Next, to streamline the otherwise inconsistent and labour-intensive process of select-
ing training and predictor data in spatially-coincident multi-temporal image stacks, we
employed the AASG algorithm. AASG rst delineates stable (no-change) sites between
images, dened as core areas within a scene whose cover class designation remains
unchanged between the date of a reference image (I
R
) and a target image (I
T
). Stable
sites are algorithmically determined by rst selecting pixels within a pre-dened dis-
tance (c) from the mean (μ) of the image dierence histogram (I), where:
ΔI¼IR";1½$%IT";1½$ (1)
Landsat 5 TM (1997-2011)
Landsat image stack (262 images): Landsat 7 ETM+ (1999-2012)
Landsat 8 OLI (2013-2017)
Gap mask
(a)
CR: NLCD
(2001, 2006, 2011)
IT: Annual PCA10
(4 scenes x 21 years)
IT:Image PCA3
(4 scenes x 21 years)
IR: Annual PCA10
(2001, 2006, 2011)
IR: Image PCA3
(2001, 2006, 2011)
AASG - RF (1) AASG - RF (2)
(b)
Class membership posterior
predictive distributions of annual
PCA10 stack (with gaps)
(9 classes x 21 years)
Class membership posterior
predictive distributions all single-date
PCA3 composites (no gaps)
(9 classes x 262 images)
(c)
Annual ensemble class membership probabilities (no gaps)
(9 classes x 21 years)
Gap-filling
Ensemble
classification
Scene
mosaicking
Spatio-temporal
contextual filtering
(d)
CT: Annual classification
(1997-2017)
Estimating
Developed - Open
Figure 3. Methods owchart. (a) Input imagery and cloud/shadow/SLC-omasking; (b) Model
training and prediction, generating class membership posterior distributions; (c) Annual ensemble
classication, including gap-lling and scene mosaicking; (d) Spatio-temporal ltering and derivation
of nal land cover time series. Inputs (yellow); process (blue); outputs (green).
6C. R. HAKKENBERG ET AL.
such that, in this case, [·,1] corresponds to the rst PCA axis derived from all spectral
bands. Stable sites are selected from within the interval:
μΔI&ck'σΔI(2)
where μ
I
and σ
I
is the mean and standard deviation of I, respectively, and c
k
is a class-
specic threshold parameter for each class k. Candidate stable sites are additionally
subjected to a class-specic spatial erode lter to mitigate against errors arising due to
image misregistration and edge eects along class boundaries. Once delineated, scene-
specic spectral signatures can be sampled from stable sites in both I
R
and I
T
images,
and subsequently combined with a reference classication (C
R
) corresponding to the
date of the I
R
for model training and prediction. By adapting to the unique atmospheric,
radiometric, and phenological characteristics of each image, the AASG procedure facil-
itates automated image ingestion and classication processes that require neither atmo-
spheric correction nor data normalization, while maintaining semantic consistency in
class denitions between the reference and target classication (C
T
) (Song et al. 2001).
As an automated training and predictor data selection algorithm, AASG is agnostic
to the choice of classier. We ultimately selected RF, an ensemble of classication
trees based on votes across bootstrap replicates, for its computational eciency and
its record of high performance in terms of predictive accuracy and generalizability
(Belgiu and Drăgu 2016). The nonparametric RF algorithm produces highly accurate
and unbiased predictions that eciently handles highly collinear neighbouring pre-
dictor pixels in each stable site, is robust to noise, and largely immune to over-tting
of interest due to the requirement that identical training data generalize to so
many dierent target images in the Landsat stack (Gislason, Benediktsson, and
Sveinsson 2006).
RF classication models for each scene/year were parameterized with 200 trees per
model, with 3 predictors sampled at each split using training data from AASG-dened
stable sites (Maxwell, Warner, and Fang 2018). Each training class was proportional to
the relative abundance of each reference class and capped at 100,000 pixels per class
(Chen, Liaw, and Breiman 2004). The three NLCD reference classications (C
R-2001
,
C
R-2006
,andC
R-2011
) were paired with reference imagery from each respective year
(I
R-2001
,I
R-2006
,andI
R-2011
), and applied to the most temporally-proximate target
imagery for all 21 years (i.e. I
R-2001
and C
R-2001
corresponds with I
T-1997
I
T-2003
,while
I
R-2006
and C
R-2006
was paired with I
T-2004
I
T-2008
,andI
R-2011
and C
R-2011
with I
T-2009
I
T-
2017
). Raw predictions, in the form of posterior membership probabilities (p) for each
class (i), are based on the distribution of votesfrom the ensemble of classication
trees in the RF classier, such that:
X
k
i¼1
pi¼1 (3)
for kclasses per pixel (Wang et al. 2015). All RF models were run using the randomForest
package (Liaw and Wiener 2002) and derived products and analyses were calculated
using the raster package (Hijmans 2017) in the software R, v. 3.3.1 (R Core Team 2017).
INTERNATIONAL JOURNAL OF REMOTE SENSING 7
2.4. Annual ensemble classication
Annual PCA10 composites (see Section 2.3), which incorporate spectral data from multi-
ple images across three phenological states within a given calendar year, serve as the
primary predictor in all classications (Figure 4(a)). Owing to data gaps in the PCA10
predictor set which represent the superset of all algorithmically delineated clouds and
cloud shadows (Zhu, Wang, and Woodcock 2015) as well as ETM+ SLC-ogaps and
radiometrically-saturated or contaminated pixels identied in quality assessment bands
(Figure 4(b)) a parallel classication was simultaneously conducted on the PCA3
composite from each single image. Specically, AASG-RF was implemented on each
PCA3 in the annual stack and used to generate per-pixel posterior predictive distribu-
tions for each class (Figure 4(c-d)). From these posteriors, an ensemble prediction
was derived from the geometric mean of the set of 37 posterior classication prob-
abilities in a given year and used as the basis for designating pixelsclass membership
(Figure 4(e)). These classied pixels were then used to ll data gaps in the original PCA10
classication (Figure 4(f)). Unlike gap-lling algorithms that interpolate pixel values
before classication, this two-part classication procedure ensures all classications are
derived from original reectance values, thereby retaining pixel-level spatial consistency
(Yin et al. 2017). This ensemble classication approach utilizes the added information
content of the full stack of all acceptable imagery in a calendar year to mitigate the
potential for contagion or classication error of any one image, as well as inter-image
pixel misalignment due to discrepancies in georegistration. Because PCA3s are only
Figure 4. Multi-date classication procedure. (a) Annual PCA10, with ETM+ SLC-oand cloud/
shadows masked (black); (b) PCA10 classication with data gaps (black); (c-e) single-date PCA3
image composites with data gaps (black); (f) classication of PCA10 gaps based on annually-
aggregated, mean membership probabilities of all PCA3 classications; (g) gap-lled classication
(combining panels b and e). Bounding box corresponds with Figure 6(a), box 2.
8C. R. HAKKENBERG ET AL.
impacted by data gaps resulting from stochastic phenomena (e.g. cloud location) in any
one image, overlap in missing data pixels for all multi-temporal images in a given year is
extremely rare, and can be interpolated during temporal stabilization (See Section 2.4.1).
To ensure a seamless transition between neighbouring scenes, mean classication
probabilities in the 24 overlapping scene edge areas were used to replace those
produced for each scene.
2.5. Spatio-temporal ltering
2.5.1. Spatialtemporal contextual ltering
To mitigate against error propagation due to misclassication and ensure consistency in
automated time series classications, we adopted a spatiotemporal contextual ltering
approach that exploits two statistical properties of the classied time series namely,
spatial autocorrelation and temporal dependence to identify potential spurious classi-
cations and adjust them accordingly (Lu and Weng 2007;Lietal.2014). Contextual
lters exploit information between a target pixel and neighbouring pixels within spatial
and temporal windows of varying size to impose constraints on the nal classication of
the target pixel. Contextual ltering consisted of three steps: (1) temporal smoothing, (2)
spatial ltering, and (3) label modication for illogical temporal transitions.
For temporal stabilization of classication probabilities, especially where class prob-
abilities exhibit pronounced peaks and troughs in the temporal domain, we applied a
temporal low pass lter using a Gaussian kernel in a ve year window (Hamilton 2015).
Spatially-weighted kernel lters were then applied to each classication in the time
series to remove spurious spatial heterogeneity (e.g. salt-and-pepper) in otherwise
homogeneous land cover patches. In addition to spatial kernel lters, a minimum
mapping unit (MMU) criteria was applied following Homer et al. (2015), whereby a 5-
pixel MMU was required for all classes except Cultivated Crops (which required a 12-
pixel MMU) and Developed classes, which were not subjected to the MMU requirement.
Lastly, a rule-based label adjustment procedure was used to eliminate illogical temporal
transitions in the time series identied when the class of maximum posterior probability
exhibits pronounced uctuations within a short time period (Wang et al. 2015; Zhang
and Weng 2016). For example, for cover classes exhibiting relatively discrete spatial
boundaries (e.g. the four Developed classes), a three-year temporal window (t1, t,
t+ 1) was employed such that the classication at time twas modied to that for time
t-1, when t-1 = t+ 1 and tt-1 (Pouliot et al. 2014;He, Lee, and Warner 2017). For land
cover classes exhibiting more continuous temporal variation in land surface properties
(e.g. Grassland/Pasture and Forest) a more conservative ve-year temporal lter (t-2:
t+ 2) was employed to distinguish long-term (genuine) trends from short-term (spur-
ious) uctuations (Cai et al. 2014).
2.5.2. Special consideration for the Developed-Open class
Following NLCD denitions, the four Developed classes Open, Low, Medium, and High
Intensity are dened by impervious surface fractional covers of 020%, 2049%,
5079%, and 80100%, respectively (Appendix 3). Of particular concern for the current
study is the characterization of Developed-Open pixels that, being dened as < 20%
impervious cover, would otherwise possess the spectral characteristics of the
INTERNATIONAL JOURNAL OF REMOTE SENSING 9
predominant fractional cover class, such as water or vegetation. Because the Developed-
Open class is dened by an impervious fractional cover far below 50%, its delineation in
the NLCD protocol requires additional non-spectral data unavailable at annual time
scales, as well as manual boundary delineation (Jon Dewitz, personal communication,
24 January 2018). And while the results of this resource-intensive process are highly
satisfactory, the approach is neither reproducible nor feasible for automated classica-
tion at an annual scale. In response, several studies have simply eschewed classifying the
Developed Open class altogether (Sexton et al. 2013; Dannenberg, Song, and
Hakkenberg 2018). However, as a central component of the low density, sprawling
development characteristic of Greater Houston, as well as its disproportionate impact
on urban ood risk, mitigation, and planning, we deemed it necessary to include a
spectrally-determined, high-delity proxy for the Developed Open class (Brody, Kim,
and Gunn 2013).
We therefore approximated the Developed Open class as all pixels falling within a
aggregated urban extent, that otherwise do not possess the fractional impervious cover
proportions dening the three higher-intensity Low, Medium, and High Developed
classes. To do this, posterior probabilities from raw classications were assessed to
identify pixels whose RF modal vote prediction falls within one of the four Developed
classes (Figure 5(b-c)). Because this spectrally-determined impervious layer fails to
capture the full extent of the NLCDs Developed classes (including the partly manu-
ally-determined < 20% impervious Developed Open) especially where vegetated yards
or overhanging tree canopies in suburban areas were misclassied as vegetation, we
applied a 3 ×3 and 5 ×5 anisotropic spatial lter to the classied output to identify
interstitial and edge pixels that should be included within the urban base map
(Figure 5(d)). Given this urban extent, the three higher-intensity Low, Medium, and
High Intensity impervious classes (Figure 5(e)) were superimposed within the urban
extent (Figure 5(f)), such that all remaining pixels are classied as Developed Open.
Approximated urban extents show a strong resemblance to concurrent NLCD maps
(Figure 5(g)) with F-scores, representing the harmonic mean of the users and producers
accuracies between the two urban extents, achieving values of 0.894, 0.887, 0.885 for
2001, 2006 and 2011, respectively (Appendix 4). Thereafter, the urban mask was updated
annually in a manner consistent with other temporal ltering processes in addition to
one illogical transition criterion based on an irreversibility assumption adopted from Gao
et al. (2012): once a pixel is classied as one of the Developed categories for a minimum
of three consecutive years, it is suciently unlikely to be converted again in the study
time period. Supporting this assumption in other studies, there were zero pixels that
transitioned from a Developed to a non-Developed category in the NLCD map of the
study area from 2001 to 2011 (Homer et al. 2007,2015)aresult similarly observed in
Washington DC by Sexton et al. (2013).
2.6. Accuracy Assessment
Maps were assessed for accuracy by comparison with coincident NLCD maps and via a
three-part validation procedure with independent, multi-temporal, ne resolution imagery
based on a sampling and response design modied from Olofsson et al. (2014): (1) full
class, crisp accuracy assessment; (2) reduced class, crisp accuracy assessment; and (3) fuzzy
10 C. R. HAKKENBERG ET AL.
accuracy assessment. First, classied maps were compared with spatially, temporally, and
thematically coincident NLCD maps to assess overall agreement (O
AG
), producers agree-
ment (P
AG
), and users agreement (U
AG
) for the three nominal years where the two
products overlap (e.g. 2001, 2006, and 2011) (Congalton 1991).
Second, we conducted a multi-temporal independent accuracy assessment using a
stratied random sampling design whereby samples corresponding to the nominal
resolution of classied maps were established in advance in 30 ne-resolution (3m)
satellite and aerial images (Appendix 2). Validation imagery is adequately distributed in
space (as measured by correspondence in total areal cover by class in the full study area
extent versus that for reference imagery only) and time (14 of 21 years) throughout the
study area (Figure 1(b); Appendix 5). Total sample size (n= 3036) across the 14 reference
dates was determined by a priori expectations for the average standard error in the
overall agreement with the three NLCD products, and adjusted upwards to account for
rare classes (Olofsson et al. 2014). All samples were allocated proportionally by cover
class strata and across reference imagery by year and spatial extent. Specic sample
locations were determined independently from AASG stable sites and, given stratica-
tion constraints, randomized.
Thereafter, trained technicians conducted a blind interpretation of land cover within
the areal extent of each sample pixel, allowing for mixed pixels and other ambiguities by
assigning proportional membership when class identity was not otherwise unambigu-
ous (e.g. membership score p
k
1). To ensure a monotonic ranking, no two membership
probabilities were equal. Due to the possibility of interpretation error and inconsistency,
Figure 5. Estimation of urban extent. (a) Fine resolution aerial reference imagery; (b) RF posterior
probability of combined Developed classes from 2012; (c) raw urban extent derived from modal
posterior probabilities; (d) spatially-ltered urban extent; (e) classied pixels in the Developed Low,
Medium, and High classes; (f) nal classication; (g) coincident and concurrent NLCD classication.
Bounding box corresponds with Figure 6(a), box 2.
INTERNATIONAL JOURNAL OF REMOTE SENSING 11
all samples were classied by more than one technician, with disagreement in the class
of maximum probability leading to secondary expert review. Thereupon, accuracy
assessment results follow standard protocols for reporting overall accuracy (OA), users
accuracies (UA), and producers accuracies (PA), with all corresponding 95% condence
intervals (CIs) based on the area-weighted population error matrix (Olofsson et al. 2014;
Foody 2002). As a single statistic for classication accuracy, the area-weighted overall
accuracy was favoured to alternative approaches like the kappa coecient (Pontius and
Millones 2011).
Owing to the relatively coarse resolution of Landsat imagery in relation to end-
member fractional cover, as well the inherent subjectivities in assigning a single crisp
class membership in reference imagery, crisp accuracy may have limited utility, espe-
cially for highly heterogeneous urban land cover (Foody 2002). The uncertainty and
ambiguity inherent in crisp accuracy assessments is non-trivial and especially apparent
in Developed mixed pixels which, despite existing on a continuum of surface imper-
viousness, are binned into discrete categories. Added to this uncertainty is the lack of
condence in the consistency of reference labels based on techniciansvisual estimate
of surface imperviousness. We therefore implemented a fuzzy accuracy assessment
based on a translation of visually determined membership probabilities, using a three-
level linguistic-measurement scale to characterize the magnitude of membership
probability, with the highest single class probability dened as absolutely right
(Right), the second highest as a good answer (Good), and all other non-zero prob-
abilities (maximum of two) assigned as reasonable or acceptable (Acceptable)
(Woodcock and Gopal 2000;Foody2002). Because the Rightcategory is, by denition,
equivalent to crisp UAs in the area-weighted population error matrix, we limit results
to the Goodand Acceptablecategories.
2.7. Cover class area estimation
Annual class area estimates were determined based on a stratied estimator of areal
proportions derived from independent reference imagery. Accepting that the accu-
racy assessment sampling design yielded estimates with relatively small standard
errors, as well as the premise that the quality of the independent reference imagery
is superior to that of the map classication, class areas can be estimated by multi-
plying area proportions derived from the population error matrix of the independent
reference imagery (i.e. column totals of the contingency table) by the total map area
(Stehman 2013). This sampling design likewise allows for the estimation of unbiased
standard errors for each class area (Olofsson et al. 2014). For simplicity, the con-
tingency table used for area estimates was constrained to single, crisp membership
consisting of the highest probability class among all independent samples. While the
derivation of area estimation parameters from an error matrix populated with crisp
set memberships tentatively assumes mutually exclusive and collectively exhaustive
categories at odds with fuzzy logic, it simultaneously allows class areas to sum to
one, and thereby better facilitates consistent inter-annual comparisons of class areas
(Woodcock and Gopal 2000).
12 C. R. HAKKENBERG ET AL.
3. Results
3.1. Annual land cover time series
Classied maps for the 21-year time series (Figure 6)showstrongvisualdelity to
known land cover patterns and demonstrate the expansive scale of the Greater
Houston region, which in the absence of signicant topographic constraints assumes
a symmetrical, hub and spoke urban form (Galster et al. 2001). As such, developed
areas are tightly clustered in the urban core, while sprawling suburbs expand out-
wards in all directions along major transportation corridors and emerging satellite
communities populate the urban periphery where they have leap-frogged non-
Developed classes (Jaret et al. 2009). The outer periphery consists largely of
Cultivated Crops, Grassland/Pasture, and Forest cover types, within which older
ranching and agricultural settlements and communities are scattered. Zoomed sub-
sets of the study area demonstrate the capacity for characterizing the texture of
intergrading impervious surfaces across an urban density gradient (Figure 7).
3.2. Classication accuracies
While dierences exist between this land cover time series and coincident NLCD maps (e.g.
thematic categories), they nonetheless still demonstrate a substantial degree of agreement
(Table 2). The largest disparities between the two products occur with Barren/Sand and the
four Developed classes, while natural and semi-natural classes (e.g. Forest, Grassland/Pasture,
and Water) exhibit close agreement for concurrent dates, on the order of 7399%. Based on a
random stratied sampling design with multi-date ne-resolution images, we found the full
2
Figure 6. Land cover classications of Greater Houston (2017). (a) HGA study area; (b) Houston (box
1). Bounding box 2 (Figures 4 and 5); boxes 35(Figure 7); boxes 67(Figure 11).
INTERNATIONAL JOURNAL OF REMOTE SENSING 13
nine-class maps to achieve an overall accuracy of 78% (± 1.5%), with usersaccuracieslowest
for the Developed classes, mostly due to confusion among the dierent intensities of
Developed land rather than confusion with non-Developed cover types (Table 3;Appendix
6). Accordingly, with all Developed classes merged, overall accuracy reaches 86% (± 1.4%)
(Table 4). Fuzzy accuracy assessment results demonstrate a 90.6% (± 1.5%) accuracy for good
matches and 94.2% (± 1%) accuracy for acceptableagreement (Table 5).
3.3. Greater Houston land cover change area estimates
Unbiased cover class areas were estimated from areal proportions in the the population
error matrix (Table 3). The largest land cover changes observed in the study area
occurred in the Developed classes, especially the Developed Medium category,
which grew by 62% over the 21-year period (2.3% compound annual) and the
Developed High class, with 52% total growth (2.0% compound annual) observed
(Figure 8). In total, combined Developed classes witnessed an increase of
Table 2. Agreement with NLCD maps for 2001, 2006, and 2011 (%). U
AG
users agreement;
P
AG
producers agreement; O
AG
overall agreement. NLCD as reference.
2001 2006 2011
U
AG
(%) P
AG
(%) U
AG
(%) P
AG
(%) U
AG
(%) P
AG
(%)
Barren/Sand 71.0 30.5 63.9 28.9 67.6 29.6
Developed-Open 58.1 76.9 53.9 74.9 52.0 74.4
Developed-Low 48.1 43.6 44.0 48.0 43.2 49.4
Developed-Medium 60.7 45.6 57.3 51.4 60.0 50.7
Developed-High 58.0 71.0 54.7 73.7 56.4 73.1
Cultivated Crops 80.0 76.7 80.1 76.3 80.1 76.2
Grassland/Pasture 75.3 81.5 75.8 79.9 76.1 78.9
Forest 86.8 76.9 88.2 75.1 87.1 75.2
Water 84.4 99.5 84.9 99.4 85.7 97.7
O
AG
75.7 74.9 74.3
Figure 7. Zoomed urban classication insets. Fine resolution aerial reference imagery (a-c) and
corresponding classications (d-f) corresponding to 2017, with increasing levels of urbanization
(from light to dark red). Specic locations correspond to bounding boxes in Figure 6(a): (a, d) box 3;
(b, e) box 4; (c, f) box 5. Classication coloration is consistent with legends in Figure 46.
14 C. R. HAKKENBERG ET AL.
Table 3. Area-weighted confusion matrix (full). UA users accuracy; PA producers accuracy; OA overall accuracy. Accuracies are listed as proportions of the
total study area, followed by 95% condence intervals.
Reference
Barren/Sand
(%)
Developed
-Open (%)
Developed
-Low (%)
Developed
-Med (%)
Developed
-High (%)
Cultivated
Crops (%)
Grassland/
Pasture (%) Forest (%) Water (%) UA (%)
Map Barren/Sand 0.2 0 0 0 0 0 0 0 0 96.1 ± 4.4
Developed-Open 0 3.6 2.9 0.7 0.3 0 0.5 0.3 0 42.2 ± 5.6
Developed-Low 0 0.4 2.7 1.4 0.3 0 0 0 0 54.8 ± 5.6
Developed-Med 0 0.1 0.6 1.7 0.6 0 0 0 0 56.2 ± 5.1
Developed-High 0 0 0.1 0.4 1.7 0 0 0 0 76.5 ± 4.6
Cultivated Crops 0 0.1 0 0 0 8.2 3.0 0.3 0 69.2 ± 7.2
Grassland/Pasture 0 1.7 0.5 0 0 1.3 29.5 1.1 0.2 85.7 ± 2.5
Forest 0 1.2 0.4 0 0 0 1.7 20.6 0.2 85.4 ± 3.0
Water 0.1 0 0 0 0 0 0.5 0.1 9.7 90.7 ± 3.9
PA 45.8 ± 11.4 50.1 ± 33.4 36.4 ± 23.8 40.2 ± 21.6 56.9 ± 14.9 85.6 ± 12.8 83.3 ± 4.1 91.9 ± 2.6 94.7 ± 2.1 OA
78.0 ± 1.5
INTERNATIONAL JOURNAL OF REMOTE SENSING 15
2040 km
2
± 400 km
2
(Figure 9), with the Low, Medium, and High Intensity Developed
classes accounting for 41%, 34%, and 21% of that growth, respectively. The remaining
4% of the change is attributable to expansion of the Developed Open class. While the
higher-intensity Developed classes experienced the largest rates of change over the 21-
year period, developed cover in the study area was still dominated by the low-intensity,
spatially-dispersed urban morphology of the Developed Open (33% of total developed
area) and Developed Low (34% of total developed area) categories. Growth in
Developed classes was largely oset by declines of 4.3% and 15.6% (0.2% and
0.8% compound annual) in the Grassland/Pasture and Forest classes, respectively. In
total, Forest cover decreased by 1350 km
2
(± 460 km
2
), while Cultivated Crops and
Grassland/Pasture experienced nonsignicant declines of 100 km
2
(± 490 km
2
) and
550 km
2
(± 670 km
2
), respectively (Figure 9).
4. Discussion
4.1. Land cover change trends in Greater Houston
Areal change maps corroborate what census data likewise indicate: rapid growth in the
13-county region over the past two decades, whereby an estimated 59% growth in
population corresponds to a 30.3% (± 3.3%) increase in Developed cover (U.S. Census
Table 4. Area-weighted confusion matrix (reduced). Developed classes combined. UA users
accuracy; PA producers accuracy; OA overall accuracy. Accuracies are listed as proportions of
the total study area, followed by 95% condence intervals.
Reference
Barren/
Sand (%)
Developed
-combined
(%)
Cultivated
Crops (%)
Grassland/
Pasture
(%) Forest (%) Water (%) UA (%)
Map Barren/Sand 0.2 0 0 0 0 0 96.1 ± 4.4
Developed-combined 0 18.0 0 0.4 0.1 0 96.2 ± 1.0
Cultivated Crops 0 0.2 8.2 3.0 0.3 0 69.2 ± 7.2
Grassland/Pasture 0 2.3 1.3 29.5 1.1 0.002 85.7 ± 2.5
Forest 0 1.6 0 1.7 20.6 0.002 85.4 ± 3.0
Water 0.1 0.2 0 0.5 0.1 0.097 90.7 ± 4.5
PA 45.2 ± 11.4 80.8 ± 10.6 85.8 ± 12.7 83.9 ± 4.0 92.4 ± 2.4 94.9 ± 2.0 OA
86.2 ± 1.4
Table 5. Fuzzy accuracy assessment. UA users accuracy; OA overall accuracy, followed by 95%
condence intervals. Fuzzy linguistic scale following Woodcock and Gopal (2000): good answer
(Good) and reasonable or acceptable (Acceptable).
Reference
UA Good(%) UA Acceptable(%)
Map Barren/Sand 96.3 ± 4.2 97.5 ± 3.4
Developed-Open 71.8 ± 5.6 86.5 ± 4.4
Developed-Low 83.0 ± 3.8 97.2 ± 1.7
Developed-Med 83.8 ± 3.8 95.3 ± 2.1
Developed-High 92.5 ± 2.9 96.2 ± 2.0
Cultivated Crops 85.9 ± 5.5 87.8 ± 5.1
Grassland/Pasture 95.1 ± 1.6 96.4 ± 1.3
Forest 93.0 ± 2.2 95.2 ± 1.8
Water 95.8 ± 2.7 95.8 ± 2.7
OA 90.6 ± 1.1 94.2 ± 1.0
16 C. R. HAKKENBERG ET AL.
Bureau 2018). Despite Houstons ranking as among the most sprawling large American
cities as measured by density and nuclearity (Galster et al. 2001), that the rate of
urbanization is half that of population growth reects some degree of densication,
however modest. Notably, the largest proportional land cover changes observed in the
study area occurred in the higher density Developed classes, especially the Developed
Medium and DevelopedHigh categories, which grew by 62.1% (± 9.8%) and 51.8%
(± 9.4%), respectively, versus the lower density DevelopedOpen and DevelopedLow
categories, which grew by 38.8% (± 9.1%) and 39.6% (± 8.5%), respectively. This nding
of increased high-density growth largely corroborates the conclusions of other studies
which nd land availability constraints and growing commute times in Houston to be
primary factors driving inll and multi-story developments (Brody, Kim, and Gunn 2013).
Increases in Developed cover were mostly oset by 4.3% (± 2.6%) and 15.6% (± 2.7%)
declines in the Grassland/Pasture and Forest categories, respectively. The disparity in the
magnitude of the positive versus negative change rates in the zero-sum game of land
Figure 8. Greater Houston land cover change rates. Barren/sand and water are excluded. Linear
regression line added for reference. Error bars based on standard error.
1000
0
1000
2000
Developed Crops Grass/Pasture Forest
Landcover class
Total change (km2)
Figure 9. Greater Houston land cover change totals between 19972017. Total change and standard
error for the four largest land cover change classes.
INTERNATIONAL JOURNAL OF REMOTE SENSING 17
cover conversion is explained by considering the vastly larger total area of the
Grassland/Pasture and Forest classes in the study area versus the Developed categories.
The relatively smaller decline in the Grassland/Pasture category compared with Forest
cover is due to the far greater frequency of Forest conversion to Grassland/Pasture (e.g.
deforestation) compared with the reverse trend (e.g. aorestation/reforestation). Based
on a comparison with the National Oceanic and Atmospheric Administrations Coastal
Change Analysis Programs land cover product, 14% of all land cover urbanized between
19972017 was classied as wetlands prior to conversion (NOAA C-CAP 2011). The
magnitude of wetland conversion over the past two decades in the Greater Houston
area has important implications for wetlands ecological conservation, storm water
management, and ood hazards (Bullock and Acreman 2003).
While bi-temporal change detection provides an estimate of net land cover change
with associated uncertainties, it is insucient for characterizing spatially-variant change
trajectories as well as temporal dynamics of urbanization morphologies (Yu and Zhou
2017). Multi-temporal classications, on the other hand, are capable of detecting higher-
order dynamics of land cover change (Li, Gong, and Liang 2015; Song et al. 2016).
Periodic uctuations in land cover growth trajectories are evident in the land cover time
series, with the timing and magnitude of acceleration in the growth of urbanization
mirroring the periodicity observed in socio-economic indicators including total popula-
tion, Greater Houstons Gross Domestic Product (GDP), and the Harris County Housing
Price Index (HPI) (Figure 10). Over the 21-year period, the rate of urbanization peaked
between 20052007, followed by a considerable reduction relative to baseline growth
after the start of the Great Recessionin the United States in late 2007. Interestingly, the
timing of satellite-observable development is temporally oset from the underlying
socio-economic forces partly driving it, shedding light on the magnitude of the temporal
lag between the two related trends.
The spatial imprint of temporal processes of urbanization is particularly visible in
change year maps. Using the example of The Woodlands and Cinco Rancho large-scale
developments, we observe that while both exhibit some similar growth characteristics
(e.g. expansion from an initial seed area), their growth morphologies are in fact quite
dierent (Figure 11). For example, the stringently-zoned western extension of The
Woodlands expands within a constrained area bounded by green space to the north,
west, and south. The largely unzoned Cinco Rancho, on the other hand, expands out-
ward in all directions, largely undeterred by zooming, topography, or hydrology in the
process of converting former agricultural land to large-scale residential developments
(Qian 2010). These individual developments exemplify the scale and pace of urbaniza-
tion in the Greater Houston area, with the former area adding 27 km
2
(6.6% compound
annual) in Developed cover over the 21-year period, while the latter added 115 km
2
(5.7% compound annual).
4.2. Classication accuracy
While no one statistic is singularly authoritative in validating the accuracy of dense land
cover time series, the use of multiple assessments helps to clarify spurious or misleading
confusion in the crisp classications, while simultaneously providing a more robust
ceiling for actual (rather than sampled) map accuracy. Among all classes, crisp
18 C. R. HAKKENBERG ET AL.
classications of Developed cover exhibited the lowest per-class accuracies owing
largely to the subjectivity inherent in techniciansassignment of a single imperviousness
value to the spatially complex, multi-endmember impervious cover types (Wang, Huang,
and De Colstoun 2017; Weng 2012). Per-class agreement with the NLCD was likewise
relatively low for these four Developed classes, though when combined into a single
Developed class, users accuracies achieve 96% overall accuracy and 89% agreement
with the NLCD. It should be stressed that inference of accuracy from a test of agreement
is problematic owing to the lack of an unambiguous reference map (errors exist in both
products). Fuzzy accuracy assessments largely compensate for these misleadingly low
Figure 10. Land cover and socio-economic trends in the HGA. Standardized residuals from the slope
(β1) of a linear regression of urbanization, Greater Houstons Gross Domestic Product (GDP), Harris
County House Price Index (HPI), and population. Land cover points and standard error bars represent
class-specic areal estimates. The land cover trend line is represented by a loess function, plus 95%
condence interval. Socio-economic data from U.S. Bureau of Economic Analysis (2018) and U.S.
Census Bureau (2018).
(a) (b) (c)
(d) (e) (f)
Figure 11. Change year maps for large-scale developments. The Woodlands, corresponding with
bounding box 6 in Figure 6(a) (a-c) and Cinco Ranch, corresponding with bounding box 7 in Figure 6(a)
(d-f). Classication coloration is consistent with legends in Figure 46, with darker reds indicating higher
proportions of impervious surface.
INTERNATIONAL JOURNAL OF REMOTE SENSING 19
accuracies for the four distinct Developed classes, albeit at the expense of thematic
precision, reaching 9094% overall accuracy (Woodcock and Gopal 2000).
The Cultivated crops and Grassland/Pasture classes exhibited signicant confusion,
partly owing to the ambiguity regarding the taxonomic identity of vegetation in the two
classes, as well as uncertainty in labelling samples in reference imagery. Confusion was
likewise observed in the Barren/Sand and Cultivated Crop classes, which both tend to
exhibit high reectance values that may be easily confused with impervious surfaces
(Wang, Huang, and De Colstoun 2017; Wickham et al. 2017). Furthermore, Barren/Sand
(< 1% of total area) may represent a transitional state in the urbanization process
(ground clearing and early construction) that could simultaneously be accurately char-
acterized as a Developed class.
The accuracy of this Greater Houston land cover product generally compares favour-
ably to those observed in similar studies, though caution is advised with direct compar-
ison owing to idiosyncrasies in ground cover complexity among regions, as well as the
distinct dierences in spatial and thematic resolution, reference data, and assessment
method (Gómez, White, and Wulder 2016). In a meta-analysis of over 500 studies
between 1989 and 2003, Wilkinson (2005) observed a mean accuracy of 76% (15.6%
sd). Furthermore, Herold et al. (2016) notes that map accuracies since 2011 generally
range from 61% to 87%. Interestingly, despite the advancements in satellite data
acquisition and classication algorithms, classication accuracies have not improved
signicantly in the past 30 years (Herold et al. 2016; Yu et al. 2014).
4.3. Towards ne-resolution, large-extent, annual land cover time series
The demand for map products capable of assessing increasingly ne-scale spatio-tem-
poral dynamics over large extents and long durations has accelerated in recent years for
research elds spanning the realms of urban socio-economics, hazard and risk mitiga-
tion/reduction, and ecosystem modelling (Jensen and Cowen 1999; Yu and Zhou 2017).
To meet this demand, international eorts have proceeded swiftly to operationalize
continuous, wall-to-wall monitoring of land cover change across the globe. The fast pace
of satellite deployments over the past few years, coupled with the profusion of increas-
ingly sophisticated data fusion techniques, has enabled near-daily monitoring of the
Earth surface (Zhu et al. 2015; Gómez, White, and Wulder 2016). However, cloud-free
historical imagery from workhorse satellites like those in the Landsat program remains
relatively sparse. This circumstance has forced researchers to compromise between
(among other things) resolution and extent in both temporal and spatial domains
(Lunetta et al. 2004). Classied maps derived from imagery at a medium spatial resolu-
tion typically possess coarse temporal resolution over a single scene (Dou and Chen
2017; Fenta et al. 2017) or over multi-scene extents (Gong et al. 2013; Sun et al. 2017), or
ne temporal but coarse spatial resolution (He, Lee, and Warner 2017; Xu, Zhang, and
Lin 2018). Most recently, studies have increasingly sought to create medium spatial
resolution land cover time series at an annual temporal resolution, though these
products may be limited in thematic resolution and spatial extent (Li, Gong, and Liang
2015; Song et al. 2016; Zhang and Weng 2016).
To mitigate the impact of limited scene availability as well as data gaps (e.g., due to
failure of the scan line corrector of the ETM+ sensor), researchers have increasingly
20 C. R. HAKKENBERG ET AL.
employed data fusion for multi-temporal classications (Gómez, White, and Wulder
2016). One popular approach to ensure spatio-temporally consistent imagery, espe-
cially for large-area classications in heavily-clouded or undersampled regions, is the
generation of best-available-pixels (BAP) composites for a given time period (White
et al. 2014). Other approaches include data blending methods whereby data gaps are
interpolated using temporally proximate imagery (Yin et al. 2017), as well as multi-
sensor data fusion for the production of synthetic images with high temporal precision
for a given date (Gong et al. 2013;Zhuetal.2015). In this study, where annual
classication accuracy was prioritized over subannual temporal precision compositing,
gap-lling, and multi-date data fusion were performed at the classication stage. Using
all acceptable imagery within the calendar year, classiers were parameterized with
the original reectance retrievals and benet from the added information content of
multi-seasonal imagery, while reducing the impact of any one image on classication
results. Because data fusion occurs at the classication stage (and not preceding it),
pixel-wise uncertainties can be readily derived from the posterior membership prob-
abilities of the ensemble prediction.
Despite the performance of AASG, that ensures that each automated training set was
adapted to the radiometric idiosyncrasies of each new scene, and robust nonparametric
classiers like RF, numerous factors remain to aect the accuracy and consistency of land
cover classications derived from spectral data (Gray and Song 2013). Classication
errors due to signal noise from subpixel heterogeneity and bidirectional reectance
distribution function eects, atmospheric contamination, as well as classier confusion
among cover classes tend to manifest in space (Song et al. 2015). At the same time,
inconsistent surface reectance retrievals due to varying specications among sensors,
sensor degradation through time, radiometric and atmospheric changes between
images, as well as geolocational misalignment between dates may result in temporal
inconsistencies along the classication time series (Roy et al. 2016). Spatial and temporal
classication errors may then, in turn, propagate in multi-temporal classications. To
ensure greater spatio-temporal consistency in dense land cover map time series, post-
classication stabilization of time series results is a critical step for improving classica-
tion accuracy and consistency (Li et al. 2014; Lu and Weng 2007). Rule-based ltering
techniques based on the spatio-temporal context of a focal pixel are highly ecient for
processing very large classication time series (He, Lee, and Warner 2017; Wang et al.
2015; Pouliot et al. 2014; Gao et al. 2012), while more computationally-intensive sto-
chastic model-based approaches allow for uncertainty estimates to propagate through
all steps (Wang et al. 2015; Liu and Cai 2012).
5. Conclusion
In this study, we developed an innovative automated classication algorithm that takes
advantage of the synergistic value of all acceptable Landsat images in a single year,
using aggregate votes from the posterior predictive distributions of multiple image
composites to mitigate against misclassications in any one image in the annual stack,
and ll gaps due to missing and contaminated data, such as those from clouds and
cloud shadows. Using this ensemble classication algorithm, we produced a multi-scene,
annual land cover time series characterizing 21 years of dynamic land cover change
INTERNATIONAL JOURNAL OF REMOTE SENSING 21
trends in the 35,000 km
2
Greater Houston area. Importantly, all input data were con-
strained to their corresponding calendar year to ensure temporal precision sucient for
researchers seeking a land cover dataset from which to investigate higher-order patterns
in humanenvironment interactions. Land cover products of ne spatio-temporal reso-
lution provide the means to isolate specic drivers of regional change (including
environmental disturbances, economic cycles, and policy feedbacks) from their obser-
vable footprint on the ground. Furthermore, they provide sucient temporal detail from
which to estimate periodicity and temporal lags for parametrizing forecast models of
future development. For this study, ecological categories were designed to be su-
ciently broad to allow for temporal consistency within the hierarchical classication
scheme, but still readily supplemented with the most up-to-date spatial distributions
of, for instance, ecological transitions, biomass estimates, and wetland delineations that
are otherwise beyond the scope of the current study.
Rapid and vast urbanization trends, coupled with more frequent and intense hurri-
canes, could have devastating consequences for cities like Houston in the coming
decades, and especially for their most vulnerable inhabitants. Planning for these con-
tingencies will, at the regional scale, require a concerted eort to ensure that resistance
and resilience is built into future development plans. Continued advances in near-
continuous, wall-to-wall Earth observation and automated land cover characterization
will provide planners and policy-makers the requisite tools to make informed choices.
Acknowledgments
The authors thank the Houston Endowment, the Kinder Institute for Urban Research, and the Rice
University Academy of Fellows for support of this research. DigitalGlobe data were provided by
NASAs Commercial Archive Data (cad4nasa.gsfc.nasa.gov) under the National Geospatial-
Intelligence Agencys NextView license agreement. We would also like to thank Eric Smith and
the Kinder Institute Urban Data Platform team.
Disclosure statement
No potential conict of interest was reported by the authors.
Data availability statement
The data that support the ndings of this study are openly available at the Kinder Institute for
Urban Research Urban Data Platform: www.kinderudp.org/#/datasetCatalog/zbn96g5x658z
ORCID
C.R. Hakkenberg http://orcid.org/0000-0002-6579-5954
References
Belgiu, M., and L. Drăgu. 2016.Random Forest in Remote Sensing: A Review of Applications and
Future Directions.ISPRS Journal of Photogrammetry and Remote Sensing 114: 2431.
doi:10.1016/j.isprsjprs.2016.01.011.
22 C. R. HAKKENBERG ET AL.
Breiman, L. 2001.Random Forests.Machine Learning 45: 532. doi:10.1023/A:1010933404324.
Brody, S., H. Kim, and J. Gunn. 2013.Examining the Impacts of Development Patterns on Flooding
on the Gulf of Mexico Coast.Urban Studies 50 (4): 789806. doi:10.1177/0042098012448551.
Bullock, A., and M. Acreman. 2003.The Role of Wetlands in the Hydrological Cycle.Hydrology and
Earth System Sciences 7: 358389. doi:10.5194/hess-7-358-2003.
Cai, S., D. Liu, D. Sulla-Menashe, and M. A. Friedl. 2014.Enhancing MODIS Land Cover Product
with a Spatial-Temporal Modeling Algorithm.Remote Sensing of Environment 147: 243255.
Elsevier Inc. doi:10.1016/j.rse.2014.03.012.
Chen, C., A. Liaw, and L. Breiman. 2004.Using Random Forest to Learn Imbalanced Data.Journal
of Machine Learning Research, No 666: 112.
Congalton, R. G. 1991.A Review of Assessing the Accuracy of Classications of Remotely Sensed
Data.Remote Sensing of Environment 37 (1): 3546. doi:10.1016/0034-4257(91)90048-B.
Dannenberg, M. P., C. R. Hakkenberg, and C. Song. 2016.Consistent Classication of Landsat Time
Series with an Improved Automatic Adaptive Signature Generalization Algorithm.Remote
Sensing 8: 8. doi:10.3390/rs8080691.
Dannenberg, M. P., C. Song, and C. R. Hakkenberg. 2018.A Long-Term, Consistent Land Cover
Database for the Southeastern United States Using Automatic Adaptive Signature
Generalization (AASG).Photogrammetric Engineering & Remote Sensing 84 (9): 3544.
Dou, P., and Y. Chen. 2017.Dynamic Monitoring of Land-Use/Land-Cover Change and Urban
Expansion in Shenzhen Using Landsat Imagery from 1988 to 2015.International Journal of
Remote Sensing 38 (19): 53885407. doi:10.1080/01431161.2017.1339926.
Emanuel, K. 2017.Assessing the Present and Future Probability of Hurricane Harveys Rainfall.
Proceedings of the National Academy of Sciences 201716222. doi:10.1073/pnas.1716222114.
Fenta, A. A., H. Yasuda, N. Haregeweyn, A. S. Belay, Z. Hadush, M. A. Gebremedhin, and G.
Mekonnen. 2017.The Dynamics of Urban Expansion and Land Use/Land Cover Changes
Using Remote Sensing and Spatial Metrics: The Case of Mekelle City of Northern Ethiopia.
International Journal of Remote Sensing 38 (14): 41074129. Taylor & Francis. doi:10.1080/
01431161.2017.1317936.
Foody, G. M. 2002.Status of Land Cover Classication Accuracy Assessment.Remote Sensing of
Environment 80 (1): 185201. doi:10.1016/S0034-4257(01)00295-4.
Fry, J. A., G. Xian, S. Jin, J. A. Dewitz, C. G. Homer, L. Yang, C. A. Barnes, N. D. Herold, and J. D.
Wickham. 2011.Completion of the 2006 National Land Cover Database for the Conterminous
United States.Photogrammetric Engineering and Remote Sensing 77 (9): 566858.
Galster, G., R. Hanson, M. R. Ratclie, H. Wolman, S. Coleman, and J. Freihage. 2001.Wrestling
Sprawl to the Ground: Dening and Measuring an Elusive Concept.Housing Policy Debate
12 (4): 681717. doi:10.1080/10511482.2001.9521426.
Gao, F., E. B. De Colstoun, R. Ma, Q. Weng, J. G. Masek, J. Chen, Y. Pan, and C. Song. 2012.Mapping
Impervious Surface Expansion Using Medium-Resolution Satellite Image Time Series: A Case Study
in the Yangtze River Delta, China.International Journal of Remote Sensing 33 (24): 76097628.
doi:10.1080/01431161.2012.700424.
Gislason, P. O., J. A. Benediktsson, and J. R. Sveinsson. 2006.Random Forests for Land Cover
Classication.Pattern Recognition Letters 27 (4): 294300. doi:10.1016/j.patrec.2005.08.011.
Gómez, C., J. C. White, and M. A. Wulder. 2016.Optical Remotely Sensed Time Series Data for Land
Cover Classication: A Review.ISPRS Journal of Photogrammetry and Remote Sensing 116: 5572.
doi:10.1016/j.isprsjprs.2016.03.008.
Gong, P., J. Wang, L. Yu, Y. Zhao, Y. Zhao, L. Liang, Z. Niu, et al. 2013.Finer Resolution Observation
and Monitoring of Global Land Cover: First Mapping Results with Landsat TM and ETM+ Data.
International Journal of Remote Sensing 34 (7): 26072654. doi:10.1080/01431161.2012.748992.
Gray, J., and C. Song. 2013.Consistent Classication of Image Time Series with Automatic
Adaptive Signature Generalization.Remote Sensing of Environment 134 (July): 333341.
Elsevier Inc. doi:10.1016/j.rse.2013.03.022.
Hamilton, N. 2015.Smoother: Functions Relating to the Smoothing of Numerical Data.R Package
Version 1.1. https://cran.r-project.org/package=smoother
INTERNATIONAL JOURNAL OF REMOTE SENSING 23
He, Y., E. Lee, and T. A. Warner. 2017.A Time Series of Annual Land Use and Land Cover Maps of
China from 1982 to 2013 Generated Using AVHRR GIMMS NDVI3g Data.Remote Sensing of
Environment 199 (September): 201217. Elsevier Inc. doi:10.1016/j.rse.2017.07.010.
Herold, M., L. See, N. E. Tsendbazar, and S. Fritz. 2016.Towards an Integrated Global Land Cover
Monitoring and Mapping System.Remote Sensing 8 (12): 111. doi:10.3390/rs8121036.
HGAC. 2018.Houston-Galveston Area Council.Accessed 10 July 2017. http://www.h-gac.com
Hijmans, R. J. 2017.Raster: Geographic Data Analysis and Modeling.R Package Version 2.67.
https://cran.r-project.org/package=raster
Homer, C., J. Dewitz, L. Yang, S. Jin, P. Danielson, G. Xian, J. Coulston, N. Herold, J. Wickham, and K.
Megown. 2015.Completion of the 2011 National Land Cover Database for the Conterminous
United States-Representing a Decade of Land Cover Change Information.Photogrammetric
Engineering and Remote Sensing 81 (5): 345354. doi:10.14358/PERS.81.5.345.
Homer, C., J. Dewitz, J. Fry, M. Coan, N. Hossain, C. Larson, N. Herold, A. Mckerrow, J. Nick Vandriel,
and J. Wickham. 2007.Completion of the 2001 National Land Cover Database for the
Conterminous United States.Photogrammetric Engineering & Remote Sensing 73 (4): 337341.
Jaret, C., R. Ghadge, L. W. Reid, and R. M. Adelman. 2009.The Measurement of Suburban Sprawl:
An Evaluation.City and Community. doi:10.1111/j.1540-6040.2009.01270.x.
Jensen, J. R., and D. C. Cowen. 1999.Remote Sensing of Urban Suburban Infrastructure and Socio-
Economic Attributes.Photogrammetric Engineering and Remote Sensing 65 (5): 611622.
Kinder Institute. 2018.Urban Data Platform.https://www.kinderudp.org/.
Knutson, T. R., J. L. McBride, J. Chan, K. Emanuel, G. Holland, C. Landsea, I. Held, J. P. Kossin, A. K.
Srivastava, and M. Sugi. 2010.Tropical Cyclones and Climate Change.Nature Geoscience 3:
157163. doi:10.1038/ngeo779.
Li, M., S. Zang, B. Zhang, S. Li, and C. Wu. 2014.A Review of Remote Sensing Image Classication
Techniques: The Role of Spatio-Contextual Information.European Journal of Remote Sensing 47 (1):
389411. doi:10.5721/EuJRS20144723.
Li, X., P. Gong, and L. Liang. 2015.A 30-Year (1984-2013) Record of Annual Urban Dynamics of
Beijing City Derived from Landsat Data.Remote Sensing of Environment 166: 7890.
doi:10.1016/j.rse.2015.06.007.
Liaw, A., and M. Wiener. 2002.Classication and Regression by RandomForest.RNews2 (3): 1822.
Liu, D., and S. Cai. 2012.A Spatial-Temporal Modeling Approach to Reconstructing Land-Cover
Change Trajectories from Multi-Temporal Satellite Imagery.Annals of the Association of
American Geographers 102 (6): 13291347. doi:10.1080/00045608.2011.596357.
Loveland, T. R., and J. L. Dwyer. 2012.Landsat: Building a Strong Future.Remote Sensing of
Environment 122: 2229. doi:10.1016/j.rse.2011.09.022.
Lu, D., and Q. Weng. 2007.ASurveyofImageClassication Methods and Techniques for Improving
Classication Performance.International Journal of Remote Sensing 28 (5): 823870. doi:10.1080/
01431160600746456.
Lunetta, R. S., D. M. Johnson, J. G. Lyon, and J. Crotwell. 2004.Impacts of Imagery Temporal
Frequency on Land-Cover Change Detection Monitoring.Remote Sensing of Environment 89 (4):
444454. doi:10.1016/j.rse.2003.10.022.
Maxwell, A. E., T. A. Warner, and F. Fang. 2018.Implementation of Machine-Learning Classication in
Remote Sensing: An Applied Review.International Journal of Remote Sensing 39 (9): 27842817.
Taylor & Francis. doi:10.1080/01431161.2018.1433343.
NOAA. 2018.U.S. Billion-Dollar Weather & Climate Disasters 19802017.Accessed 10 January
2018. https://www.ncdc.noaa.gov/billions/events.pdf
NOAA C-CAP. 2011.Coastal Change Analysis Program (C-CAP) Regional Land Cover. Charleston, SC:
NOAA Oce for Coastal Management. Accessed March 2018. www.coast.noaa.gov/ccapftp.
Olofsson, P., G. M. Foody, M. Herold, S. V. Stehman, C. E. Woodcock, and M. A. Wulder. 2014.Good
Practices for Estimating Area and Assessing Accuracy of Land Change.Remote Sensing of
Environment 148: 4257. doi:10.1016/j.rse.2014.02.015.
Pontius, R. G., and M. Millones. 2011.Death to Kappa: Birth of Quantity Disagreement and
Allocation Disagreement for Accuracy Assessment.International Journal of Remote Sensing 32:
44074429. doi:10.1080/01431161.2011.552923.
24 C. R. HAKKENBERG ET AL.
Pouliot, D., R. Latifovic, N. Zabcic, L. Guindon, and I. Olthof. 2014.Development and Assessment of
a 250m Spatial Resolution MODIS Annual Land Cover Time Series (20002011) for the Forest
Region of Canada Derived from Change-Based Updating.Remote Sensing of Environment 140:
731743. Elsevier B.V. doi:10.1016/j.rse.2013.10.004.
Qian, Z. 2010.Without Zoning: Urban Development and Land Use Controls in Houston.Cities 27 (1):
3141. doi:10.1016/j.cities.2009.11.006.
R Core Team. 2017.R: A Language and Environment for Statistical Computing. Vienna, Austria.: R
Foundation for Statistical Computing.
Roy, D. P., V. Kovalskyy, H. K. Zhang, E. F. Vermote, L. Yan, S. S. Kumar, and A. Egorov. 2016.
Characterization of Landsat-7 to Landsat-8 Reective Wavelength and Normalized Dierence
Vegetation Index Continuity.Remote Sensing of Environment 185: 5770. doi:10.1016/j.
rse.2015.12.024.
Sexton, J. O., D. L. Urban, M. J. Donohue, and C. Song. 2013.Long-Term Land Cover Dynamics by
Multi-Temporal Classication across the Landsat-5 Record.Remote Sensing of Environment 128:
246258. doi:10.1016/j.rse.2012.10.010.
Song, C., J. M. Chen, T. Hwang, A. Gonsamo, H. Croft, Q. Zhang, M. Dannenberg, Y. Zhang, C. R.
Hakkenberg, and J. Li. 2015.Ecological Characterization of Vegetation Using Multi-Sensor
Remote Sensing in the Solar Reective Spectrum.In Remote Sensing Handbook, Vol 2. Land
Resources: Monitoring, Modeling, and Mapping, edited by P. S. Thenkabail, 533575. London, UK.:
Taylor and Francis.
Song, C., and C. E. Woodcock. 2003.Monitoring Forest Succession with Multitemporal Landsat
Images: Factors of Uncertainty.IEEE Transactions on Geoscience and Remote Sensing 41 (11):
25572567. doi:10.1109/TGRS.2003.818367.
Song, C., C. E. Woodcock, K. C. Seto, M. P. Lenney, and S. A. Macomber. 2001.Classication and
Change Detection Using Landsat TM Data: When and How to Correct Atmospheric Eects?
Remote Sensing of Environment 75 (2): 230244. doi:10.1016/S0034-4257(00)00169-3.
Song, X. P., J. O. Sexton, C. Huang, S. Channan, and J. R. Townshend. 2016.Characterizing the
Magnitude, Timing and Duration of Urban Growth from Time Series of Landsat-Based
Estimates of Impervious Cover.Remote Sensing of Environment 175: 113. doi:10.1016/j.
rse.2015.12.027.
Stehman, S. V. 2013.Estimating Area from an Accuracy Assessment Error Matrix.Remote Sensing
of Environment 132: 202211. doi:10.1016/j.rse.2013.01.016.
Sun, Y., X. Zhang, Y. Zhao, and Q. Xin. 2017.Monitoring Annual Urbanization Activities in
Guangzhou Using Landsat Images (19872015).International Journal of Remote Sensing 38 (5):
12581276. doi:10.1080/01431161.2016.1268283.
U.S. Census Bureau. 2018.Resident Population for Houston-The Woodlands-Sugar Land, TX (MSA)
Retrieved from FRED, Federal Reserve Bank of St. Louis.Accessed 18 May 2018. https://fred.
stlouisfed.org/series/TX.
Vogelmann, J. E., A. L. Gallant, H. Shi, and Z. Zhu. 2016.Perspectives on Monitoring Gradual
Change across the Continuity of Landsat Sensors Using Time-Series Data.Remote Sensing of
Environment 185: 258270. Elsevier B.V. doi:10.1016/j.rse.2016.02.060.
Wang, J., Y. Zhao, C. Li, L. Yu, D. Liu, and P. Gong. 2015.Mapping Global Land Cover in 2001 and
2010 with Spatial-Temporal Consistency at 250m Resolution.ISPRS Journal of Photogrammetry
and Remote Sensing 103: 3847. doi:10.1016/j.isprsjprs.2014.03.007.
Wang, P., C. Huang, and E. B. de Colstoun. 2017.Mapping 20002010 Impervious Surface Change
in India Using Global Land Survey Landsat Data.Remote Sensing 9 (4): 366. doi:10.3390/
rs9040366.
Weng, Q. 2012.Remote Sensing of Impervious Surfaces in the Urban Areas: Requirements,
Methods, and Trends.Remote Sensing of Environment 117: 3449. Elsevier Inc. doi:10.1016/j.
rse.2011.02.030.
White, J. C., M. A. Wulder, G. W. Hobart, J. E. Luther, T. Hermosilla, P. Griths, N. C. Coops, et al.
2014.Pixel-Based Image Compositing for Large-Area Dense Time Series Applications and
Science.Canadian Journal of Remote Sensing 40 (3): 192212. doi:10.1080/
07038992.2014.945827.
INTERNATIONAL JOURNAL OF REMOTE SENSING 25
Wickham, J., S. V. Stehman, L. Gass, J. A. Dewitz, D. G. Sorenson, B. J. Granneman, R. V. Poss, and L.
A. Baer. 2017.Thematic Accuracy Assessment of the 2011 National Land Cover Database
(NLCD).Remote Sensing of Environment 191: 328341. Elsevier Inc. doi:10.1016/j.
rse.2016.12.026.
Wilkinson, G. G. 2005.Results and Implications of a Study of Fifteen Years of Satellite Image
Classication Experiments.IEEE Transactions on Geoscience and Remote Sensing 43 (3): 433440.
doi:10.1109/TGRS.2004.837325.
Wilson, S. G., D. A. Plane, P. J. Mackun, T. R. Fischetti, and J. Goworowska. 2012.Patterns of
Metropolitan and Micropolitan Population Change: 2000 to 2010.Report Number: C2010SR-01.
Washington DC.
Woodcock, C. E., and S. Gopal. 2000.Fuzzy Set Theory and Thematic Maps: Accuracy and Area
Estimation.International Journal of Geographical Information Science 14 (2): 2. doi:10.1080/
136588100240895.
Xu, R., H. Zhang, and H. Lin. 2018.Annual Dynamics of Impervious Surfaces at City Level of Pearl
River Delta Metropolitan.International Journal of Remote Sensing 39 (11): 35373555.
doi:10.1080/01431161.2018.1444290.
Yin, G., G. Mariethoz, Y. Sun, and M. F. McCabe. 2017.A Comparison of Gap-Filling Approaches for
Landsat-7 Satellite Data.International Journal of Remote Sensing 38 (23): 66536679. Taylor &
Francis. doi:10.1080/01431161.2017.1363432.
Yu, L., L. Liang, J. Wang, Y. Zhao, Q. Cheng, L. Hu, S. Liu, et al. 2014.Meta-Discoveries from a
Synthesis of Satellite-Based Land-Cover Mapping Research.International Journal of Remote
Sensing 35 (13): 45734588. doi:10.1080/01431161.2014.930206.
Yu, W., and W. Zhou. 2017.The Spatiotemporal Pattern of Urban Expansion in China: A
Comparison Study of Three Urban Megaregions.Remote Sensing 9 (1): 1921. doi:10.3390/
rs9010045.
Zhang, L., and Q. Weng. 2016.Annual Dynamics of Impervious Surface in the Pearl River Delta,
China, from 1988 to 2013, Using Time Series Landsat Imagery.ISPRS Journal of Photogrammetry
and Remote Sensing 113: 8696. doi:10.1016/j.isprsjprs.2016.01.003.
Zhu, Z., S. Wang, and C. E. Woodcock. 2015.Improvement and Expansion of the Fmask Algorithm:
Cloud, Cloud Shadow, and Snow Detection for Landsats 47, 8, and Sentinel 2 Images.Remote
Sensing of Environment 159: 269277. doi:10.1016/j.rse.2014.12.014.
Zhu, Z., C. E. Woodcock, C. Holden, and Z. Yang. 2015.Generating Synthetic Landsat Images Based
on All Available Landsat Data: Predicting Landsat Surface Reectance at Any Given Time.
Remote Sensing of Environment 162: 6783. doi:10.1016/j.rse.2015.02.009.
26 C. R. HAKKENBERG ET AL.
... Many classification algorithms, including support vector machine (SVM), maximum likelihood (ML), naive Bayesian (NB), decision tree (DT), K-nearest neighbour (KNN), multi-layer perceptron (MLP) and deep belief network (DBN), have been developed and applied to LULC production (Dou and Chen, 2017a;Maxwell et al., 2018;Du, et al., 2020). However, single-temporal RS image only provides the instantaneous spectrum of the land surface, and the features that can be used for classification is infrequent, resulting in very few recognitions of LULC categories (Gómez, et al., 2016;Hakkenberg et al., 2018). Achieving high classification accuracy using single-temporal RS image is difficult especially for the categories of different crops and other vegetations (Zhong et al., 2019;Tsai et al., 2018). ...
... Classification methods can be roughly divided into two types according to the structure of TSI. The first is to stack multi-temporal images by time sequence and classify them with SVM, random forest (RF), and other classifiers (Hakkenberg et al., 2018;Whelen and Siqueira, 2018;Gong et al., 2019). This method performs well in producing annual LULC, such as finer resolution observation and monitoring of global land cover (Gong et al., 2019). ...
Article
Full-text available
Recently, time series image (TSI) has been reported to be an effective resource to mapping fine land use/land cover (LULC), and deep learning, in particular, has been gaining growing attention in this field. However, deep learning methods using single classifier need further improvement for accurate TSI classification owing to the 1D temporal properties and insufficient dense time series of the remote sensing images. To overcome such disadvantages, we proposed an innovative approach involving construction of TSI and combination of deep learning and multiple classifiers system (MCS). Firstly, we used a normalised difference index (NDI) to establish an NDIs-based TSI and then designed a framework consisting of a deep learning-based feature extractor and multiple classifiers system (MCS) based classification model to classify the TSI. With the new approach, our experiments were conducted on Landsat images located in two counties, Sutter and Kings in California, United States. The experimental results indicate that our proposed method achieves great progress on accuracy improvement and LULC mapping, outperforming classifications using comparative deep learning and non-deep learning methods.
... Driven by a demand for spatio-temporal accuracy and consistency across the multidecadal imagery time series (where individual images may vary due to seasonal illumination angles and atmospheric conditions), the land cover change dataset was generated using a threepart algorithmic procedure: (1) automatic adaptive signature generalization (Dannenberg et al. 2016) for automated training data selection from NLCD classifications from 2001, 2006, and 2011(Homer et al. 2015, (2) machine learning image classification using random forests to classify atmospherically-corrected image spectra to one of the four aforementioned developed classes (Hakkenberg et al. 2020), and (3) spatio-temporal filtering to reduce erroneous classifications due to clouds, atmospheric contamination, and other sources of data noise and model errors among the 153 billion pixels classified (Hakkenberg et al. 2019). All classifications were validated using independent, multi-temporal fine-resolution imagery from the Ikonos, Quickbird, and Worldview sensors (Hakkenberg et al. 2019). ...
... Driven by a demand for spatio-temporal accuracy and consistency across the multidecadal imagery time series (where individual images may vary due to seasonal illumination angles and atmospheric conditions), the land cover change dataset was generated using a threepart algorithmic procedure: (1) automatic adaptive signature generalization (Dannenberg et al. 2016) for automated training data selection from NLCD classifications from 2001, 2006, and 2011(Homer et al. 2015, (2) machine learning image classification using random forests to classify atmospherically-corrected image spectra to one of the four aforementioned developed classes (Hakkenberg et al. 2020), and (3) spatio-temporal filtering to reduce erroneous classifications due to clouds, atmospheric contamination, and other sources of data noise and model errors among the 153 billion pixels classified (Hakkenberg et al. 2019). All classifications were validated using independent, multi-temporal fine-resolution imagery from the Ikonos, Quickbird, and Worldview sensors (Hakkenberg et al. 2019). ...
Article
Full-text available
Urbanization results in increasing impervious surfaces with the potential to threaten fragile environments and heighten flood risks. In the United States, research on the social processes driving urbanization has tended to focus on the twenty-first century, but less is known about how temporal trends arose from the spatial layout of the urban land upon which this growth was founded. To address this gap, we present a novel interdisciplinary synthesis using neighborhood-level census data in tandem with a satellite-derived annual land cover change time series to assess the role of race, affluence, and socioeconomic status in shaping spatio-temporal urbanization in the Houston metropolitan area from 1997−2016. Results from cross-sectional and temporal regression models indicate that while social dynamics associated with historical versus recent urbanization are related, they are not identical. Thus, while temporal change in Houston's urbanization is driven primarily by socioeconomic status, the social dynamics associated with spatial disparities in urbanization relate primarily to race, regardless of socioeconomic status. These results are noteworthy as urbanization in Houston does not fully comport with existing theoretical perspectives or with empirical findings nationally. Instead, we suggest these findings reflect the city’s politics and culture surrounding land use. Thus, beyond its important social and environmental implications, this study affirms the utility of fusing socio-demographic data with satellite remote sensing of urban growth, and highlights the value of the socioenvironmental succession framework for characterizing urbanization as a recursive process in space and time.
... In addition to flooding resulting from extreme precipitation from hurricanes and cyclones over the study area, intense rainfall in the upstream watersheds of the San Jacinto and Brazos rivers that pass through the study area ( Figure 1) further complicate the flooding problem [21]. Over time, urbanization and other intense anthropogenic activities across the study area have led to changes in the land cover and land surface elevation, which are believed to have aggravated the impacts of flooding [14,22,23]. [22] estimated that over the past two decades , nearly 2040 km 2 (±400 km 2 ) of land in the Greater Houston area has been changed to less permeable developed land cover, out of which 14% were wetland areas. ...
... Over time, urbanization and other intense anthropogenic activities across the study area have led to changes in the land cover and land surface elevation, which are believed to have aggravated the impacts of flooding [14,22,23]. [22] estimated that over the past two decades , nearly 2040 km 2 (±400 km 2 ) of land in the Greater Houston area has been changed to less permeable developed land cover, out of which 14% were wetland areas. ...
Article
Full-text available
The Greater Houston metropolitan area has experienced recurring flooding events in the past two decades related to tropical cyclones and heavy inland rainfall. With the projected recurrence of severe weather events, an approach that outlines the susceptibility of different localities within the study area to potential floods based on analyses of the impacts from earlier events would be beneficial. We applied a novel C-band Sentinel-1 Synthetic Aperture Radar (SAR)-based flood detection method to map floodwater distribution following three recent severe weather events with the goal of identifying areas that are prone to future flood hazards. Attempts were made to calibrate and validate the C-band-based results and analyses to compensate for possible sources of error. These included qualitative and quantitative assessments on L-band aerial SAR data, as well as aerial imagery acquired after one of the events. The findings included the following: (1) most urban centers of Harris county, with few exceptions, are not believed to be prone to flooding hazards in contrast to the densely populated areas on the outskirts of Harris county; (2) nearly 44% of the mapped flood-prone areas lie within a 1 km distance of major drainage networks; (3) areas experiencing high subsidence rates have persistently experienced flooding, possibly exacerbated by morphological changes to the land surface induced by subsidence.
... Driven by a demand for spatio-temporal accuracy and consistency across the multidecadal imagery time series (where individual images may vary due to seasonal illumination angles and atmospheric conditions), the land cover change dataset was generated using a threepart algorithmic procedure: (1) automatic adaptive signature generalization (Dannenberg et al. 2016) for automated training data selection from NLCD classifications from 2001, 2006, and 2011(Homer et al. 2015, (2) machine learning image classification using random forests to classify atmospherically-corrected image spectra to one of the four aforementioned developed classes (Hakkenberg et al. 2020), and (3) spatio-temporal filtering to reduce erroneous classifications due to clouds, atmospheric contamination, and other sources of data noise and model errors among the 153 billion pixels classified (Hakkenberg et al. 2019). All classifications were validated using independent, multi-temporal fine-resolution imagery from the Ikonos, Quickbird, and Worldview sensors (Hakkenberg et al. 2019). ...
... Driven by a demand for spatio-temporal accuracy and consistency across the multidecadal imagery time series (where individual images may vary due to seasonal illumination angles and atmospheric conditions), the land cover change dataset was generated using a threepart algorithmic procedure: (1) automatic adaptive signature generalization (Dannenberg et al. 2016) for automated training data selection from NLCD classifications from 2001, 2006, and 2011(Homer et al. 2015, (2) machine learning image classification using random forests to classify atmospherically-corrected image spectra to one of the four aforementioned developed classes (Hakkenberg et al. 2020), and (3) spatio-temporal filtering to reduce erroneous classifications due to clouds, atmospheric contamination, and other sources of data noise and model errors among the 153 billion pixels classified (Hakkenberg et al. 2019). All classifications were validated using independent, multi-temporal fine-resolution imagery from the Ikonos, Quickbird, and Worldview sensors (Hakkenberg et al. 2019). ...
Article
Full-text available
Urbanization results in increasing impervious surfaces with the potential to threaten fragile environments and heighten flood risks. In the United States, research on the social processes driving urbanization has tended to focus on the twenty-first century, but less is known about how temporal trends arose from the spatial layout of developed land upon which this growth was founded. To address this gap, we present a novel interdisciplinary synthesis using neighborhood-level census data in tandem with a satellite-derived annual land cover change time series to assess the role of race, affluence, and socioeconomic status in shaping spatio-temporal urbanization in the Houston metropolitan area from 1997-2016. Results from cross-sectional and temporal regression models indicate that while social dynamics associated with historical versus recent urbanization are related, they are not identical. Thus, while temporal change in urbanization is driven primarily by socioeconomic status, the social dynamics associated with spatial disparities in urbanization relate primarily to race, regardless of socioeconomic status. The results are noteworthy as urbanization in Houston does not fully comport with existing theoretical perspectives or with empirical findings nationally. Instead, we suggest these findings reflect the city's politics and culture surrounding land use. Thus, beyond its important social and environmental implications, this study affirms the utility of fusing socio-demographic data with satellite remote sensing of urban growth, and highlights the value of the socioenvironmental succession framework for characterizing urbanization as a recursive process in space and time.
... The first satellite of Landsat series launched in 1972, while the last version is working until these days. Landsat satellite images are freely provided by the United State Geological Survey USGS with a medium spectral and spatial resolution (Hakkenberg et al., 2019). ...
... Among the clustering methods, common methods include means clustering and fuzzy C-means (FCM) clustering. Hakkenberg uses the principal component analysis method to extract change features, and use the means method to cluster pixels into two categories to obtain a different map [45]. ...
Article
Full-text available
This paper proposes a multi-temporal image change detection algorithm based on adaptive parameter estimation, which is used to solve the problems of severe interference of coherent speckle noise and the retention of detailed information about changing regions in synthetic aperture radar remote sensing images. The change area in the initial differential image has local consistency and global prominence. By detecting the significant area to locate similar change areas, the coherent speckle noise outside the area can be eliminated. The use of hierarchical FCM clustering to automatically generate training samples can improve the reliability of training samples. In addition, in order to increase the distinction between the changed area and the non-changed area, a sparse automatic encoder is used to extract the changed features and generate a change detection map. Experiments using 4 sets of SAR images show that the algorithm can effectively reduce the effect of speckle noise on detection accuracy, the extraction of changing areas is more complete and meticulous, and the false detection rate is greatly reduced. Since the images in different time phases will be disturbed by weather, clouds, sea water, etc., the target segmentation algorithm can be used to extract the target of interest and highlight the changing area. Principal component analysis and kmeans clustering method are used to reduce the influence of isolated pixels, and change information is extracted to obtain different images. The experiment uses four sets of image data of islands and reefs. The experiment proves that the algorithm can well eliminate external interference, improve the accuracy of change detection, and have a good detection effect on the area of islands and reefs. The adaptive parameter estimation plays a good role in the detection of changing areas, and the visual effect is better, which can improve the accuracy of the detection results.
... Bi-temporal change detection using image pairs has been used effectively to quantify state change (e.g. land cover class) or relative change in surface characteristics between two dates, but is unable to capture higher-order temporal dynamics, including gradual change, periodicity, and change rates [2]. Reflecting the demand for more temporally-frequent land surface data products for disparate applications from land cover change to biophysical land surface models [3], [4], the use of multitemporal image time series has increased rapidly [5]. ...
Article
Full-text available
The characterization of fine temporal-resolution land surface dynamics from broadband optical satellite sensors is constrained by sparse acquisitions of high-quality imagery; interscene variation in radiometric, phenological, atmospheric, and illumination conditions; and subpixel variability in heterogeneous environments. In this letter, we address these concerns by developing and testing the automatic adaptive signature generalization and regression (AASGr) algorithm. Provided a robust reference map corresponding to the date of one image, AASGr automates the prediction of continuous fields maps from imagery time series that is adaptive to the spectral and radiometric characteristics of each target image and thereby requires neither atmospheric correction nor data normalization. We tested AASGr on a 22-year Landsat time series to quantify subannual impervious fractional cover dynamics in Houston, TX--an area characterized by a high degree of spatial heterogeneity in surface cover and high frequency in land cover change. The map time series achieved high accuracy in a three-part validation procedure and reveals spatio-temporal dynamics of urban intensification and extensification at a level of detail previously elusive in discrete classifications or coarse temporal-resolution map products. The automation of continuous fields time series is enabling a new generation of land surface products capable of characterizing precise morphologies along a continuum of spatio-temporal change. While AASGr was applied here to predict subpixel impervious fractional cover from Landsat imagery, the method is generalizable to a range of imagery and applications requiring dense continuous fields time series with uncertainty estimates of geophysical and biochemical characteristics, such as leaf area index, biomass, and albedo.
Article
Full-text available
Typhoon Hagibis passed through Japan on October 12, 2019, bringing heavy rainfall over half of Japan. Twelve banks of seven state-managed rivers collapsed, flooding a wide area. Quick and accurate damage proximity maps are helpful for emergency responses and relief activities after such disasters. In this study, we propose a quick analysis procedure to estimate inundations due to Typhoon Hagibis using multi-temporal Sentinel-1 SAR intensity images. The study area was Ibaraki Prefecture, Japan, including two flooded state-managed rivers, Naka and Kuji. First, the completely flooded areas were detected by two traditional methods, the change detection and the thresholding methods. By comparing the results in a part of the affected area with our field survey, the change detection was adopted due to its higher recall accuracy. Then, a new index combining the average value and the standard deviation of the differences was proposed for extracting partially flooded built-up areas. Finally, inundation maps were created by merging the completely and partially flooded areas. The final inundation map was evaluated via comparison with the flooding boundary produced by the Geospatial Information Authority (GSI) and the Ministry of Land, Infrastructure, Transport, and Tourism (MLIT) of Japan. As a result, 74% of the inundated areas were able to be identified successfully using the proposed quick procedure.
Article
Full-text available
We estimate, for current and future climates, the annual probability of areally averaged hurricane rain of Hurricane Harvey’s magnitude by downscaling large numbers of tropical cyclones from three climate reanalyses and six climate models. For the state of Texas, we estimate that the annual probability of 500 mm of area-integrated rainfall was about 1% in the period 1981–2000 and will increase to 18% over the period 2081–2100 under Intergovernmental Panel on Climate Change (IPCC) AR5 representative concentration pathway 8.5. If the frequency of such event is increasingly linearly between these two periods, then in 2017 the annual probability would be 6%, a sixfold increase since the late 20th century.
Article
Full-text available
Understanding and monitoring the environmental impacts of global urbanization requires better urban datasets. Continuous field impervious surface change (ISC) mapping using Landsat data is an effective way to quantify spatiotemporal dynamics of urbanization. It is well acknowledged that Landsat-based estimation of impervious surface is subject to seasonal and phenological variations. The overall goal of this paper is to map 2000-2010 ISC for India using Global Land Survey datasets and training data only available for 2010. To this end, a method was developed that could transfer the regression tree model developed for mapping 2010 impervious surface to 2000 using an iterative training and prediction (ITP) approachAn independent validation dataset was also developed using Google Earth™ imagery. Based on the reference ISC from the validation dataset, the RMSE of predicted ISC was estimated to be 18.4%. At 95% confidence, the total estimated ISC for India between 2000 and 2010 is 2274.62 ± 7.84 km².
Article
Full-text available
To analyse changes in human settlement in Shenzhen City during the past three decades, changes in land use/land cover (LULC) and urban expansion were investigated based on multi-temporal Landsat Thematic Mapper/Enhanced Thematic Mapper Plus/Operational Land Imager (TM/ETM+/OLI) images. Using C4.5-based AdaBoost, a hierarchical classification method was developed to extract specific classes with high accuracy by combining a specific number of base-classifier decisions. Along with a classification post-processing approach, the classification accuracy was greatly improved. The statistical analysis of LULC changes from 1988 to 2015 shows that built-up areas have increased 6.4-fold, whereas cultivated land and forest continually decreased because of rapid urbanization. Urban expansion driven by human activities has considerably affected the landscape change of Shenzhen. The urban-expansion pattern of Shenzhen is a mixture of three urban-expansion patterns. Among these patterns, traffic-driven urban expansion has been the main form of urban expansion for some time, especially in the Non-Special Economic Zone. In addition, by taking 8 to 10 year periods as time intervals, urban expansion in Shenzhen was divided into three stages: the early-age urbanization stage (1988–1996), the rapid urbanization stage (1996–2005), and the intensive urbanization stage (2005–2015). For different stages, the state of urban expansion is different. In long-term LULC dynamic monitoring and urban-expansion detection, it was possible to obtain 11 LULC maps, which took 2 to 4 years as a research interval. With regard to the short research periods, LULC changes and urban expansion were investigated in detail.
Article
Full-text available
Information on the rate and pattern of urban expansion is required by urban planners to devise proper urban planning and management policy directions. This study evaluated the dynamics and spatial pattern of Mekelle City’s expansion in the past three decades (1984–2014). Multi-temporal Landsat images and Maximum Likelihood Classifier were used to produce decadal land use/land cover (LULC) maps. Changes in LULC and spatial pattern of urban expansion were analysed by post-classification change detection and spatial metrics, respectively. The results showed that in the periods 1984–1994, 1994–2004, and 2004–2014, the built-up area increased annually by 10%, 9%, and 8%, respectively; with an average annual increment of 19% (100 ha year−1 ), from 531 ha in 1984 to 3524 ha in 2014. Between 1984 and 2014, about 88% of the gain in built-up area was from conversion of agricultural lands, which decreased by 39%. Extension of existing urban areas was the dominant growth type, which accounted for 54%, 75%, and 81% of the total new development during 1984–1994, 1994–2004, and 2004–2014, respectively. The spatial metrics analyses revealed urban sprawl, with increased heterogeneity and gradual dispersion in the outskirts of the city. The per capita land consumption rate (ha per person) increased from 0.009 in 1984 to 0.014 in 2014, indicating low density urban growth. Based on the prediction result, the current (2014) built-up area will double by 2035, and this is likely to have multiple socioeconomic and environmental consequences unless sustainable urban planning and development policies are devised.
Article
Land-Cover/Land-Use (LCLU) change is a critical aspect of global environmental change, with profound social and ecological consequences. The southeastern U.S. in particular is changing rapidly, but a long-term, consistent LCLU history at fine spatial resolution does not exist for the region. Here, we present a new LCLU history of the southeastern U.S. based on temporal extension of the 2011 National Land Cover Database (NLCD) back to 1986. We used Automatic Adaptive Signature Generalization (AASG) to generate this product from Landsat TM/ETM+ imagery and ancillary topographic information. AASG identifies stable sites between two images and uses these stable sites to generate a new training dataset for updating a classification from one date to the next. Our long-term LCLU classifications are broadly consistent with the NLCD while providing a much longer historical record for characterizing recent changes in the southeastern U.S. and contextualizing their consequences for ecosystem services in the region.
Article
Human activities have transformed natural land surface to artificial constructions, which causes the urban regions increasingly expanding. Urban development can be reflected by the dynamic changes of impervious surfaces, which can be assessed using time-series satellite data. The Pearl River Delta (PRD) metropolitan has witnessed fast urban development over past decades. Even though many studied have focused on monitoring the urbanization of megacities, there is still a lack of comprehensive understanding about the long-term annual dynamics of urban impervious surfaces at city level in PRD. In this study, MOD09A1 (MODIS Surface Reflectance 8 Day L3 Global 500 m) products were chosen to assess the annual dynamics of impervious surface at city level in PRD. Temporal filtering was utilized to improve the accuracy of the impervious surface estimation. Moreover, a new index, normalized difference urban development index (NDUDI) was proposed to investigate relative development of impervious surface in PRD. Additionally, the economy factor, Gross Domestic Product (GDP) was introduced to reveal the correlation with the impervious surfaces. Results demonstrated that the overall accuracy for impervious surface extraction was about 91.00–94.00%. The impervious surfaces area in PRD dramatically increased from 6000 km² to 115,840 km² in the past 16 years. Guangzhou and Foshan have the fastest developing speed, while Dongguan, Shenzhen and Foshan sacrificed much more lands to the construction of PRD than other cities. Besides, areas of impervious surface per city show very good correlations with GDP in city of Guangzhou, Huizhou, Foshan, Zhongshan, Hong Kong and Zhuhai-Macau. The scatter plot between NDUDI and GDP data display different clustering pattern between different cities.
Article
Machine learning offers the potential for effective and efficient classification of remotely sensed imagery. The strengths of machine learning include the capacity to handle data of high dimensionality and to map classes with very complex characteristics. Nevertheless, implementing a machine-learning classification is not straightforward, and the literature provides conflicting advice regarding many key issues. This article therefore provides an overview of machine learning from an applied perspective. We focus on the relatively mature methods of support vector machines, single decision trees (DTs), Random Forests, boosted DTs, artificial neural networks, and k-nearest neighbours (k-NN). Issues considered include the choice of algorithm, training data requirements, user-defined parameter selection and optimization, feature space impacts and reduction, and computational costs. We illustrate these issues through applying machine-learning classification to two publically available remotely sensed data sets.
Article
The purpose of this study is to assess the relative performance of four different gap-filling approaches across a range of land-surface conditions, including both homogeneous and heterogeneous areas as well as in scenes with abrupt changes in landscape elements. The techniques considered in this study include: (1) Kriging and co-Kriging; (2) geostatistical neighbourhood similar pixel interpolator (GNSPI); (3) a weighted linear regression (WLR) algorithm; and (4) the direct sampling (DS) method. To examine the impact of image availability and the influence of temporal distance on the selection of input training data (i.e. time separating the training data from the gap-filled target image), input images acquired within the same season (temporally close) as well as in different seasons (temporally far) to the target image were examined, as was the case of using information only within the target image itself. Root mean square error (RMSE), mean spectral angle (MSA), and coefficient of determination (R²) were used as the evaluation metrics to assess the prediction results. In addition, the overall accuracy (OA) and kappa coefficient (kappa) were used to assess a land-cover classification based on the gap-filled images. Results show that all of the gap-filling approaches provide satisfactory results for the homogeneous case, with R² > 0.93 for bands 1 and 2 in all cases and R² > 0.80 for bands 3 and 4 in most cases. For the heterogeneous example, GNSPI performs the best, with R² > 0.85 for all tested cases. WLR and GNSPI exhibit equivalent accuracy when a temporally close input image is used (i.e. WLR and GNSPI both have an R² equal to 0.89 for band 1). For the case of abrupt changes in scene elements or in the absence of ancillary data, the DS approach outperforms the other tested methods.
Article
A time series of annual land use and land cover (LULC) maps that cover an extended period of time is a key dataset for climatological studies investigating land-atmosphere interaction. Change in LULC can influence regional climate by altering the surface roughness, soil moisture, heat flux partition, and terrestrial carbon storage. Although annual global LULC maps are generated from Moderate-resolution Imaging Spectroradiometer (MODIS) data, the earliest MODIS LULC map is for 2001, which limits the potential time period for climatological analyses. This study produced a continuous series of annual LULC maps of China from 1982 to 2013 using random forest classification of 19 phenological metrics derived from Advanced Very High Resolution Radiometer (AVHRR) Global Inventory Modeling and Mapping Studies (GIMMS) third generation NDVI (NDVI3g) data. The classifier was trained using reference data derived from the MODIS land cover type product (MCD12Q1). Based on a comparison with Google Earth images, the overall accuracy of a simplified eight-class version of our 2012 LULC map is 73.8%, which is not significantly different from the accuracy of the MODIS map of the same year. Our maps indicate that for the three decades studied, the area of croplands and forests in China increased, and the area of grasslands decreased. These annual maps of land cover will be an important dataset for future climate studies, and the methodologies used in this study can be applied to other geographical regions where availability of continuous time series of LULC maps is limited.