Xem mẫu
- Environmental Advances 5 (2021) 100073
Contents lists available at ScienceDirect
Environmental Advances
journal homepage: www.sciencedirect.com/journal/environmental-advances
Changes in precipitation patterns in Houston, Texas
Madeline D. Statkewicz, Robert Talbot 1, Bernhard Rappenglueck *
University of Houston, Department of Earth and Atmospheric Science, Houston TX, United States
A R T I C L E I N F O A B S T R A C T
Keywords: There has been an alarming increase in the frequency of major flooding events along the Gulf Coast over the last
Precipitation trends three decades, primarily due to events of unprecedented, or extreme, rainfall. Using data from 63 rain gauges
Climate change maintained by the Harris County Flood Control District’s Flood Warning System (HCFCD FWS), this study ex
Urbanization
amines the changes in daily precipitation amounts in the highly urbanized city of Houston, Texas, USA. The
Flooding
Houston
potential shift in annual precipitation patterns over a period of three decades (1989-2018) was examined by
investigating the numbers of dry and wet days as well as annual precipitation totals over the study period. Wet
days were then further scrutinized based on daily rainfall amounts (e.g., R10, R20, R30, R40, R50, R100) to
determine if extreme events are beginning to dominate annual rainfall amounts. Trends were analyzed for sta
tistical significance temporally using the Mann-Kendall and Sen’s slope methods and for spatial trends using GIS
applications. The results indicate a statistically significant increase in extreme rainfall at the expense of light,
moderate, and heavy rainfall over time. The only negative relationship is found in dry days. The most statistically
significant trends exist in the 99th percentile, maximum, and R100 parameters with p-values of 0.07, 0.08, and
0.11, respectively. There has been rapid growth and intensive development in the Houston area in recent decades
that continues to this day, and land cover change has been significant as 12.6% of Harris County changed to one
of four National Land Cover Database (NLCD) Urban classes (e.g., developed: barren, low intensity, medium
intensity, high intensity). This confirms that urbanization has continued to increase while total vegetative and
wetland coverage has decreased. The findings of this study provide essential guidance for city and state planners
and engineers.
1. Introduction are at risk. Brody et al. (2008) examined the relationship between the
built environment and flood impacts, finding that Texas consistently
Global climate is changing (IPCC, 2013), and a warming climate has sustains the most damage from flooding when compared to any other
been linked with an amplification of precipitation extremes, where state in the United States (US). According to the National Oceanic and
heavy rainfall events tend to increase during warm months and decrease Atmospheric Association’s National Centers for Environmental Infor
during cold months (Allan and Soden, 2008). Locally, there is a mation (NOAA NCEI), the State of Texas has experienced the greatest
noticeable shift in precipitation patterns which is influencing individual number of inflation-adjusted billion-dollar disasters between 1980 and
storm rainfall amounts, their intensity, and duration. Cases in point are 2019, and it also claims the highest cumulative damage costs in the
the number of 100- and 500-year storms and major hurricanes that are nation (Smith, 2020).
impacting the Gulf Coast region. Indeed, analysis of extreme rainfall The costs and impacts of extreme precipitation events may be
events along the Gulf Coast have been shown to be on the rise (van magnified in the future because of climate change (Melillo et al., 2014).
Oldenborgh et al., 2017; van der Wiel, 2017; van Oldenborgh, 2019). Further, most regions receive a greater fraction of total seasonal pre
Floods continue to pose the greatest threat to the property and safety cipitation from extreme events. These results imply that fewer but
of human communities among all-natural hazards in the United States heavier and more intense precipitation events may be the new normal in
(Brody et al., 2008). Moreover, a recent study by Quinn et al. (2017) the future, leading to more frequent wet and dry extremes in most re
concluded that flood risk in the United States is severely under gions of the U.S. (Singh et al., 2013). According to van Oldenborgh et al.
estimated, and that 40 million people and $5.5 trillion worth of assets (2018), the intensity of extreme precipitation increased about two times
* Corresponding author at: University of Houston, Department of Earth and Atmospheric Science, 4800 Calhoun Rd, Houston, TX 77204-5007, United States.
E-mail addresses: mdstatke@central.uh.edu (M.D. Statkewicz), brappenglueck@uh.edu (B. Rappenglueck).
1
deceased
https://doi.org/10.1016/j.envadv.2021.100073
Received 16 November 2020; Received in revised form 2 June 2021; Accepted 2 June 2021
Available online 6 June 2021
2666-7657/© 2021 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
- M.D. Statkewicz et al. Environmental Advances 5 (2021) 100073
the increase of moisture holding capacity of the atmosphere expected for 2. Materials and methods
1◦ C warming based on observations since 1880. Oldenborgh et al.
(2017) argued that the moisture flux was increased by both the moisture 2.1. Study area
content and stronger updrafts driven by enhanced release of latent heat.
Many modeling studies have been carried out regarding extreme Home to 22 individual watersheds, the “Bayou City,” is linked to the
precipitation and flooding events. Using a regional climate model for the Gulf of Mexico through the Houston Ship Channel, which is ranked first
area of Toronto, Canada, Wang et al. (2014) found that the intensity of by total tonnage among ports in the United States (https://www.bts.
extreme precipitation events is expected to increase at all durations and gov/content/tonnage-top-50-us-water-ports-ranked-total-tons).
frequencies, reaffirming that events of extreme rainfall are more likely Located along the North American Gulf Coast of Mexico, Harris County
to occur in the future. Further, results showed that events with longer (Figure 1) encompasses the greater Houston area, which lies approxi
return periods are expected to change the most. This is imperative for mately 50 miles inland from Galveston within the Gulf Coastal Plain and
the City of Houston and Harris County, since Hurricane Harvey was the experiences a complicated land-sea breeze circulation due to its diverse
third 500-year flood in three years in the area at the time of its occur coastline (Caicedo et al., 2019; Darby, 2005; Day et al., 2010; Kocen,
rence. Wang et al. (2019) derives and tests a new model of urban 2013;). The local climate is humid subtropical and is characterized by
flooding prediction under extreme precipitation events, using the 2016 springtime supercell thunderstorms, hot humid summers, and temperate
Louisiana flood as a test bed. Using parameters such as elevation, land winters (https://www.britannica.com/place/Houston). The single daily
cover, soil texture, groundwater level, precipitation, and initial soil rainfall extreme over the entire study period is 536.4 mm, recorded at
moisture content, the model appears to perform best at a resolution of the intersection of Cedar Bayou with U.S. 90 on 27 August 2017. This
100 m. While the model suffers some biases based upon input data, it rainfall is associated with Hurricane Harvey. A table of multiday
reproduces the reverse flow phenomenon particularly well and could be maximum rainfall values is given in Table 1. Most notably, each mul
useful in future studies regarding urban flood prediction in the study tiday total is associated with Hurricane Harvey, just as is the single daily
area. rainfall extreme.
Another modeling study by Didovets et al. (2019) assesses possible Often, winds from the southeast prevail, indicating the dominance of
climate change impacts on flooding and flood frequency in the Carpa the sea-bay breeze circulation (Shepherd et al., 2010). It covers a total
thain region of Central and Eastern Europe using five General Circula area of approximately 4,412 km2 and has a population of 4,713,325 as of
tion Models (GCMs) and two climate scenarios (e.g., RCP4.5 and July 2019 (https://www.census.gov/quickfacts/harriscountytexas).
RCP8.5). Generally, the findings were highly uncertain due to both a The population density is approximately 1,068 people km− 2, illustrating
relatively small number of ensemble members and to limited climate that population dispersion and urban sprawl are prevalent characteris
data availability in the region. The analysis in this study could techni tics. In addition to this, the study area is low-lying and flat with. The
cally be reproduced following the same procedure using our own input elevation of the Houston downtown area is about 15 m above sea level
data, but no direct comparison is possible due to huge differences in (asl), while the northwestern portion of the area exhibits the highest
geographical features. Further, the method is applicable for areas with point of about 46 m. Thus, the average elevation is 36 m asl. The urban
sparse monitoring stations, whereas the HCFCD FWS stations are dense. sprawl is vast, lending to an already burgeoning flooding problem in
For example, the Carpathian region is mountainous, located at notice Houston. This flood risk is exacerbated by a lack of zoning laws in Harris
able distances from oceans and has pronounced seasons including County, leading to at least five major business districts and development
extended snow-covered winters, which likely contributes regional within the 100-year floodplain. Increases in population, temperature,
flooding scenarios from snowmelt. Houston, on the other hand, is and impervious surface coverage demonstrate the significance of urban
topographically flat, located close to the Gulf of Mexico and has a sub sprawl in Houston. A recent citizen-science campaign found an
tropical climate. Using a hydrological modeling approach, Li et al. urban-rural temperature difference of over 8.3◦ C (17◦ F) (https://www.
(2020) evaluated the impacts of urbanization, antecedent rainfall events h3at.org/) largely due to highly developed urban environments and
(AREs) and storm tracks on streamflow during Hurricane Harvey. differing amounts of tree canopy coverage citywide. In addition, a pre
Findings included that (1) urbanization accelerated the Addicks and vious reliance on groundwater has led to increased subsidence in the
Barker Reservoirs reaching their peak elevations; (2) AREs closer to a area, which could further exacerbate the risk of flooding if the trend in
storm event are more impactful; and (3) the storm track could easily subsidence continues in the future. This has been corrected by elimi
have been more adverse for Houston and caused worse flooding than nating the use of groundwater and instituting the use of man-made
actually occurred. Given that extreme events are projected to increase in reservoirs (e.g., Lakes Houston and Conroe), but the land subsidence
the future, as found by Wang et al. (2019), each of these findings still persists today to some degree (Buckley, 2003; Kearns et al, 2015).
highlights the importance of this work to decision makers, including city There has been a wealth of literature produced concerning Harris
planners and engineers, in the area. This importance must be empha County, or, more generally, Houston, Texas, regarding air quality
sized, as flooding in the Houston area is not only confined to extreme (Banta et al., 2005; Jin, 2005; Liu et al., 2015; Rappenglueck et al.,
single events, but also occurs with heavy rainfall whose frequency tends 2008), flooding (Bass et al., 2017; Brody et al., 2015), sea breeze (Chen
to increase according to our findings and may likely increase the fre et al., 2011), storm surge (Davlasheridze et al., 2018; Torres, 2015),
quency of flooding events. tropical cyclones (D’Sa et al., 2018; Fang et al., 2014; Kao et al, 2019;
In this study, we investigate the shifting patterns exhibited in rainfall Trepanier and Tucker, 2018; Zhu et al, 2015), the Urban Heat Island
in the highly urbanized city of Houston, Texas, USA, based on daily (UHI) effect (Ganeshan et al., 2013; Streutker, 2002; Streutker, 2003),
precipitation data for a period of thirty years (1989-2018). The city of and urbanization (Brody et al., 2007; Burian and Shepherd, 2005; Khan,
Houston has been impacted by various notable events over the last de 2005; Shepherd and Burian, 2003).
cades, e.g. Tropical Storm Allison and Hurricane Harvey (e.g., Blood, Concerning flooding, the greater Houston area has notably experi
2014; Kao et al., 2019), which may become a more prevalent feature of a enced several 100- and 500-year floods in the last decade, spawning
potential consequence of a future warming climate. Results of this study studies on Tropical Storm Allison in 2001 (Fang et al., 2014), the Me
will be pertinent to city planners and engineers for many coastal urban morial Day Flood in 2015 (Bass et al., 2017), Hurricane Harvey in 2017
areas and beyond as adaptation for a changing climate becomes more (van Oldenborgh et al., 2017; Kao et al., 2019), and Tropical Storm
pressing. Imelda in 2019 (van Oldenborgh, 2019), the latter two of which are
rapid attribution studies. Additionally, some papers have investigated
specific watersheds in the study area (Bass et al., 2017; Zhu et al., 2015).
However, a spatiotemporal trend analysis over a 30-year time period
2
- M.D. Statkewicz et al. Environmental Advances 5 (2021) 100073
Fig. 1. Location of Harris County, Texas, relative to the State of Texas and the Gulf Coast.
county’s Flood Warning System (FWS), a densely populated network
Table 1
(Fig. 2) of instruments monitoring rainfall, water levels, wind speed and
Maximum multiday rainfall totals and associated events
direction, pressure, temperature, humidity, and other parameters (https
Duration Dates Rainfall (mm) Associated Event ://www.harriscountyfws.org/About). Daily precipitation data were
2 26-27 August 2017 589.3 collected from all 185 rain gauges and screened further for suitability for
3 26-28 August 2017 765.0
Hurricane Harvey (2017) analysis. This included, several steps. Initially, the data were checked for
4 25-28 August 2017 845.3
temporal coverage of the study period (1989-2018). Any station that
5 26-30 August 2017 866.6
began collecting data after 1 January 1989 was removed from the data
set. Through this process, 116 stations were eliminated. Of the 69
using different rainfall classification bins has never been performed. remaining stations, only those with at least 90% of daily precipitation
data availability over the study period were included in the final anal
2.2. Materials and methods ysis. In addition, stations with outliers attributable to instrumentation
malfunction were excluded. Often, these sites exhibited extreme
The Harris County Flood Control District (HCFCD) operates the anomalous values which did not occur at any other site on the same date.
Fig. 2. Locations of all 185 rain gauge stations (dark) maintained by the Harris County Flood Control District’s Flood Warning System and those selected for
analysis (light).
3
- M.D. Statkewicz et al. Environmental Advances 5 (2021) 100073
A total of 63 stations were ultimately retained yielding a very dense rain and to account for spatial correlation between monitoring stations (Ly
gauge network of ~70 km2 station− 1. et al., 2013). More specifically, universal kriging was employed so as to
Thresholds for classification of wet days by rainfall amount vary account for the local prevailing wind pattern.
worldwide. For example, the Indian Meteorological Department (IMD) It should be noted that interpolation itself is not necessary but has
define in their glossary five spatial (e.g., widespread, scattered, isolated, been employed to visualize the spatial distribution of any detected
etc.) and ten temporal (e.g., trace, light, very light, etc.) definitions of trends. No statistical significance, or any analysis therein, is either
rainfall (http://www.imdgoa.gov.in/index.php?option=com_content implied, investigated, or displayed, and all commentary on these figures
&view=article&id=271). In New Zealand, heavy rainfall occurs any is observational in nature.
time more than 100 mm of rainfall are recorded in a 24-hour period. In
past literature, the definition of extreme precipitation as being 50.4 mm (2 3. Results and discussion
inches) of rainfall in one day is well established (Groisman, 1999; Karl
and Knight, 1998). For the purposes of our analysis, we defined a dry day 3.1. Time series analysis
(R0) to be a day on which no rainfall was recorded. Conversely, we
defined a wet day to be a day on which rainfall was recorded. This In Fig. 3, the left panel depicts a scatter plot of daily rainfall depth
contrasts starkly with the IMD’s definition of a rainy day as one that (mm) where each daily measurement in the study period is the median
records 2.5 mm of rainfall. Wet days were further classified by daily value of rainfall experienced at each station employed in this analysis.
rainfall totals using specified rainfall depths (RN) where N = 0, 10, 20, Many of the days plotted were, by definition herein, dry, and the ma
30, 40, 50, and 100 mm, respectively. This methodology is the same that jority of daily measurements appear to fall near or beneath 50 mm. Days
is used by the Expert Team on Climate Change Detection and Indices with rainfall amounts exceeding 50 mm are notably fewer and increase
(ETCCDI) in calculating their 27 climate indices (see http://etccdi.pacifi in sparsity with increasing daily rainfall amount. Further, the days
cclimate.org/list_27_indices.shtml). Each class was analyzed for fre exceeding 100 mm of rainfall produce a database from which to choose
quency of occurrence as well as for percentage contribution to total individual case studies for future work in attribution analysis where the
annual precipitation. Annual precipitation totals were also investigated GEV and GP distributions may be employed. The overall maximum daily
for trend. rainfall plotted measures 399.3 mm on 27 August 2017 and is associated
To perform a detailed statistical analysis, we employed the Mann- with Hurricane Harvey. This measurement is notably less than the global
Kendall Test (Gilbert, 1987; Kendall, 1975; Mann, 1945) to detect sta maximum value associated with the same date (536.4 mm) due to the
tistically significant trends and the Sen’s slope method (Sen, 1968; use of the daily median values.
Theil, 1950) to distinguish the magnitude of trends. Although several On the right of Fig. 3, the time series of total annual precipitation is
extreme value attribution studies have made use of much more complex shown to vary heavily yet generally increase over the study period, most
statistical techniques (e.g., generalized extreme value (GEV) and likely due to natural variability (e.g., ENSO). Notably, the overall min
generalized Pareto (GP) distributions) (Coles, 2001) focusing on various imum occurred in 2011 with a total annual rainfall amount of 542.5 mm
single-event case studies, including the Houston area (van der Wiel, and is due to a drought, unprecedented in its severity, experienced by
2017; van Oldenborgh, 2017; van Oldenborgh, 2019), the use of the the whole state of Texas during that year (Nielsen-Gammon, 2012). The
aforementioned methods in statistical analysis of time series data is well increasing trend of 6.6 mm in annual totals most likely owes itself to the
established (Gocic and Trajkovic, 2013; Hodgkins et al., 2019; Partal increase in extreme events exhibited (e.g., Memorial Day flood, 2015;
and Kahya, 2006; Sayemuzzaman and Jha, 2014). Tax Day flood, 2016; Hurricane Harvey, 2017). The magnitude of trend
The Sen’s slopes derived from the statistical analysis were then im from the slope of simple linear regression (SLR) is nearly twice that
ported into ArcMap to visualize the spatial characteristics of any trends estimated by the Sen’s slope.
detected. Interpolating trends to locations where no gauge stations are For further analysis, several 30-year time series were created (e.g.,
available is the basis of proximity interpolation and is concretely R0, 90th percentile, etc.). This was accomplished by calculating these
established in the literature as early as the 1910s (Thiessen, 1911). The parameters for each station over each year of the study period and taking
use of spatial interpolation for temporal trends is very well established the median value from the results of all monitoring stations as the final
and is used frequently (Dassou et al., 2016; Guo et al., 2020; Kilibarda, time series. We use the median here to minimize the impact of single
2014; Malik et al., 2019; Malik et al., 2020). In fact, many hydrological extreme events on this data ensemble. The different daily rainfall bins
models require it as a step and have been improved by the spatial are presented in Fig. 4.
interpolation of precipitation data (Ly et al., 2013). Spatial interpolation It is apparent that dry days are decreasing at the expense of increases
methods vary in type (e.g., deterministic, stochastic) as well as in in all other rainfall bins considered in this analysis. The largest trend can
application and include such examples as inverse distance weighted be observed in R10 with an increase of ~0.13 days yr− 1, while the most
(IDW), spline, kriging, etc. In this case, a kriging interpolation method significant increases occur in the R40 and R50 rain fall bins, indicating a
was used due to its ability to give quality predictions with low variance general trend toward more intense rainfall events and that short bursts
Fig. 3. Scatter plot of daily precipitation values (left) and a time series of total annual precipitation with trend (right), both over the period 1989-2018.
4
- M.D. Statkewicz et al. Environmental Advances 5 (2021) 100073
Fig. 4. Time series plots of dry days and rainfall bins R10, R20, R30, R40, and R50 for the period 1989-2018.
of rainfall are becoming more frequent. For R50, the positive slope is relatively short period of study. The SLR slope of ~0.28 mm yr− 1 for the
~0.06 days yr− 1. While it is about a factor of two smaller than R10, this 95th percentile is nearly twice that of the 90th. It also shows a more
rainfall bin includes 5 times less rainfall than R10 and indicates that significant increase. This analysis further supports a likely trend toward
extreme events may increase further in the future and possibly dominate a higher frequency of heavier rainfall events, which will eventually
annual totals. Notably, days in some rainfall bins occur more often than overtake a larger fraction of other rainfall bins.
do others. For example, R10 occurs the most frequently, while R50 oc The trends in each parameter, as well as their p-value, Sen’s Slope,
curs least. and SLR slope, are presented in Table 2. Both slopes are measured in mm
Trends in the 90th and 95th percentiles of precipitation can be seen in yr− 1. The most statistically significant trends exist in the 99th percentile,
Fig. 5. These parameters are more robust in depicting a temporal trend maximum, and R100 parameters with p-values of 0.07, 0.08, and 0.11,
than is the 99th percentile, which is highly event-driven over such a respectively. There are many instances where the Sen’s slope is zero
Fig. 5. Time series plots of the 90th (left) and 95th (right) percentiles of rainfall, both over the period 1989-2018.
5
- M.D. Statkewicz et al. Environmental Advances 5 (2021) 100073
Table 2 Table 3
Results from statistical analysis. Brief summary of interpolation results
Parameter P-Value Sen’s Slope SLR Slope Surface Year Minimum Maximum Range
R0 0.66 -0.13 -0.16 Annual Total (mm) 1989 627.4 1,299.6 672.2
R10 0.32 0.13 0.13 2018 674.2 1,791.5 1,117.3
R20 0.69 0.00 0.04 R10 (days) 1989 13.7 29.4 15.7
R30 0.47 0.05 0.06 2018 13.1 49.8 36.7
R40 0.33 0.00 0.06 R50 (days) 1989 1.0 6.1 5.1
R50 0.18 0.00 0.06 2018 3.4 10.1 6.7
R100 0.11 0.00 0.03
P90 0.34 0.11 0.14
P95 0.24 0.16 0.28 this point in time, urban sprawl was not nearly as widespread, and
P99 0.07 0.68 1.31 development rates were not quite as intense. Although the minimum
Total 0.50 3.72 6.55
Minimum 0.49 0.00 0.00
number of days has remained nearly unchanged, the occurrence of R10
Q1 0.90 0.00 0.00 days has more than doubled, as mentioned above. Meanwhile, the
Median 0.24 0.00 0.02 minimum and maximum number of R50 days have increased, though
Mean 0.30 0.04 0.09 the range of values has not changed much. This reflects an overall in
Q3 0.27 0.05 0.07
crease in the number of R50 days in the region.
Maximum 0.08 1.03 2.42
Of all maps, the prevailing wind pattern is most visible in the number
of R10 days and the total rainfall, both for 2018. However, its prevalence
while the SLR slope is nonzero, and, in general, the SLR slope exceeds is apparent in all panels. Generally, its influence is more visible in 2018
the Sen’s slope. than in 1989, which is in line with the findings of Liu et al. (2015) that
the sea breeze circulation has intensified by a factor of ~2.5. Further, a
pattern of high values over the coast, city, and downwind areas emerges
3.2. Geospatial analysis
most pointedly in R10 days and total rainfall in 2018. These findings are
in accordance with urban areas interacting with storm systems and
Fig. 6 shows the geospatial distribution of total rainfall in years 1989
potentially bifurcating them (Bornstein and Lin, 2000; Trevino, 2012).
(right) and 2018 (left). The appropriateness of the universal kriging
Additionally, a study by Liu and Niyogi (2019) found that urban rainfall
method is apparent in each interpolated surface due to patterns
modification enhances rainfall by 18% downwind of the city, 16% over
exhibited along the prevailing wind’s line of propagation. A brief sum
the city, 2% to the left of the city, and 4% right of the city, all within a
mary of minimum and maximum values is included in Table 3 to com
radius of 20-50 km from the city center.
plement the legend of each Harris County map.
Liu and Niyogi (2019) found that urban precipitation is enhanced by
Certainly, choosing only two years for geospatial analysis would not
up to 19% downwind of a city center. Considering a prevailing wind
necessarily imply a trend, per se. However, both years represent the first
direction from the southeast for the Houston area (Shepherd et al., 2010)
and the final year of the period 1989-2018 presented in Fig. 6, which
the main city downwind area would be found in western part of Harris
shows the overall trend in total annual precipitation. Also, both years are
county. A recent study documented an intensification of the sea-bay
not among any extreme years during this period. Table 3 indicates some
breeze circulation for the Houston area (Liu et al., 2015). This in turn
notable results: (i) the range of annual precipitation increased by about
would lead to an enhanced advection of humid air from the Gulf of
66% despite the relatively small change in the minimum value (7%), and
Mexico, coinciding with the aforementioned observations (Liu et al.,
the increase in the range of this parameter is primarily due to the in
2015; Liu and Niyogi, 2019; Shepherd et al., 2010) and suggesting
crease in the maximum annual rainfall, which has increased by about
increased extreme rainfall events with an intensification of the sea-bay
38%; (ii) it should be noted that this increase is not due to single-event
breeze.
extreme rainfall. According to Table 3, the maximum number of inter
mediate rainfall days R10 and R50 rose by 69%. It should be mentioned
that these days, albeit not defined as extreme, already have the potential 3.3. Land cover change analysis
to trigger flooding in low-lying areas and contribute to higher frequency
of occurrence and warrant protective measures. Urbanization has a marked effect on flooding and is suspected to
Additionally, the early stages of urbanization are exhibited in the affect precipitation and the hydrological cycle as a whole on a greater
total rainfall distribution of 1989, where the highest values of nearly scale as well. A study by Brody et al. (2008) examined the relationship
1,300 mm are localized over what is now the core urban area in Hous between the built environment and flood impacts and found that Texas
ton. By 2018, the emergence of the UHI effect is visible in that the consistently sustains the most damage from flooding when compared
coastal and downwind areas receive the most rainfall. with any other state in the United States. Additionally, Harris County
The geospatial distribution of R10 and R50 days are shown in Fig. 7. has seen a explosive population growth during the period of study, with
Again, the relative lack of the UHI effect is visible in both maps for 1989, growth of 41% since 2000 and 70% since 1990. Over 80% of the pop
reflecting once more the early stages of development in the county. At ulation growth since 2000 has occurred in the county’s unincorporated
Fig. 6. Interpolated surfaces depicting total rainfall amounts (mm) in 1989 (left) and 2018 (right) were created using a universal kriging method.
6
- M.D. Statkewicz et al. Environmental Advances 5 (2021) 100073
Fig. 7. Interpolated surfaces depicting the number of rainfall days in the R10 (top) and R50 (bottom) bins were created using a universal kriging method for years
1989 (left) and 2018 (right).
area (Harris County Budget Management Department, 2019), which create a map using methods within ArcGIS 10.7 to illustrate the complex
reflects and emphasizes the extent to which urban sprawl characterizes nature of land cover change in the Harris County area.
the area. Consequently, overall land cover change in Harris County has Before any analysis was performed, it was first necessary to derive all
been significant. According to the Houston-Galveston Area Council of the areas (km2) of each land cover change type from the available
(H-GAC), developed land increased from 46.3% in 1996 to 56.9% in NLCD data. This task was completed using the Tabulate Area tool within
2011. In this same time frame, agricultural land cover decreased from the ArcGIS 10.7 (Esri Inc., 2019) software package. For example, the
18.1% to 14.3%; forest decreased from 19.3% to 13.8%; and wetlands area of Harris County was calculated to be approximately 4,604 km2
decreased from 8.8% to 7.7% (H-GAC). Although naturally occurring using this tool and the shapefile HCAD Harris County Boundary (see htt
wetlands decreased by only 1.1%, it is still worth noting due to the ps://geo-harriscounty.opendata.arcgis.com/datasets/hcad-harris-c
crucial role they play in mitigating flood damage (Zahran et al., 2007). ounty-boundary). It should be noted that 4,604 km2 will be the total area
Streutker (2003) used satellite data from Advanced Very upon which all calculations and tabulated values are based (Table 4)
High-Resolution Radiometer (AVHRR) to determine that the urban heat because it is approximately 8 km2 less than the area reported by the
island effect of Houston experienced a mean growth of 35% between two Census.
time periods (1985-1987 and 1999-2001) based on nighttime scene Between 2001 and 2016, most of the change that occurred in Harris
analyses. In a prior analysis, Streutker (2002) characterized the County occurred in the periphery of the downtown area. Indicated in
magnitude and spatial extent of Houston’s UHI. During the same time dark green, 83.2% (3,829.1 km2) of the land area has seen no change.
period the mean area of that urban heat island may have risen between Many of these areas (e.g., West University Place, Bunker Hill Village,
38-88% according to the same study. Khan (2005) used a neural network Pasadena, Baytown, indicated in Fig. 8) have already been developed or
technique to show that asphalt and concrete increased 21% from 1984 to existed as a park (e.g., Bear Creek Park near Addicks and Barker Res
1994, 39% from 1994 to 2000, and 114% from 2000 to 2003, illus ervoirs) or cultivated area. The greatest difference can be seen in that of
trating that both urbanization and the rate of urbanization were urban change, indicating that 12.6% (579.8 km2) of Harris County
increasing as total vegetative cover decreased. Shepherd and Burian changed from or to one of the four urban land cover classes (e.g.,
(2003) presented evidence that the urban heat island’s influence is of developed: open space, low intensity, medium intensity, high intensity).
primary significance in causing the observed precipitation anomalies in
the Houston area.
Table 4
As the area of impervious surface coverage increases, there is a
Areal analysis of NLCD land cover change index
corresponding decrease in infiltration and increase in surface run-off
Description Area Area
(Dunne and Leopold, 1978; Paul and Meyer, 2001). As the percent
(km2) (%)
drainage basin impervious surface cover increases 10-20%, run-off in
creases two-fold (Arnold and Gibbons, 1996). Greater surface run-off No Change 3829.1 83.2
Change from or to Water 40.1 0.9
volume tends to result in increased frequency and severity of flooding.
Change from or to any of the four Urban classes 579.8 12.6
The built environment is also linked to higher peak discharges (Brezonik Change from Herbaceous Wetland to Woody Wetland, or 63.7 1.4
and Stadelmann, 2002; Burges et al., 1998; Leopold, 1994) where lag vice versa
time is compressed, leading to floods peaking more rapidly. Change from or to Herbaceous Wetland 2.7 0.1
Change from Cultivated Crops to Hay/Pasture, or vice 8.3 0.2
In the US, one repository for descriptive land cover data at a 30 m
versa
resolution is the National Land Cover Database (NLCD) that is released Change from or to Cultivated Crops 1.6 0.0
multiple times per decade by the U.S. Geological Survey (USGS) and is Change from or to Hay/Pasture 27.4 0.6
available publicly through the Multi-Resolution Land Characteristics Persistent Grass or Shrubland Change 0.3 0.0
Consortium (MRLC) (see https://www.mrlc.gov/data). For the purposes Change from or to Barren 0.8 0.0
Change from or to any of the three Forest classes 50.7 1.1
of this analysis, the NLCD Land Cover Change Index product was used to
Change from or to Woody Wetland 0.1 0.0
7
- M.D. Statkewicz et al. Environmental Advances 5 (2021) 100073
Fig. 8. NLCD land cover change index for Harris County, TX.
Given the population growth in the area of study, it can be deduced increase while varying heavily over the study period. Dry days are either
that the change was to, not from, urban land cover. The locations of decreasing at the expense of all other rainfall bins or are increasing due
urban change (e.g., Katy, Tomball, Spring, Aldine, Humble, Sheldon to more rainfall falling in less days overall. Geospatially, the dominance
indicated in Fig. 8) reflect that ongoing urbanization continues to occur of the prevailing wind pattern is exhibited by greater annual rainfall
outside of the core urban area already established. This is indicative of amounts, more R10 and R50 days along its path, and larger ranges of
the urban sprawl already characteristic of Houston spreading to the values for all three. Analysis of land cover change revealed that 12.6%
unincorporated areas of Harris County and is emphasized by smaller (579.8 km2) of the county has urbanized between 2001 and 2016,
percentage changes in other land cover classes such as any type of particularly in the peripheral area of the city’s urban core. This has
wetland or forest. Notably, the construction of the Grand Parkway is occurred at the expense of wetlands and forests, which experienced
visible in the northwestern corner of the county. changes of 1.4% (63.7 km2) and 1.1% (50.7 km2), respectively. The
The spatial distribution of wetlands likely changes with regional results of this study are crucial for residents, elected officials, city
hydrology, which will undoubtedly be affected by urbanization and planners, and engineers, not only for the Houston area, but for numerous
urban sprawl. Wetlands within class change account for only 1.4% (63.7 low-lying coastal cities worldwide, most critical for those many rapidly
km2), but wetlands are crucial for flood control and mitigating flood growing megacities in tropical regions (e.g. Jakarta, Manila, Dhaka, to
damage (Zahran et al., 2007). Flooding of these wetlands can lead to name a few). Options for future work include case studies from R100
changes in soil moisture, vegetation, and further alter other land cover days, an expanded analysis of land cover change, and investigations of
classes (e.g., herbaceous). changes in other meteorological parameters in Harris County.
Changes from or to any of the three forest classes similarly account
for a small percentage of 1.1% (50.7 km2), but changes in tree canopies Authors’ contribution
have larger implications on the city’s Urban Heat Island effect and its
intensity. The change in forests may be due to land clearing (e.g., forest M. Statkewicz: Formal analysis, Methodology, Writing - review &
to barren land) and urbanization (e.g., barren land to any of the four editing. R. Talbot: Conceptualization, Project Initiation. B. Rappenglück:
urban classes). These changes seem to occur concurrently with changes Investigation, Writing - review & editing.
in wetlands and in urban classes, which could imply an overall inverse
relationship between tree canopies and urbanization locally. Funding
The remainder of changes amount to 0.89% (41.2 km2) and are too
small to be speculated upon at this point. In future work, the full suite of This research was funded by the HuRRI (Hurricane Resilience
NLCD Land Cover products (e.g., urban imperviousness, urban de Research Institute) Seed Grants Program (SGP) provided by the Uni
scriptors, tree canopy) will be employed to delve into when and, versity of Houston Division of Research.
perhaps, why these changes have occurred between 2001 and 2016.
Declaration of Competing Interest
4. Conclusions
The authors declare that they have no known competing financial
This study analyzes daily and annual rainfall totals as well as days
interests or personal relationships that could have appeared to influence
that receive particular rainfall amounts in a 24-hour period in the highly
the work reported in this paper.
urbanized area of Houston, TX, over a period of three decades (1989-
2018). According to the results in Figure 7 and Table 3, the only negative
Acknowledgments
trend is in R0 days, while the most statistically significant trends exist in
the 99th percentile, maximum, and R100 parameters with p-values of
We appreciate data provided by the Harris County Flood Control
0.07, 0.08, and 0.11, respectively. Total annual precipitation appears to
District (HCFCD) and would like to thank each of the reviewers for their
8
- M.D. Statkewicz et al. Environmental Advances 5 (2021) 100073
commitment to the betterment of this manuscript. The time spent and Gilbert, R.O., 1987. Statistical Methods for Environmental Pollution Monitoring. Wiley,
NY.
effort put forth has been imperative in the creation of this document in
Gocic, M., Trajkovic, S., 2013. Analysis of changes in meteorological variables using
its current form. Mann-Kendall and Sen’s slope estimator statistical tests in Serbia. Global Planet.
Change 100, 172–182. https://doi.org/10.1016/j.gloplacha.2012.10.014.
References Groisman, P.Ya., Karl, T.R., Easterling, D.R., Knight, R.W., Jamason, P.F., Hennessy, K.J.,
Suppiah, R., Page, C.M., Wibig, J., Fortuniak, K., Razuvaev, V.N., Douglas, A.,
Førland, E., Zhai, P.-M, 1999. Clim. Change 42 (1), 243–283. https://doi.org/
Allan, R.P., Soden, B.J., 2008. Atmospheric Warming and the Amplification of 10.1023/a:1005432803188.
Precipitation Extremes. Science 321 (5895), 1481–1484. https://doi.org/10.1126/ Guo, B., Zhang, J., Meng, X., Xu, T., Song, Y., 2020. Long-term spatio-temporal
science.1160787. precipitation variations in China with precipitation surface interpolated by
Arnold Jr., C.L., Gibbons, C.J, 1996. Impervious Surface Coverage: The Emergence of a ANUSPLIN. Sci. Rep. 10 (1) https://doi.org/10.1038/s41598-019-57078-3.
Key Environmental Indicator. J. Am. Plann. Assoc. 62 (2), 243–258. https://doi.org/ Harris County Budget Management Department (2019). Fiscal Year 2020 Population
10.1080/01944369608975688. Report. Retrieved from https://budget.harriscountytx.gov.
Banta, R.M., Senff, C.J., Nielsen-Gammon, J., Darby, L.S., Ryerson, T.B., Alvarez, R.J., Hodgkins, G.A., Dudley, R.W., Archfield, S.A., Renard, B., 2019. Effects of climate,
Sandberg, S.P., Williams, E.J., Trainer, M., 2005. A Bad Air Day in Houston. Bull. regulation, and urbanization on historical flood trends in the United States.
Am. Meteorol. Soc. 86 (5), 657–670. https://doi.org/10.1175/bams-86-5-657. J. Hydrol. 573, 697–709. https://doi.org/10.1016/j.jhydrol.2019.03.102.
Bass, B., Juan, A., Gori, A., Fang, Z., Bedient, P., 2017. 2015 Memorial Day Flood Impacts IPCC, 2013. Climate Change 2013: The Physical Science Basis. In: Stocker, T.F., Qin, D.,
for Changing Watershed Conditions in Houston. Nat. Hazard. Rev. 18 (3), 05016007 Plattner, G.-K., Tignor, M., Allen, S.K., Boschung, J., Nauels, A., Xia, Y., Bex, V.,
https://doi.org/10.1061/(asce)nh.1527-6996.0000241. Midgley, P.M. (Eds.), Contribution of Working Group I to the Fifth Assessment
Blood, P., 2014. The optimization of southeast Texas extreme rainfall prediction utilizing Report of the Intergovernmental Panel on Climate Change. Cambridge University
the Weather Research and Forecast - Environmental Modeling System. MS Thesis. Press, Cambridge, United Kingdom and New York, NY, USA, p. 1535. https://doi.
University of Houston. org/10.1017/CBO9781107415324.
Bornstein, R., Lin, Q., 2000. Urban heat islands and summertime convective Jin, M., 2005. Urban aerosols and their variations with clouds and rainfall: A case study
thunderstorms in Atlanta: three case studies. Atmos. Environ. 34 (3), 507–516. for New York and Houston. J. Geophys. Res. 110 (D10) https://doi.org/10.1029/
https://doi.org/10.1016/s1352-2310(99)00374-x. 2004jd005081.
Brezonik, P.L., Stadelmann, T.H., 2002. Analysis and predictive models of stormwater Kao, S.-C., DeNeale, S.T., Watson, D.B., 2019. Hurricane Harvey Highlights: Need to
runoff volumes, loads, and pollutant concentrations from watersheds in the Twin Assess the Adequacy of Probable Maximum Precipitation Estimation Methods.
Cities metropolitan area, Minnesota, USA. Water Res. 36 (7), 1743–1757. https:// J. Hydrol. Eng. 24 (4), 05019005 https://doi.org/10.1061/(asce)he.1943-
doi.org/10.1016/s0043-1354(01)00375-x. 5584.0001768.
Brody, S.D., Zahran, S., Highfield, W.E., Grover, H., Vedlitz, A., 2007. Identifying the Karl, T.R., Knight, R.W., 1998. Secular Trends of Precipitation Amount, Frequency, and
impact of the built environment on flood damage in Texas. Disasters 32 (1), 1–18. Intensity in the United States. Bull. Am. Meteorol. Soc. 79 (2), 231–241.https://doi.
https://doi.org/10.1111/j.1467-7717.2007.01024.x. org/10.1175/1520-0477(1998)0792.0.co;2
Brody, S.D., Sebastian, A., Blessing, R., Bedient, P.B., 2015. Case study results from Kearns, T.J., Wang, G., Bao, Y., Jiang, J., Lee, D., 2015. Current Land Subsidence and
southeast Houston, Texas: identifying the impacts of residential location on flood Groundwater Level Changes in the Houston Metropolitan Area (2005–2012). J. Surv.
risk and loss. J. Flood Risk Manage. 11, S110–S120. https://doi.org/10.1111/ Eng. 141 (4), 05015002 https://doi.org/10.1061/(asce)su.1943-5428.0000147.
jfr3.12184. Kendall, M.G., 1975. Rank Correlation Methods, 4th edition. Charles Griffin, London.
Buckley, S.M., 2003. Land subsidence in Houston, Texas, measured by radar Khan, S.D., 2005. Urban development and flooding in Houston Texas, inferences from
interferometry and constrained by extensometers. J. Geophys. Res. 108 (B11) remote sensing data using neural network technique. Environ. Geol. 47 (8),
https://doi.org/10.1029/2002jb. 1120–1127. https://doi.org/10.1007/s00254-005-1246-x.
Burges, S.J., Wigmosta, M.S., Meena, J.M., 1998. Hydrological Effects of Land-Use Kilibarda, M., Hengl, T., Heuvelink, G.B.M., Gr¨ aler, B., Pebesma, E., Perˇcec Tadi´c, M.,
Change in a Zero-Order Catchment. J. Hydrol. Eng. 3 (2), 86–97. https://doi.org/ Bajat, B, 2014. Spatio-temporal interpolation of daily temperatures for global land
10.1061/(asce)1084-0699(1998)3:2(86. areas at 1 km resolution. J. Geophys. Res.: Atmospheres 119 (5), 2294–2313.
Burian, S.J., Shepherd, J.M., 2005. Effect of urbanization on the diurnal rainfall pattern https://doi.org/10.1002/2013jd020803.
in Houston. Hydrol. Processes 19 (5), 1089–1103. https://doi.org/10.1002/ Kocen, M., 2013. Observations of Sea-Breeze Fronts along the Houston Gulf Coast. MS
hyp.5647. Thesis. University of Houston.
Caicedo, V., Rappenglueck, B., Cuchiara, G., Flynn, J., Ferrare, R., Scarino, A.J., Leopold, L.B., 1994. A View of the River. Harvard University Press, Cambridge, MA.
Berkoff, T., Senff, C., Langford, A., Lefer, B., 2019. Bay Breeze and Sea Breeze Liu, J., Niyogi, D., 2019. Meta-analysis of urbanization impact on rainfall modification.
Circulation Impacts on the Planetary Boundary Layer and Air Quality From an Sci. Rep. 9 (1) https://doi.org/10.1038/s41598-019-42494-2.
Observed and Modeled DISCOVER-AQ Texas Case Study. J. Geophys. Res.: Liu, L., Talbot, R., Lan, X., 2015. Influence of Climate Change and Meteorological Factors
Atmospheres. https://doi.org/10.1029/2019jd030523. on Houston’s Air Pollution: Ozone a Case Study. Atmosphere 6 (5), 623–640.
Chen, F., Miao, S., Tewari, M., Bao, J.-W., Kusaka, H., 2011. A numerical study of https://doi.org/10.3390/atmos6050623.
interactions between surface forcing and sea breeze circulations and their effects on Ly, S., Charles, C., Degr´e, A., 2013. Different methods for spatial interpolation of rainfall
stagnation in the greater Houston area. J. Geophys. Res. 116 (D12) https://doi.org/ data for operational hydrology and hydrological modeling at watershed scale: a
10.1029/2010jd015533. review. Biotechnol. Agron. Soc. Environ. 17 (2), 392–406.
Coles, S., 2001. An Introduction to Statistical Modeling of Extreme Values. Springer Malik, A., Kumar, A., Guhathakurta, P., Kisi, O., 2019. Spatial-temporal trend analysis of
Series in Statistics. Springer, London. https://doi.org/10.1007/978-1-4471-3675-0. seasonal and annual rainfall (1966–2015) using innovative trend analysis method
Darby, L.S., 2005. Cluster Analysis of Surface Winds in Houston, Texas, and the Impact of with significance test. Arabian J. Geosci. 12 (10) https://doi.org/10.1007/s12517-
Wind Patterns on Ozone. J. Appl. Meteorol. 44 (12), 1788–1806. https://doi.org/ 019-4454-5.
10.1175/jam2320.1. Malik, A., Kumar, A., 2020. Spatio-temporal trend analysis of rainfall using parametric
Dassou, E.F., Ombolo, A., Chouto, S., Mboudou, G.E., Essi, J.M.A., Bineli, E, 2016. Trends and non-parametric tests: case study in Uttarakhand, India. Theoret. Appl. Climatol.
and Geostatistical Interpolation of Spatio-Temporal Variability of Precipitation in 140 (1–2), 183–207. https://doi.org/10.1007/s00704-019-03080-8.
Northern Cameroon. Am. J. Climate Change 05 (02), 229–244. https://doi.org/ Mann, H.B., 1945. Non-parametric tests against trend. Econometrica 13, 163–171.
10.4236/ajcc.2016.52020. Melillo, J.M., Richmond, T., 2014. Climate Change Impacts in the United States: The
Davlasheridze, M., Atoba, K.O., Brody, S., Highfield, W., Merrell, W., Ebersole, B., Third National Climate Assessment. U.S. Global Change Research Program. https://
Purdue, A., Gilmer, R.W., 2018. Economic impacts of storm surge and the cost- doi.org/10.7930/j0z31wj2.
benefit analysis of a coastal spine as the surge mitigation strategy in Houston- Nielsen-Gammon, J.W., 2012. The 2011 Texas Drought. Texas Water J. 3 (1) https://doi.
Galveston area in the USA. Mitigation Adaptation Strat. Glob. Change 24 (3), org/10.21423/TWJ.V3I1.6463 (2012).
329–354. https://doi.org/10.1007/s11027-018-9814-z. Partal, T., Kahya, E., 2006. Trend analysis in Turkish precipitation data. Hydrol. Process.
Day, B.M., Rappenglück, B., Clements, C.B., Tucker, S.C., Alan Brewer, W., 2010. 20 (9), 2011–2026. https://doi.org/10.1002/hyp.5993.
Nocturnal boundary layer characteristics and land breeze development in Houston, Paul, M.J., Meyer, J.L., 2001. Streams in the Urban Landscape. Ann. Rev. Ecol. Syst. 32
Texas during TexAQS II. Atmos. Environ. 44 (33), 4014–4023. https://doi.org/ (1), 333–365. https://doi.org/10.1146/annurev.ecolsys.32.081501.114040.
10.1016/j.atmosenv.2009.01.031. Quinn, N., Wing, Oliver, Smith, Andrew, Sampson, Christopher Charles, Neal, Jeffrey C.,
D’Sa, E.J., Joshi, I., Liu, B., 2018. Galveston Bay and Coastal Ocean Optical-Geochemical Bates, Paul D., 2017. The generation of a stochastic flood event catalogue for
Response to Hurricane Harvey From VIIRS Ocean Color. Geophys. Res. Lett. 45 (19) continental USA. In: Poster H21J-1615, AGU Fall Meeting. New Orleans, Louisiana.
https://doi.org/10.1029/2018gl079954, 10,579-10,589. Rappenglück, B., Perna, R., Zhong, S., Morris, G.A., 2008. An analysis of the vertical
Dunne, T., Leopold, L.B., 1978. Water in environmental planning. Freeman, New York. structure of the atmosphere and the upper-level meteorology and their impact on
Esri Inc., 2019. ArcMap 10.7.1. Esri Inc., Redlands, CA. Software. surface ozone levels in Houston, Texas. J. Geophys. Res. 113 (D17) https://doi.org/
Fang, Z., Dolan, G., Sebastian, A., Bedient, P.B., 2014. Case Study of Flood Mitigation and 10.1029/2007jd009745.
Hazard Management at the Texas Medical Center in the Wake of Tropical Storm Sayemuzzaman, M., Jha, M.K., 2014. Seasonal and annual precipitation time series trend
Allison in 2001. Nat. Hazard. Rev. 15 (3), 05014001 https://doi.org/10.1061/(asce) analysis in North Carolina, United States. Atmos. Res. 137, 183–194. https://doi.
nh.1527-6996.0000139. org/10.1016/j.atmosres.2013.10.012.
Ganeshan, M., Murtugudde, R., Imhoff, M.L., 2013. A multi-city analysis of the UHI- Sen, 1968. Pranab Kumar Estimates of the regression coefficient based on Kendall’s tau.
influence on warm season rainfall. Urban Climate 6, 1–23. https://doi.org/10.1016/ J. Am. Stat. Assoc. 63, 1379–1389.
j.uclim.2013.09.004.
9
- M.D. Statkewicz et al. Environmental Advances 5 (2021) 100073
Shepherd, J.M., Burian, S.J., 2003. Detection of Urban-Induced Rainfall Anomalies in a rainfall–runoff for the Houston–Galveston region. Coastal Eng. 106, 7–19. https://
Major Coastal City. Earth Interact. 7 (4), 1–17. https://doi.org/10.1175/1087-3562 doi.org/10.1016/j.coastaleng.2015.09.004.
(2003)0072.0.co;2. Trepanier, J., Tucker, C., 2018. Event-Based Climatology of Tropical Cyclone Rainfall in
Shepherd, J.M., Carter, M., Manyin, M., Messen, D., Burian, S., 2010. The Impact of Houston. Atmosphere, TexasMiami, Florida, p. 170. https://doi.org/10.3390/
Urbanization on Current and Future Coastal Precipitation: A Case Study for Houston. atmos9050170, 9.
Environ. Plann. B: Plann. Des. 37 (2), 284–304. https://doi.org/10.1068/b34102t. Trevino, Anna Marie, 2012. Mechanisms of urban influence on precipitation in the
Singh, D., Tsiang, M., Rajaratnam, B., Diffenbaugh, N.S., 2013. Precipitation extremes southeastern United States: precipitation enhancement, storm bifurcation, and
over the continental United States in a transient, high-resolution, ensemble climate synoptic characteristics. LSU Doctoral Dissertations, p. 1893. https://digitalcommon
model experiment. J. Geophys. Res.: Atmospheres 118 (13), 7063–7086. https://doi. s.lsu.edu/gradschool_dissertations/1893.
org/10.1002/jgrd.50543. van der Wiel, K., Kapnick, S.B., van Oldenborgh, G.J., Whan, K., Philip, S., Vecchi, G.A.,
Smith, A.B., 2020. U.S. Billion-dollar Weather and Climate Disasters, 1980 - present Singh, R.K., Arrighi, J., Cullen, H., 2017. Rapid attribution of the August 2016 flood-
(NCEI Accession 0209268) [Data set]. NOAA National Centers for Environmental inducing extreme precipitation in south Louisiana to climate change. Hydrol. Earth
Information. https://doi.org/10.25921/STKW-7W73. Syst. Sci. 21 (2), 897–921. https://doi.org/10.5194/hess-21-897-2017.
Streutker, D.R., 2002. A remote sensing study of the urban heat island of Houston, Texas. van Oldenborgh, G.J., van der Wiel, K., Sebastian, A., Singh, R., Arrighi, J., Otto, F.,
Int. J. Remote Sens. 23 (13), 2595–2608. https://doi.org/10.1080/ Haustein, K., Li, S., Vecchi, G., Cullen, H., 2017. Attribution of extreme rainfall from
01431160110115023. Hurricane Harvey, August 2017. Environ. Res. Lett. 12 (12), 124009 https://doi.org/
Streutker, D., 2003. Satellite-measured growth of the urban heat island of Houston, 10.1088/1748-9326/aa9ef2.
Texas. Remote Sens. Environ. 85 (3), 282–289. https://doi.org/10.1016/s0034-4257 van Oldenborgh, et al., 2019. Rapid attribution of the extreme rainfall in Texas from
(03)00007-5. Tropical Storm Imelda. https://www.worldweatherattribution.org/rapid-attributio
Theil, H., 1950. A rank-invariant method of linear and polynomial regression analysis. I. n-of-the-extreme-rainfall-in-texas-from-tropical-storm-imelda/.
Nederl. Akad. Wetensch. Proc 53, 386–392. = Indagationes Math. 12, 85–91 (1950). Zahran, S., Vedlitz, A., Grover, H., Brody, S.D., Highfield, W.E., 2007. Identifying the
Thiessen, A.H., 1911. Precipitation Averages for Large Areas. Mon. Weather Rev. 39 (7), impact of the built environment on flood damage in Texas. Disasters 32 (1), 1–18.
1082–1089. https://doi.org/10.1175/1520-0493(1911)392.0.co;2. https://doi.org/10.1111/j.1467-7717.2007.01024.x.
Torres, J.M., Bass, B., Irza, N., Fang, Z., Proft, J., Dawson, C., Kiani, M., Bedient, P., 2015. Zhu, L., Quiring, S.M., Guneralp, I., Peacock, W.G., 2015. Variations in tropical cyclone-
Characterizing the hydraulic interactions of hurricane storm surge and related discharge in four watersheds near, 7. Climate Risk Management, Houston,
Texas, pp. 1–10. https://doi.org/10.1016/j.crm.2015.01.002.
10
nguon tai.lieu . vn