Project ID: NTC2016-SU-R-04

SUSTAINABILITY AND SCALING OF URBAN TRANSPORTATION NETWORKS

Final Report

by

Zhihua Wang, PI zhwang@asu.edu; 480-727-2933

Kamil E. Kaloush, Co-PI Kamil.kaloush@asu.edu; 480-965-5509

Arizona State University

Chenghao Wang, Research Associate Arizona State University

for

National Transportation Center at Maryland (NTC@Maryland) 1124 Glenn Martin Hall University of Maryland

College Park, MD 20742

November, 2017

iii

ACKNOWLEDGEMENTS This project was funded by the National Transportation Center @ Maryland (NTC@Maryland), one of the five National Centers that were selected in this nationwide competition, by the Office of the Assistant Secretary for Research and Technology (OST-R), U.S. Department of Transportation (US DOT).

DISCLAIMER The contents of this report reflect the views of the authors, who are solely responsible for the facts and the accuracy of the material and information presented herein. This document is disseminated under the sponsorship of the U.S. Department of Transportation University Transportation Centers Program in the interest of information exchange. The U.S. Government assumes no liability for the contents or use thereof. The contents do not necessarily reflect the official views of the U.S. Government. This report does not constitute a standard, specification, or regulation.

v

TABLE OF CONTENTS

EXCUTIVE SUMMARY ............................................................................................................. 1 1.0 INTRODUCTION............................................................................................................. 3 2.0 DEVELOPMENT OF A SCALING MODEL TO EVALUATE URBAN WARMING WITH POPULATION GROWTH........................................................................ 5

2.1 METHODS ......................................................................................................................... 5 2.1.1 Urban Population Model ............................................................................................. 5 2.1.2 Population–Temperature Regression .......................................................................... 6

2.2 CASE STUDY IN THE PHOENIX METROPOLITAN AREA ....................................... 7 2.2.1 The Study Area ........................................................................................................... 7 2.2.2 Correlating the Urban Climate Change with Urban Population ................................. 7 2.2.3 Simulating the Historical Population Growth ............................................................. 9 2.2.4 Projecting the Future Population Growth ................................................................. 12 2.2.5 Predicting Future Urban Warming............................................................................ 13

3.0 STATISTICAL ANALYSIS IN 10 U.S. MEGAPOLITANS ...................................... 15 4.0 URBAN TRANSPORTATION NETWORKS AND BEYOND ................................. 19

4.1 ROAD NETWORK .......................................................................................................... 19 4.2 OTHER TRANSPORTATION NETWORKS ................................................................. 22

4.2.1 Air Traffic ................................................................................................................. 22 4.2.2 Railroad System ........................................................................................................ 25

5.0 CONCLUSION ............................................................................................................... 29 6.0 REFERENCES ................................................................................................................ 31

LIST OF TABLES

Table 1: Emerging cities during different population growth stages in the PMA. ....................... 10 Table 2: Calibrated parameters in the proposed model for different stages of population growth

in PMA. ................................................................................................................................. 11 Table 3: Summary of predicted temperature increase (in °C) from the reference in 2016. .......... 14 Table 4: General geographic information of the 10 selected major megapolitan areas in the U.S.

............................................................................................................................................... 15 Table 5: Public road length (in miles) by functional system in the United States in 2013. .......... 20 Table 6: Summary of Enplanements at the top 50 U.S. Airports in 2016. ................................... 23 Table 7: Miles of Freight Railroad Operated by Class of Railroad in 2012. ................................ 26

LIST OF FIGURES Figure 1: Conceptual sketch of transportation networks for urban development (Seto et al.,

2012). ...................................................................................................................................... 3

vi

Figure 2: Geographical location of the study area with national land cover (NLCD2011) (a) overlaid in the state of Arizona, and (b) PMA and weather stations. ..................................... 8

Figure 3: Correlation between the AAT and population in the PMA using the (a) linear and (b) exponential regression. ........................................................................................................... 9

Figure 4: Historic record of population growth rate in PMA from 1969 to 2014. ....................... 10 Figure 5: Comparison between predicted and observed historical population in the PMA;

numbering of subplots follows the index in Table 2............................................................. 11 Figure 6: Projected future population in the PMA based on calibrated parameters in the most

recent period of 2012 – 2014. ............................................................................................... 12 Figure 7: Predicted trends of future temperature changes in PMA using different population

projections. ............................................................................................................................ 14 Figure 8: Population growth in the selected megapolitan areas in (a) Sun Corridor, Cascadia,

Northern California, Southern California, and Front Range, and (b) Texas Triangle, Florida, Great Lakes, Piedmont Atlantic, and Northeast during 1969 – 2014. .................................. 16

Figure 9: Same as Figure 8 but for AAT. ..................................................................................... 17 Figure 10: Statistical correlation between the AAT and population in the 10 megapolitan areas in

the period of 1969 – 2014; scatter points are the historical observations, and red line is the linear regression. ................................................................................................................... 18

1

EXCUTIVE SUMMARY

Drastic growth of urban transportation networks call for sustainable planning and designs. The dynamics of transportation networks, as well their manifestations, essentially regulate the growth, economies, scaling, and environment of cities. This study proposed a novel modeling framework for analyzing the historical urban development and its impacts, and at the same time, provided informative predictions for future transportation infrastructure growth. Urban population was selected as the primary indicator, as well as the control scaling parameter if transportation network density, while the urban environmental and socio-economic impacts are measured by the exacerbated urban thermal environment, i.e. urban warming.

Based upon the population–warming correlation, the Phoenix Metropolitan Area was selected as the testbed, and both linear and exponential regression were tested with historical as well as projected temperature and population datasets. The proposed population growth models were validated using the historical demographic data from 1969 to 2014 in the study area. Furthermore, the predicted future population growth from 2016 to 2050 using the proposed model is found in reasonable agreement with the estimates from the Arizona Department of Administration – Office of Employment and Population Statistics for the first 15 years, with higher model uncertainty manifested in longterm projections.

Similar regression analyses were conducted for ten major megapolitan areas in U.S. from 1969 to 2014. Reasonable coefficients of determination ranging from 0.164 (Great Lakes) to 0.563 (Arizona Sun Corridor) suggest that correlation between the annual average temperature and population is statistically significant in the selected megapolitan areas. Results indicate that the proposed model is also applicable for urban areas with different geographical controls.

Based on the urban dynamics derived in this study, statistical datasets for road network, air traffic, and railroad system were collected from multiple sources, for future development of more comprehensive mechanistic models. The developed mechanistic scaling model can then be used for characterizing the intra- and inter-city transportation networks and their impact on sustainable urban growth.

3

1.0 INTRODUCTION

Today, urban areas are home to more than half of the world’s population, with a projected urban population of 6.3 billion (68% total global population) in 2050 (UN, 2012). The concentration of population in urban areas has positively affected the economic growth, spurring entrepreneurship, inventions, and business innovation (Bettencourt and West, 2010). In addition, large cities are often “greener” than rural areas, because people living in denser habitats typically have smaller energy footprints, require less infrastructure and consume fewer resources per capita (Kalnay and Cai, 2003; Bettencourt and West, 2011). Despite these benefits, urban areas confront a number of sustainability challenges, resulting from the complex interaction of infrastructural, economic and social components (Arnfield, 2003; Cash et al., 2003; Patz et al. 2005; Nazaroff, 2013). The problems associated with urban growth, however, are typically treated as independent issues with a lack of proper account of intra-/inter-city “bridging networks” via transportation systems (Figure 1). Urban economies currently generate more than 90% of global gross value added, meaning few peri-urban or rural systems are unaffected by urbanization (Seto et al., 2012). Being the ultimate urban land connectors, malfunction or under-performance of transportation networks may in fact trigger the failure of other systems as these elements are inherently interdependent (Badland and Schofield, 2005; Bettencourt et al., 2007). This issue involves complex system planning of intra- and inter-city road network and the integration of the structure, function, and evolution mechanisms of the system (Batty, 2008; Hou et al., 2015).

Figure 1: Conceptual sketch of transportation networks for urban development (Seto et al., 2012).

In this project, we used Phoenix metropolitan as a hub to study the operation of connecting transportation systems and their environmental impact on the southwest US via the inter-city road network, due to its well-documented historical growth (Jiang and Jia, 2011; Kane et al., 2014). The central hypothesis of this project is that the dynamics of transportation networks, as well as their manifestations, essentially regulate the growth, economies, scaling, and environment of cities. Testing this hypothesis requires a trans-disciplinary approach that integrates analysis of conceptual framework, statistical regression, and mechanistic model. We aim to develop a framework for analyzing the historical urban development and its impacts, and

4

at the same time, providing informative predictions for potential future transformative paradigm. Urban population was selected as the manifestations of the transportation network density, and the exacerbated urban thermal environment, i.e. urban warming, was chosen as an example of the environmental impact.

The population growth is concomitant with the global economic development (Rosa et al., 2004), prompting the urban development via, e.g. technological innovations and reduced footprints of resource allocation per capita. In 1900, only about 40% of the total population was an urban resident in the U.S.; this percentage was over 75% in 1990 (Henderson, 2002). Urban growth and the associated infrastructure development/retrofit, inevitably, lead to adverse environmental issues, such as the urban heat island (UHI) (Oke, 1973), air pollution (Mage et al., 1996), and human health issues (Patz et al., 2005), to name a few. For example, the anthropogenic heat, generated from human metabolism, industries, buildings, vehicles, etc., is one of the main contributors to urban warming (Sailor, 2011). Sailor (2011) reported that the metabolic heat during daytime in U.S. cities is around 1 W m−2. In industrial sectors, sensible heat is the main form of anthropogenic heat as generated by energy consumptions; accurate estimate of its magnitude is imperative but lacking (Sailor, 2011). Heiple and Sailor (2008) pointed out that energy consumptions inside buildings (mainly induced by the heating, ventilation and air conditioning systems) result in significant anthropogenic heat emissions. Additional contribution to the anthropogenic heat (and moisture) arises from vehicular emission as a result of fossil fuel combustion.

As a developed country, the U.S. has several cities of dense population, with more than 61% of the Nation’s population distributed over 23 megapolitan areas (Nelson and Lang, 2011). Karl et al. (1988) quantified the correlation between increasing population and temperature among a network of 1219 stations in the U.S. from 1901 to 1981. The effect of urbanization on temperature has been detected even in small towns (population < 10,000). Balling and Idso (1989) also demonstrated urban warming is associated with population growth based on historical temperature data in U.S. from 1920 to 1984. The trend is expected to continue with further urbanization and population concentration, which calls for a predictive demographic model (Bettencourt et al., 2007), especially for cities, in order to devise an operative tool for future urban climate projections. The far-reaching goal, herein, is to develop a novel approach for predicting trends of urban warming by quantitatively linking the urban population–climate dynamics in metropolitan area, revealing the links between urban transportation network and environment impacts.

As mentioned above, the Phoenix Metropolitan Area (PMA) was selected first as the testbed. The development and validation of the statistical scaling model, which correlates urban population growth and urban warming for a selected metropolitan area, is presented in Section 2. We extended the proposed model to multiple study areas (U.S. metropolitan areas) to evaluate the connection between urban population growth and urban warming in Section 3. The results have been published in Wang and Wang (2017). Section 4 summaries historical data of multiple transportation networks, which can serve as a database for future model development.

5

2.0 DEVELOPMENT OF A SCALING MODEL TO EVALUATE URBAN WARMING WITH POPULATION GROWTH

2.1 METHODS

2.1.1 Urban Population Model

The population growth is an intrinsic measure of urbanization, which blends all the dynamic indicators associated with urban development such as infrastructure development, LULC changes, anthropogenic emissions, etc. Define N(t) as the population in a specific urban region at time t, the material resources at that time as Y(t), the co-evolution of the available material resources and the sustained total population can be linked using a generic exponential function (Bettencourt et al., 2007):

0( ) ( )Y t Y N t β= , (1)

where Y0 is the initial resource, and β is a parameter accounting for generic dynamic contributors in a community (a generic one, be it a human society or a biological group) development contributing to the population growth/decay. For example, a β value smaller than unity signifies the constraint of infrastructure and transportation materials/quantities in the built environment (Bettencourt et al., 2007). While the urban infrastructure, e.g. gasoline stations and length of electrical cables, is necessary in terms of urban development, its limitations of development or availability inevitably impose restrictions on continuous population increase. This is analogous to a biological community; e.g. for bees, when the food supply is ample, the available space in a beehive will ultimately set an upper limit of the bee population.

The consumption of the total material resources can be subdivided into two categories, i.e. for population maintenance and growth, respectively, which yields:

( ) ( )mdNY t R N t Edt

= + , (2)

where Rm is the average amount of resource for individual maintenance per unit time per capita, with the subscript m denotes “model”, and E is the average amount of resource required for population growth per capita.

When β = 1, the driver of population growth is to sustain individual need (e.g., housing, employment, household electrical and water consumptions), leading to an exponential growth pattern. The exponential growth is well-known as a paradigmatic “free” growth mode for a generic biological community (e.g., grey seals (Bowen et al., 2003)) or human societies (e.g., Kampala, Uganda (Vermeiren et al., 2012)), subject to no pressing external stress. Integrating Eq. (2) leads to:

6

0( ) /0( ) mY R t EN t N e −= . (3)

Note that for the exponent here, (Y0 − Rm)/E is time-independent. When β > 1, population growth is driven by wealth, information and resources creation (e.g., inventors, wages, GDP, etc.). Note that in this case, there will be an initial capacity (Rm/Y0)1/(β−1). If the initial population N0 is greater than this capacity there exists a threshold time tc for population growth, after which population will collapse:

( ) ( )

11

0

1ln 1 (0)1 1 (0)

mc

m m

RE Et NR Y R N

βββ β

−−

= − − ≈ − −

. (4)

When β < 1, the growth curve should be sigmoidal with a carrying capacity (Y0/Rm)1/(1−β), and the growth is related to optimal and efficient urban development (e.g. infrastructure and transportation materials or quantities).

For simplicity, Rm and E are assumed to be constants and independent of N, following the treatment in previous studies (Lane et al., 2009; West et al., 2001). Population growth during a few decades can be divided into several successive cycles or time periods (stages) (Bettencourt et al., 2007). In a single stage, the initial capacity serves as a threshold for population growth with scaling β > 1, while the carrying capacity serves as the constraint for β < 1. Hence, we prescribe Y0 as a constant with a fixed β value every year in a single stage. The ratio E/Rm is interpreted as the necessary average time for a human from birth to productive maturity, given by E/Rm ≈ τ × 20 years (Bettencourt et al., 2007), where τ = 1 year−1. Re-arrange Eq. (2), we have:

( )1

* * * 110 0

* * *( ) (0) exp 1m

m m

Y Y RN t N tR R E

ββ β

−−

= + − − −

. (5)

2.1.2 Population–Temperature Regression

Here we use a statistical regression model to correlate the historical climate data to demographic data:

0 1y xβ β ε= + + , (6)

where y is the response function, x is the regressor, β0 and β1 are the coefficients of linear regression, and ε is a random error term. In order to estimate the regression coefficients in Eq. (6), we adopted the least-squares estimation:

0 1ˆ ˆy xβ β= + , (7)

where the circumflex denotes estimated value (estimator), and the regression coefficients in Eq. (6) are replaced by the least-squares estimators here.

7

2.2 CASE STUDY IN THE PHOENIX METROPOLITAN AREA

2.2.1 The Study Area

In this study, we chose the Phoenix Metropolitan Area (PMA) as out testbed, partly informed by the strongest population-temperature correlation in Sun Corridor. As the largest metropolitan area in this region, the PMA is a desert metropolitan area of about 37000 km2 and has been expanding for more than 60 years since 1950 (Brazel et al., 2007) (Figure 2). During this period, extensive LULC changes have taken place, converting agricultural and natural landscapes to the built environment. The urban expansion of the PMA was one of the highest among 100 largest metropolitan areas from 1980 to 2010 (Frey, 2012). Rapid urbanization in this region results in not only a population growth, but also temperature increases and a severe UHI (Hedquist and Brazel, 2014). For example, Brazel et al. (2007) reported an increase of 2 to 4 oC in mean minimum temperature of June in this region due to urbanization between 1990 and 2004.

2.2.2 Correlating the Urban Climate Change with Urban Population

The demographical data of the Phoenix-Mesa-Glendale Metropolitan Statistical Area from U.S. BEA (2014) were used in this analysis. Population growth rates of the low, medium and high series (2015-2050) were retrieved from the Population Projections of the Arizona Department of Administration – Office of Employment and Population Statistics (ADOA-EPS) (2015). In addition, 7 NOAA land-based weather stations (NOAA NESDIS, 2016) were selected for the estimate of historical annual average temperature (AAT) data (1969 – 2014) (Figure 2b). Results for the correlation between the historical AAT and population data in the PMA using both the linear and exponential regression are shown in Figure 3.

The temperature-population relation is determined statistically as

0.0723 20.7344T N= + , (8)

using the linear regression, where T is the temperature in Celsius, and N is the population in 105; and

1.8201ln 16.8590T N= + , (9)

using the exponential regression. Significant coefficients of statistical correlation were found between the populating and urban warming in both the linear (R2 = 0.735) and exponential (R2 = 0.798) regression (at the confidence level of 95%).

8

Figure 2: Geographical location of the study area with national land cover (NLCD2011) (a) overlaid in the state of Arizona, and (b) PMA and weather stations.

Weather Stations

Los Angeles

San Diego

Las Vegas

Phoenix

Tucson

Albuquerque

El Paso

CALIFORNIA

NEVADA

ARIZONA

NEW MEXICO

TEXAS

UTAH COLORADO

Phoenix Metropolitan Area (PMA)

9

Figure 3: Correlation between the AAT and population in the PMA using the (a) linear and (b) exponential regression.

2.2.3 Simulating the Historical Population Growth

In this section, the future population growth in the PMA is projected using the prognostic population model (see Section 2.1 for details), where various growth modes represent different underlying driving forces for population increases, including technological, socio-economic, and political drivers. Relative population growth rate of the PMA over time revealed the existence of accelerating and decelerating stages (Figure 4), similar to those of New York City (Bettencourt et al., 2007). However, due to different urbanization levels in these two regions, the successive shorter stages of super-exponential growth in New York City are not observed in the PMA. The reaccelerated process during a different period was strongly modulated by new emerging urban areas in the PMA, as shown in Table 1. To test the applicability of the proposed population model in simulating different population growth modes, we divide years 1969–2014 into seven different stages, with the key parameters listed in Table 2.

Figure 5 illustrates the capacity of the population growth model in predicting population in different temporal periods as well as growth modes. Note that a superlinear (β >1) relation is found for four periods (Figure 5a–c&g) in the PMA, suggesting the information and resources creation, such as patents and financial services, as the main driving force during these periods. Taking innovation for example, metropolitan areas are often more appealing to inventors and industries (Batty, 2008; Feldman and Audretsch, 1999). The larger β value found in the PMA suggests rapid population growth rate associated with strong driving force in the area during the above superlinear growing periods.

Annual average temperature (oC)

Popu

latio

n(1

06 )

20 21 22 23 24 250

1

2

3

4

5R2 = 0.735

(a)

Annual average temperature (oC)

Popu

latio

n(1

06 )

20 21 22 23 24 250

1

2

3

4

5R2 = 0.798

(b)

10

Figure 4: Historic record of population growth rate in PMA from 1969 to 2014.

Table 1: Emerging cities during different population growth stages in the PMA.

Years Emerging cities and population growth facts 1969-1976 Available data since 1969 (U.S. BEA, 2014).

The recession during 1969 – 1970 resulted in a population spurt (Rex, 2000).

1977-1981 Glendale, Tempe, Mesa experienced their highest population growth rates (U.S. BEA, 2014).

1982-1990 East Valley (Mesa, Chandler, Tempe, and Gilbert) emerged around 1982 (Ripley, 2011) http://www.eastvalleytribune.com/guide/article_ac67b832-fa97-11e0-b1eb-001cc4c03286.html). Peoria went through its highest population growth rate (over 300%) (U.S. BEA, 2014).

1991-2003 Chandler experienced an exponential growth (early 1990s), meanwhile Surprise and Avondale went through their highest population growth rates (over 300% and over 100%, respectively) (U.S. BEA, 2014).

2004-2011 Goodyear, Buckeye experienced the highest population growth rate (over 200% and over 600%, respectively). Surprise and Avondale continued to expand (population increased over 300% and over 100%, respectively). Population in El Mirage increased over 300%, and the increase in Queen Creek was over 500% (U.S. BEA, 2014).

2012-2014 Population growth has the same trend as that of the state of Arizona (ADOA-EPS, 2015), and this may be related to economy recovery.

Year

Popu

latio

ngr

owth

rate

(%)

1970 1975 1980 1985 1990 1995 2000 2005 20101

2

3

4

5

6

7

11

Table 2: Calibrated parameters in the proposed model for different stages of population growth in PMA.

Period index Stage N0 (1×105) β Error < = Rm* E* R2 (a) 1969 – 1973 10.134 1.211 1.5%N(t)real 0.50 1.00 0.992 (b) 1974 – 1979 13.020 1.300 1.5%N(t)real 0.50 1.00 0.985 (c) 1980 – 1985 16.122 1.282 1.5%N(t)real 0.50 1.00 0.989 (d) 1986 – 1990 20.133 0.853 0.5%N(t)real 0.50 1.00 0.999 (e) 1991 – 2003 23.192 0.848 1.5%N(t)real 0.50 1.00 0.998 (f) 2004 – 2011 36.373 0.877 1.5%N(t)real 0.50 1.00 0.995 (g) 2012 – 2014 43.310 1.129 0.1%N(t)real 0.50 1.00 0.999

Figure 5: Comparison between predicted and observed historical population in the PMA; numbering of subplots follows the index in Table 2.

Year

Popu

latio

n(1

05 )

1973 1974 1975 1976 1977 1978 1979 198012

13

14

15

16(b)

Year

Popu

latio

n(1

05 )

1968 1969 1970 1971 1972 1973 197410

11

12

13ObservedPredicted

(a)

Year

Popu

latio

n(1

05 )

1979 1980 1981 1982 1983 1984 1985 198615

16

17

18

19

20(c)

Year

Popu

latio

n(1

05 )

1985 1986 1987 1988 1989 1990 199120

21

22

23(d)

Year

Popu

latio

n(1

05 )

2002 2004 2006 2008 2010 2012

37

39

41

43(f)

Year

Popu

latio

n(1

05 )

1990 1992 1994 1996 1998 2000 2002 200422

25

28

31

34

(e)

Year

Popu

latio

n(1

05 )

2011 2012 2013 2014 201543

44

45(g)

12

2.2.4 Projecting the Future Population Growth

Assuming the current trend of population growth in PMA continues, and based on the calibrated parameters of the last stage (2012 – 2014), we predict the future population in the PMA up to 2050, with results shown in Figure 6. Here we denote the two theoretical population models derived from Bettencourt et al. (2007) as: (a) the free growth model based on Eq. (5) with β = 1.0, and (b) the scaling model with β > 1 or β < 1 for super-exponential or sub-exponential growth respectively, each constrained by a different set of socioeconomic conditions. In addition to the theoretical predictions, the low, medium, and high series population growth rates projected empirically by the ADOA-EPS (2015) are also presented in Figure 6. A more rapid rate of population increase predicted by the scaling model (β = 1.129) is observed as compared to that of the free growth model (β = 1.0). Meanwhile, both theoretical models predict larger population sizes than the empirical ADOA-EPS population projections in 2050. At the end of 2050, using the medium series of ADOA-EPS projection as the reference, the deviation of the population estimated by the scaling model from the reference is 24.29%, while the discrepancy of the free growth model is 13.74%. This discrepancy apparently suggests that the uncertainty of the theoretical predictions of future population becomes larger as expected, where a local stationarity is assumed in the population growth pattern and extended for a long period of time (in this case 35 years from 2016 to 2050).

Figure 6: Projected future population in the PMA based on calibrated parameters in the most recent period of 2012 – 2014.

Both the free growth and the scaling models can predict the population size in the first 15 years (2016 - 2030) with reasonable accuracy when compared against projection of the medium series,

Year

Popu

latio

n(1

05 )

1970 1980 1990 2000 2010 2020 2030 2040 205010

20

30

40

50

60

70

80

90

100Historical dataFree growth modelScaling modelADOA-EPS high seriesADOA-EPS mediumADOA-EPS low

13

with an error less than 1%. The theoretical models underestimate population size from 2015 to 2025, but with slight overestimates from 2025 to 2030. A more significant deviation is observed between predicted and projected population from 2030 to 2050, indicating that key parameters for the theoretical population model should be updated periodically to represent dynamic population growth mode in a city, especially for regions with rapidly varying socioeconomic conditions (e.g. technology hubs).

2.2.5 Predicting Future Urban Warming

The population-temperature correlation provides a simple means to forecast future urban temperature if a feasible population growth model is given, where both the linear regression (hereafter referred to as LR) in Eq. (8) and the exponential regression (hereafter ER) in Eq. (9) are used in subsequent analysis. The predicted temperature trends in the PMA using the population projections estimated from the free growth and scaling models, together with that from the medium series projection by ADOA-EPS, are shown in Figure 7, based on LR and ER population-temperature correlations respectively. It is clear that predicted future temperatures using LR are higher than those using ER, due to the superlinear characteristic of ER (i.e., the temperature increase rate slows down when population size is large, c.f. Figure 3b). All projections of temperature increase are summarized in Table 3. The scaling model with LR predicts the strongest warming from 24.1 °C to 27.5 °C over the next 35 years, while the low series population growth with ER yields the weakest warming from 23.8 °C to 24.5 °C. In 2050, the predicted temperature increase using LR relative to the reference temperature in 2016 ranges from 1.44 °C to 3.40 °C, while the increase using ER varies from 0.65 °C to 1.27 °C for different population growths.

Year

Ann

uala

vera

gete

mpe

ratu

re(o C

)

2015 2020 2025 2030 2035 2040 2045 2050

24

25

26

27

28Free growth model LRFree growth model ERScaling model LRScaling model ERADOA-EPS medium projection LRADOA-EPS medium ER

14

Figure 7: Predicted trends of future temperature changes in PMA using different population projections.

Table 3: Summary of predicted temperature increase (in °C) from the reference in 2016.

Year Linear regression (LR) Exponential regression (ER) SM FG HS MS LS SM FG HS MS LS

2016 − − − − − − − − − − 2020 0.25 0.25 0.32 0.27 0.21 0.13 0.13 0.17 0.14 0.11 2025 0.60 0.59 0.73 0.59 0.46 0.30 0.29 0.35 0.30 0.23 2030 1.01 0.96 1.13 0.91 0.70 0.48 0.46 0.52 0.44 0.34 2035 1.47 1.36 1.54 1.23 0.92 0.66 0.62 0.68 0.57 0.44 2040 2.02 1.80 1.94 1.52 1.11 0.86 0.78 0.82 0.68 0.52 2045 2.65 2.29 2.33 1.80 1.29 1.06 0.94 0.95 0.78 0.59 2050 3.40 2.82 2.71 2.06 1.44 1.27 1.11 1.07 0.87 0.65

* SM: Scaling model; FG: Free growth model; HS: High series; MS: Medium series; LS: Low series

In addition, we found that with LR relation, the predicted temperature increase by the scaling model of population growth is close to that by the free growth model in the period 2016 – 2035, with a difference of only 0.1 °C. Differences of the same magnitude are also observed with ER projections from 2016 to 2050. Despite the difference in population growth modes (free versus super- or sub-exponential), the predicted temperature in 2050 using the theoretical population models is higher than that based on the empirical ADOA-EPS population projections, as the latter takes into more socioeconomic constraints. Though the future PMA temperature increase predicted by the ADOA-EPS is lower, considerable discrepancies still exist between different series (vary from 0.42 °C to 1.27 °C), indicating uncertain warming effects with different population growth modes. Note that the warming range results from assumed different total fertility rates of all race groups in 2100 (ADOA-EPS, 2015), viz., ADOA-EPS projects total fertility rates will converge to 1.4, 1.9, and 2.4 for low, medium, and high series in 2100, respectively.

15

3.0 STATISTICAL ANALYSIS IN 10 U.S. MEGAPOLITANS

The development and retrofit of urban areas have significant impact on local and regional climatic changes (Kalnay and Cai, 2003). In order to deduce the relation between the trends of climatic change and population growth in cities, we collected data from 10 major megapolitan areas in the U.S. (Nelson and Lang, 2011) to extend the application of the model. The demographic data from 1969 to 2014 were retrieved from the U.S. Department of Commerce, Bureau of Economic Analysis (BEA) (Brown and Wardwell, 1980; U.S. BEA, 2014). General geographic information, in particular, main cities, states, and the number of counties located within or overlapping the geographical boundaries of these megapolitan areas are summarized in Table 4.

Table 4: General geographic information of the 10 selected major megapolitan areas in the U.S.

Megapolitan areas

Main cities Number of counties

State and counties inside megapolitan areas

Sun Corridor Phoenix, Tucson 4 AZ Cascadia Seattle, Portland, Vancouver 24 OR, WA and Canada Florida Tampa, Miami, Orlando, Jacksonville 25 FL Front Range Albuquerque, Santa Fe, Colorado Springs,

Denver 15 CO, NM

Great Lakes Chicago, Detroit, Pittsburgh, Columbus, Cleveland, Minneapolis, St. Louis, Indianapolis

169 IL, IN, MI, MN, OH, PA, WI, Canada

Northeast New York, Philadelphia, Washington D.C., Boston, Baltimore

165 CT, DC, DE, MA, MD, NY, NJ, PA, VA, WV, NH, RI

Northern California (CA)

Oakland, Reno, Sacramento, San Jose, San Francisco

27 CA, NV

Piedmont Atlantic Atlanta, Birmingham, Raleigh-Durham, Charlotte

121 NC, SC, GA, AL

Southern California (CA)

Los Angeles, San Diego, Anaheim, Long Beach, Las Vegas

12 CA, NV and Mexico

Texas Triangle Austin, Dallas/Fort Worth, Houston, San Antonio

67 TX

Population in each megapolitan area is computed based on the demographical data collected at the county level, i.e., the total population of a megapolitan area is the sum of population from individual counties. For example, population in Great Lakes is the sum of population from 169 counties in this area (see Table 4). During the past 46 years (1969 – 2014), the population in all the 10 megapolitans has experienced continuous increases (Figure 8), except for the Great Lakes where fluctuations of population were observed around 1979 – 1986. Though not the largest megapolitan area in the U.S., the Northeast region is the most densely populated megapolitan

16

with the largest population size as well as significant growth rate. Among these regions, Arizona’s Sun Corridor has seen the most rapid rate of urbanization with a population growth of more than 309% during the past five decades. The rate of population growth is lowest in Great Lakes and Northeast, which is partly due to the high level of urbanization preexisted in these areas prior to 1969.

Figure 8: Population growth in the selected megapolitan areas in (a) Sun Corridor, Cascadia, Northern California, Southern California, and Front Range, and (b) Texas Triangle, Florida, Great Lakes, Piedmont Atlantic, and Northeast during 1969 – 2014.

The temperature data were downloaded from the National Environmental Satellite, Data, and Information Service (NESDIS) of the National Oceanic and Atmospheric Administration (NOAA) (NOAA NESDIS, 2016). We averaged the archived near-surface (5 ± 1 ft) air temperature of all climate divisions in each megapolitan area to obtain the AAT of an individual megapolitan. All the 10 selected regions exhibit a trend of urban warming with some fluctuations (Figure 9) under the background of global climate changes accelerated in the past half century (Murray & Colle, 2011). Florida, Sun Corridor and Texas Triangle are the three warmest megapolitan areas in the U.S. during the past 46 years, while Front Range and Great Lakes are the two coldest according to the AAT data. In addition, the synchronization of the AAT change is observed, for example, 1986 – 1992 and 1996 – 2001.

Year

Popu

latio

n(1

07 )

1968 1978 1988 1998 20080

0.5

1

1.5

2

2.5

3Arizona Sun CorridorCascadiaNorthern CaliforniaSouthern CaliforniaFront Range

(a)

Year

Popu

latio

n(1

07 )

1968 1978 1988 1998 20080

1

2

3

4

5

6

7

8Texas TriangleFloridaGreat LakesPiedmont AtlanticNortheast

(b)

17

Figure 9: Same as Figure 8 but for AAT.

To quantify the correlation between the population growth and urban warming, we conducted linear regressional analysis with a 95% confidence level for the selected megapolitans. The coefficient of determination R2 is used to quantify the correlation level. It is clear from Figure 10 that the AAT is significantly correlated with population in most of these megapolitan areas. The statistical correlation is found to be most significant in Arizona’s Sun Corridor, with a R2 value of 0.563. The variance is expected as that the AAT data is obtained from the divisional level with areal averaging and embedded background changes without discriminating to detailed land use land cover classification, whereas the population growth is obtained at county level. Relation between AAT and population in Great Lakes during the past 46 years indicates large data scatter relative to Arizona’s Sun Corridor. Two megapolitan areas in California show similar high R2 values to Arizona’s Sun Corridor, especially Southern California (R2 = 0.468), revealing similar temperature-population relation in this geographical region (Arizona and California), while the correlation Great Lakes is slightly at variance with the general trend with a relatively low R2 value of 0.164.

Year

Ann

uala

vera

gete

mpe

ratu

re(o C

)

1968 1978 1988 1998 20085

10

15

20

25

30Arizona Sun CorridorCascadiaNorthern CaliforniaSouthern CaliforniaFront Range

(a)

YearA

nnua

lave

rage

tem

pera

ture

(o C)

1968 1978 1988 1998 20085

10

15

20

25

30Texas TriangleFloridaGreat LakesPiedmont AtlanticNortheast

(b)

18

Figure 10: Statistical correlation between the AAT and population in the 10 megapolitan areas in the period of 1969 – 2014; scatter points are the historical observations, and red

line is the linear regression.

Popu

latio

n(1

06 )

17 18 19 20 21

1

3

5

7 Arizona Sun CorridorR2 = 0.563

Popu

latio

n(1

06 )

7 8 9 10 111

3

5Front RangeR2 = 0.329

Popu

latio

n(1

07 )

20 21 22 230

0.5

1

1.5

2 FloridaR2 = 0.240

Popu

latio

n(1

07 )

17 18 19 20 210

0.5

1

1.5

2

2.5 Texas TriangleR2 = 0.468

Popu

latio

n(1

07 )

15 16 17 18 191

1.5

2

2.5

3 S CaliforniaR2 = 0.468

Popu

latio

n(1

06 )

9 10 11 12

3

5

7

9CascadiaR2 = 0.218

Popu

latio

n(1

07 )

7 8 9 10 11 123

3.5

4

4.5 Great LakeR2 = 0.164

Popu

latio

n(1

07 )

13 14 15 16 170.5

1

1.5

2N CaliforniaR2 = 0.360

AAT (oC)

Popu

latio

n(1

07 )

14 15 16 170

0.5

1

1.5

2

2.5Piedmont AtlanticR2 = 0.244

AAT (oC)

Popu

latio

n(1

07 )

10 11 12 13 143

4

5

6

7NortheastR2 = 0.363

19

4.0 URBAN TRANSPORTATION NETWORKS AND BEYOND

4.1 ROAD NETWORK

The dataset of the primary roads (including Interstate Highway System, United States Numbered Highway System, etc.) was retrieved from the U.S. Census Bureau's Master Address File / Topologically Integrated Geographic Encoding and Referencing (MAF/TIGER) Database (MTDB, 2017). The dataset of population came from U.S. BEA (2014). Both datasets are presented in Figure 11, showing the road connections between major cities and states.

Figure 11: Primary roads, population density, and city population over the contiguous U.S.

Note that the dataset of primary roads contains information of road name, type, and length (calculated using ArcMap software), providing opportunities of future spatial analysis. With the given speed limits, the (minimum) travel time between two selected metropolitan areas can be easily calculated using Spatial Analyst tool in ArcMap. With the developed scaling model framework, solutions could be available toward sustainable urban road network planning. For example, California and Arizona are connected mainly through three highways, i.e. I-8, I-10, and I-40. Though the distance/travel time calculation, I-8 and I-10 are identified as the fastest highways connecting California and Arizona. As the urban population keeps growing over time, the number of commuters also increases, which potentially calls widening and more frequent

20

maintenance of highways. The different patterns of population and commuters increases in different metropolitan areas and along highways can be predicted using the proposed scaling model with tuned parameters. The future pressure on a given highway in terms of rebuild or maintenance will be identified, providing suggestions to transportation planners.

The same framework is also applicable for finer scale analysis. An example for Maricopa County, Arizona, where Phoenix is located, is shown in Figure 12. There are 274832 road sections in this county. The dense spatial distribution of the road network is similar to that of developed imperviousness over urban areas, though there are also sparse networks in peripheral suburban and rural areas. Therefore more attention should be paid to regions with fast-growing population.

Figure 12: (a) Urban percent developed imperviousness (National Land Cover Database (NLCD) 2011) and (b) road network (MTDB, 2017) in Maricopa County, Arizona.

Besides, the growth of road network could be projected by the proposed scaling model as well, as it is implicitly driven by the population growth. Table 5 shows the summary of the length of public roads for all states in 2013 as an example. This dataset was retrieved from the Office of highway Policy Information, U.S. Department of Transportation (2015). The statistics are updated yearly since 1942, providing a sufficient time series for model simulation and future predictions.

Table 5: Public road length (miles) by functional system in the United States in 2013.

21

State Interstate Other principal and minor arterials*

Major and minor collectors

Local Total

Alabama 1002 9716 22386 68733 101837 Alaska 1081 1571 3300 9727 15680 Arizona 1168 6021 8072 51178 66441 Arkansas 656 7441 21061 72499 101656 California 2451 30002 32223 110313 174989 Colorado 952 9259 16245 62109 88565 Connecticut 346 3004 3206 14918 21474 Delaware 41 680 1039 4633 6393 District of Columbia 12 286 157 1047 1501 Florida 1495 13590 14560 92442 122088 Georgia 1247 14329 23037 90006 128620 Hawaii 55 824 752 2800 4430 Idaho 612 4249 10611 32611 48082 Illinois 2185 14771 22169 106583 145708 Indiana 1188 8758 22523 65084 97553 Iowa 782 9778 31629 72240 114429 Kansas 874 9688 33698 96427 140687 Kentucky 801 6169 16562 56066 79598 Louisiana 926 5685 9972 44844 61427 Maine 367 2199 5914 14401 22882 Maryland 481 4110 5059 22772 32422 Massachusetts 575 6768 4550 24478 36370 Michigan 1244 15008 24458 81431 122141 Minnesota 914 13686 30408 93759 138767 Mississippi 700 7740 15892 50784 75116 Missouri 1379 10487 25109 94925 131900 Montana 1192 6088 16245 51408 74933 Nebraska 482 8144 20772 64371 93770 Nevada 596 3471 5612 30460 40139 New Hampshire 225 1745 2642 11485 16098 New Jersey 431 6391 4437 28034 39293 New Mexico 1000 4963 9188 55620 70772 New York 1724 14601 20737 77666 114728 North Carolina 1255 10018 17351 77579 106202 North Dakota 571 5941 11929 68637 87078 Ohio 1574 11253 22869 87602 123297 Oklahoma 933 8417 25490 78100 112940 Oregon 730 7112 18589 44798 71228

Table 5. cont.

22

State Interstate Other principal and minor arterials*

Major and minor collectors

Local Total

Pennsylvania 1857 13762 19847 84470 119936 Rhode Island 70 914 887 4235 6106 South Carolina 851 7233 15089 43059 66232 South Dakota 679 6430 19004 56446 82558 Tennessee 1104 9305 17994 67132 95536 Texas 3415 33280 65154 211378 313228 Utah 937 3772 8162 33384 46254 Vermont 320 1320 3119 9506 14266 Virginia 1119 8764 14394 50472 74748 Washington 764 8412 17292 55980 82448 West Virginia 555 3498 8635 26063 38750 Wisconsin 743 12910 23501 77990 115145 Wyoming 914 3671 10279 14161 29024 United States, total 47575 417232 803807 2846848 4115462

* Includes other freeways and expressways.

4.2 OTHER TRANSPORTATION NETWORKS

4.2.1 Air Traffic

Similarly, the enplanements at airports have gone through increases during the last few decades. The visualized images of enplanements at top 50 U.S. Airports in the last 2 decades were downloaded from U.S. Department of Transportation, Bureau of Transportation Statistics (2016) and are presented in Figure 13. Increasing urban population has led to heavier air traffic over several airports of major cities, e.g. Phoenix Sky Harbor International Airport, Denver International Airport, McCarran International Airport (Las Vegas, NV), etc. An example of the statistics (and the abbreviations in Figure 13) is shown in Table 6. The proposed model can be used to predict the trend of increasing air traffic directly, or indirectly via the correlation with population growth.

23

Figure 13: Enplanements at the top 50 U.S. Airports from 1991 to 2016.

Table 6: Summary of Enplanements at the top 50 U.S. Airports in 2016.

Abbreviation Airport name Enplanements TPA Tampa International 4522421 STL Lambert-St. Louis International 9304471 SNA John Wayne Airport-Orange County 2544596 SMF Sacramento International 2104678 SLC Salt Lake City International 5323338 SJU Luis Munoz Marin International 3796906 SJC Norman Y. Mineta San Jose International 3188132 SFO San Francisco International 15010375

1991 1996

2001 2006

2011 2016

24

Table 6: cont.

Abbreviation Airport name Enplanements SEA Seattle/Tacoma International 7829458 SAT San Antonio International 2567883 SAN San Diego International 5408393 RDU Raleigh-Durham International 4309939 PIT Pittsburgh International 7713803 PHX Phoenix Sky Harbor International 10998363 PHL Philadelphia International 6523271 PDX Portland International 3156559 ORD Chicago O'Hare International 26790767 OAK Metropolitan Oakland International 2981799 MSY Louis Armstrong New Orleans International 3204472 MSP Minneapolis-St Paul International 8908643 MKE General Mitchell International 1744734 MIA Miami International 12051889 MEM Memphis International 3501115 MDW Chicago Midway International 2935166 MCO Orlando International 8293359 MCI Kansas City International 3292452 LGA LaGuardia 9474937 LAX Los Angeles International 21530661 LAS McCarran International 8296350 JFK John F. Kennedy International 11992926 IND Indianapolis International 2549344 IAH George Bush Intercontinental/Houston 8121412 IAD Washington Dulles International 5004363 HOU William P Hobby 3761490 HNL Honolulu International 9833705 FLL Fort Lauderdale-Hollywood International 3737724 EWR Newark Liberty International 10445578 DTW Detroit Metro Wayne County 9592572 DFW Dallas/Fort Worth International 22842312 DEN Denver International 12358848 DCA Ronald Reagan Washington National 6627775 DAL Dallas Love Field 2791490 CVG Cincinnati/Northern Kentucky International 4330626 CLT Charlotte Douglas International 7687233 CLE Cleveland-Hopkins International 3526049 BWI Baltimore/Washington International Thurgood Marshall 4376847 BOS Logan International 9417569

25

Table 6: cont.

Abbreviation Airport name Enplanements BNA Nashville International 3900235 AUS Robert Mueller Municipal 2019767 ATL Hartsfield-Jackson Atlanta International 17979191

Figure 14 shows the numbers of passengers from 2002 to 2016 (U.S. Department of Transportation, Bureau of Transportation Statistics, 2016). It is noteworthy that the trend of changes does not necessarily follow the increase of population (Figure 14). However, predicting the changes of air traffic for individual airports might be possible, especially for the airports located in emerging cities with intensive urbanization.

Figure 14: Numbers of passengers at U.S. Airports from 1991 to 2016.

4.2.2 Railroad System

The dataset of existing railroad systems was retrieved from MTDB (2017), and is presented in Figure 15. In addition, the miles of freight railroad by state (Association of American Railroads, 2015) are summarized in Table 7. Note that the 656 miles of track owned by Amtrak are excluded.

Year

Num

bero

fpas

seng

ers(×

108 )

2002 2004 2006 2008 2010 2012 2014 20160

1

2

3

4

5

6

7

8

9

10DomesticInternationalTotal

26

Figure 15: Railroad system over contiguous U.S. in 2017.

Table 7: Miles of Freight Railroad Operated by Class of Railroad in 2012.

State Class I Regional Local Canadian Total Linehaul Switching and

terminal Alabama 2255 236 635 68 0 3194 Alaska 0 506 0 0 0 506 Arizona 1235 0 259 149 0 1643 Arkansas 1677 0 895 126 0 2698 California 3919 0 999 377 0 5295 Colorado 2018 198 368 78 0 2662 Connecticut 6 210 148 0 0 364 Delaware 183 0 47 20 0 250 District of Columbia

15 0 0 5 0 20

Florida 1693 431 774 2 0 2900 Georgia 3251 0 1384 18 0 4653 Hawaii 0 0 0 0 0 0 Idaho 962 33 481 147 0 1623 Illinois 5851 148 649 338 0 6986 Indiana 2510 304 1076 185 0 4075

27

Table 7: cont.

State Class I Regional Local Canadian Total Linehaul Switching and

terminal Iowa 3189 364 271 45 0 3869 Kansas 2816 1429 367 243 0 4855 Kentucky 2117 270 221 0 0 2608 Louisiana 2354 0 515 58 0 2927 Maine 0 621 493 2 0 1116 Maryland 557 0 172 29 0 758 Massachusetts 261 529 159 24 0 973 Michigan 1557 0 1751 233 1 3542 Minnesota 3625 3 651 127 44 4450 Mississippi 1614 8 716 114 0 2452 Missouri 3399 0 419 139 0 3957 Montana 2061 865 274 0 0 3200 Nebraska 2567 324 469 15 0 3375 Nevada 1192 0 0 0 0 1192 New Hampshire

0 174 170 0 0 344

New Jersey 189 91 176 525 0 981 New Mexico 1431 0 96 310 0 1837 New York 1758 328 1231 128 2 3447 North Carolina

2335 0 709 214 0 3258

North Dakota 2182 766 382 0 0 3330 Ohio 3240 433 1265 350 0 5288 Oklahoma 2009 0 968 296 0 3273 Oregon 1103 321 843 129 0 2396 Pennsylvania 2428 772 1374 577 0 5151 Rhode Island 0 19 0 0 0 19 South Carolina

1948 0 266 97 0 2311

South Dakota 1494 74 98 87 0 1753 Tennessee 1836 0 751 62 0 2649 Texas 8369 0 1236 864 0 10469 Utah 1249 0 59 35 0 1343 Vermont 0 224 366 0 0 590 Virginia 2773 0 438 4 0 3215 Washington 1735 0 1272 185 0 3192 West Virginia 1855 0 365 6 0 2226 Wisconsin 2595 674 180 0 0 3449 Wyoming 1851 0 0 9 0 1860 U.S. total 95264 10355 26438 6420 47 138524

28

The publicly available datasets of annual railroad statistics are relatively rare without the membership in the Association of American Railroads. However, the proposed model can still be applied to the railroad system in a similar way as described in Section 4.1.

29

5.0 CONCLUSION

In this project, we developed a generic mechanistic scaling law to correlate the transportation networks and their impacts on the built environment, with urban population as the intrinsic driver (and the control parameter) for socio-economic growth. In addition, the warming trends in cities was used as the primary environmental indicator as the impact of socio-economic growth and the concomitant urban dynamics such as urban growth, infrastructural development, transportation network expansion, etc. The scaling law was developed and evaluated using the Phoenix metropolitan as the testbed, and can be readily extended to all major metropolitans within and outside of the United States.

The statistical analysis of historical urban population and annual air temperature (1969–2014) in Phoenix Metropolitan Area shows a strong correlation. Further regression analyses of ten major U.S. megapolitan areas show similar correlations between population growth and urban warming. The model was validated with historical demographic data in Phoenix over the past 46 years, and then was applied to predict future population growth. Predicted future population growth matches well with the estimates from the Arizona Department of Administration – Office of Employment and Population Statistics in the first few years. To extend the application of the proposed model, datasets of road network, air traffic and railroad system were retrieved from multiple data sources. These datasets will serve as the basis of future studies on spatial and temporal distribution and changes.

It is noteworthy that high uncertainty in longterm projections was found in this project, necessitates periodical updates of key model parameters signaling the expansion of inter- and intra-city transportation networks as the dynamic urban growth. This is manifesting due to that the urban transportation development and/or redevelopment in individual metropolitan areas over a long period inevitably involve different modes and socio-economic drivers. This is particularly crucial when the teleconnection among multiple metropolitans is taken into account for urban sustainability measurement. In addition, with richer sets of socio-economic and transportation data products and refined spatio-temporal analysis, the accuracy of the modeling framework is expected to be enhanced. Overall, the proposed framework provides a new perspective for the development of urban transportation networks, and is expected to be informative to urban planning and decision-making processes, including the design of transportation infrastructure.

31

6.0 REFERENCES

ADOA-EPS. (2015) Arizona Department of Administration - Office of Employment and Population Statistics (ADOA-EPS) Maricopa County 2015-2050 Projections. Retrieved July 8, 2016, from https://population.az.gov/population-projections Arnfield, A. J. (2003) Two decades of urban climate research: A review of turbulence, exchanges of energy and water, and the urban heat island. International Journal of Climatology 23(1), pp. 1-26. Association of American Railroads. (2015) Railroad Ten-Year Trends. Available at www.aar.org/StatisticsAndPublication as of July 2015. Badland, H. and Schofield, G. (2005) Transport, urban design, and physical activity: an evidence-based update. Transportation Research Part D-Transport and Environment 10(3), pp. 177-196. Balling, R. and Idso, S. (1989) Historical temperature trends in the United States and the effect of urban population growth. Journal of Geophysical Research: Atmospheres 94(D3), pp. 3359–3363. Batty, M. (2008) The size, scale, and shape of cities. Science 319(5864), pp. 769-771. Bettencourt, L.M., Lobo, J., Helbing, D., Kühnert, C. and West, G.B. (2007) Growth, innovation, scaling, and the pace of life in cities. Proceedings of the National Academy of Sciences of the United States of America 104(17), pp. 7301-7306. Bettencourt, L. and West, G. (2010) A unified theory of urban living. Nature 467(7318), pp. 912-913. Bowen, W. D., McMillan, J. and Mohn, R. (2003) Sustained exponential population growth of grey seals at Sable Island, Nova Scotia. ICES Journal of Marine Science: Journal Du Conseil 60(6), pp. 1265–1274. Brazel, A., Gober, P., Lee, S., Grossman-Clarke, S., Zehnder, J., Hedquist, B. and Comparri, E. (2007) Determinants of changes in the regional urban heat island in metropolitan Phoenix (Arizona, USA) between 1990 and 2004. Climate Research 33(2), pp. 171–182. Cash, D.W., Clark, W.C., Alcock, F., Dickson, N.M., Eckley, N., Guston, D.H., Jäger, J. and Mitchell, R.B. (2003) Knowledge systems for sustainable development. Proceedings of the National Academy of Sciences of the United States of America 100(14), pp. 8086-8091. Feldman, M. P. and Audretsch, D. B. (1999) Innovation in cities: Science-based diversity, specialization and localized competition. European Economic Review 43(2), pp. 409–429. Frey, W. H. (2012) Population growth in metro America since 1980: putting the volatile 2000s in perspective. Washington DC, USA: The Brookings Institution. Hedquist, B. C. and Brazel, A. J. (2014) Seasonal variability of temperatures and outdoor human comfort in Phoenix, Arizona, USA. Building and Environment 72, pp. 377–388. Heiple, S. and Sailor, D. J. (2008) Using building energy simulation and geospatial modeling techniques to determine high resolution building sector energy consumption profiles. Energy and Buildings 40(8), pp. 1426–1436. Henderson, V. (2002) Urbanization in developing countries. The World Bank Research Observer 17(1), pp. 89–112.

32

Hou, H., Liu, Y., Liu, Y., Wei, X., He, Q. and He, Q. (2015) Using inter-town network analysis in city system planning: A case study of Hubei Province in China. Habitat International 49, pp. 454-465. Jiang, B. and Jia, T. (2011) Zipf's law for all the natural cities in the United States: a geospatial perspective. International Journal of Geographical Information Science 25(8), pp. 1269-1281. Kalnay, E. and Cai, M. (2003) Impact of urbanization and land-use change on climate. Nature 423(6939), pp. 528-531. Kane, K., Tuccillo, J., York, A.M., Gentile, L. and Ouyang, Y. (2014) A spatio-temporal view of historical growth in Phoenix, Arizona, USA. Landscape and Urban Planning 121, pp. 70-80. Karl, T. R., Diaz, H. F. and Kukla, G. (1988) Urbanization: Its detection and effect in the United States climate record. Journal of Climate 1(11), pp. 1099–1123. Lane, D., Pumain, D., van der Leeuw, S. E. and West, G. (2009) Complexity Perspectives in Innovation and Social Change. New York, USA: Springer. Mage, D., Ozolins, G., Peterson, P., Webster, A., Orthofer, R., Vandeweerd, V. and Gwynne, M. (1996) Urban air pollution in megacities of the world. Atmospheric Environment 30(5), pp. 681–686. MTDB (2017) U.S. Census Bureau's Master Address File / Topologically Integrated Geographic Encoding and Referencing (MAF/TIGER) Database (MTDB). Retrieved June 20, 2017, from https://www.census.gov/geo/maps-data/data/tiger.html Murray, J. C. and Colle, B. A. (2011) The spatial and temporal variability of convective storms over the northeast United States during the warm season. Monthly Weather Review 139(3), pp. 992–1012. Nazaroff, W.W. (2013) Exploring the consequences of climate change for indoor air quality. Environmental Research Letters 8(1), 015022. Nelson, A. C. and Lang, R. E. (2011) Megapolitan America: A New Vision for Understanding America’s Metropolitan Geography. Chicago, USA: American Planning Association. NOAA NESDIS. (2016) National Climatic Data Center (NCDC) Climate Data Online (CDO). Retrieved June 27, 2016, from http://www7.ncdc.noaa.gov/CDO/georegion Oke, T. R. (1973) City size and the urban heat island. Atmospheric Environment 7(8), pp. 769–779. Patz, J.A., Campbell-Lendrum, D., Holloway, T. and Foley, J.A. (2005) Impact of regional climate change on human health. Nature 438(7066), pp. 310-317. Rex, T. R. (2000) Development of metropolitan Phoenix: Historical, current and future trends. Tempe, USA: Morrison Institute for Public Policy, Arizona State University. Ripley, J. (2011) The East Valley: 30 years and counting. Retrieved July 14, 2016, from http://www.eastvalleytribune.com/guide/article_ac67b832-fa97-11e0-b1eb-001cc4c03286.html Rosa, E. A., York, R. and Dietz, T. (2004) Tracking the anthropogenic drivers of ecological impacts. AMBIO: A Journal of the Human Environment 33(8), pp. 509–512. Sailor, D. J. (2011) A review of methods for estimating anthropogenic heat and moisture emissions in the urban environment. International Journal of Climatology 31(2), pp. 189–199. Seto, K.C., Reenberg, A., Boone, C.G., Fragkias, M., Haase, D., Langanke, T., Marcotullio, P., Munroe, D.K., Olah, B. and Simon, D. (2012) Urban land teleconnections and sustainability. Proceedings of the National Academy of Sciences of the United States of America 109(20), pp. 7687-7692.

33

United Nations (UN) (2012) World Urbanization Prospects: The 2011 Revision, ESA/P/WP/224, pp 33, The United Nations' Department of Economic and Social Affairs - Population Division, New York. U.S. BEA. (2014) U.S. Bureau of Economic Analysis (BEA). Retrieved June 23, 2016, from http://www.bea.gov/iTable/iTable.cfm?reqid=70&step=1&isuri=1&acrdn=2#reqid=70&step=1&isuri=1&7003=900&7022=11&7023=7&7001=711&7002=2&7090=70 U.S. Department of Transportation, Bureau of Transportation Statistics. (2016) Retrieved December 01, 2016, from https://www.bts.gov/topics/airlines-and-airports U.S. Department of Transportation, Office of Highway Policy Information. (2015) Retrieved September 20, 2016, from https://www.fhwa.dot.gov/policyinformation/statistics.cfm Vermeiren, K., Van Rompaey, A., Loopmans, M., Serwajja, E. and Mukwaya, P. (2012) Urban growth of Kampala, Uganda: Pattern analysis and scenario development. Landscape and Urban Planning 106(2), pp. 199–206. Wang, C. and Wang, Z.H. (2017) Projecting population growth as a dynamic measure of regional urban warming. Sustainable Cities and Society 32, pp.357-365. West, G. B., Brown, J. H. and Enquist, B. J. (2001) A general model for ontogenetic growth. Nature 413(6856), pp. 628–631.