This article was downloaded by: [Arizona State University] On: 03 July 2014, At: 12:57 Publisher: Routledge Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
The Professional Geographer Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/rtpg20
Do Sustainable Urban Designs Generate More Ecosystem Services? A Case Study of Civano in Tucson, Arizona a
b
V. Kelly Turner & Christopher S. Galletti a
Kent State University
b
Arizona State University Published online: 01 Jul 2014.
To cite this article: V. Kelly Turner & Christopher S. Galletti (2014): Do Sustainable Urban Designs Generate More Ecosystem Services? A Case Study of Civano in Tucson, Arizona, The Professional Geographer, DOI: 10.1080/00330124.2014.922021 To link to this article: http://dx.doi.org/10.1080/00330124.2014.922021
PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/termsand-conditions
Do Sustainable Urban Designs Generate More Ecosystem Services? A Case Study of Civano in Tucson, Arizona V. Kelly Turner Kent State University
Downloaded by [Arizona State University] at 12:57 03 July 2014
Christopher S. Galletti Arizona State University Empirical evidence of environmental performance of urban areas designed according to the principles of sustainable urbanism is limited. Using the case study of Civano, a planned development that was designed and marketed as a sustainable community in Tucson, Arizona, we quantify fine-scale differences in urban form and delivery of ecosystem services. We found that the urban design of the first phase of development translated to the lowest surface temperatures and highest albedo and vegetative density. The first and second phases of the development greatly reduced potable water consumption through the addition of nonpotable resources; however, the second phase had higher temperatures and less dense vegetation than even the conventional development. Our results show modest improvements in environmental performance through sustainable urbanism and suggest further refinement in fine-scale spatial analysis of the role of urban design in the provisioning of services. Key Words: arid environments, remote sensing, sustainability, urban planning. , (Civano) , ,
, , , , ; ,
, , , : , , , Es escasa la evidencia emp´ırica relacionada con el desempeno de acuerdo con los ˜ ambiental de las a´ reas urbanas disenadas ˜ principios del urbanismo sustentable. A partir del estudio de caso de Civano, un desarrollo planificado que se disen˜ o´ y se vendio´ a t´ıtulo de comunidad sustentable en Tucson, Arizona, nosotros cuantificamos las diferencias a fina escala en forma urbana y prestacion ´ de servicios ecosist´emicos. Encontramos que el diseno ˜ urbano de la primera fase del desarrollo se adecuo´ a las temperaturas superficiales m´as bajas y el m´as alto albedo y densidad vegetativa. La primera y segunda fases del desarrollo redujeron apreciablemente el consumo de agua potable mediante la adicion ´ de recursos no potables; sin embargo, la segunda fase tuvo temperaturas m´as altas y menos vegetacion ´ densa que el desarrollo convencional. Nuestros resultados muestran mejoras modestas en desempeno ˜ ambiental mediante el urbanismo sustentable y sugieren un mayor refinamiento en el an´alisis espacial a escala fina del papel del diseno ˜ urbano en el suministro de servicios. Palabras clave: entornos a´ ridos, percepci´on remota, sustentabilidad, planificaci´on urbana.
T
he sustainable urbanism movement within urban planning suggests a series of urban design alternatives to mitigate environmental challenges of sprawl. Yet, few studies have measured the environmental outcomes of these design alternatives once implemented (Conway 2009). This study draws on the theoretical and methodological contributions of geographers to land planning and land change science to begin to address this empirical gap. The approach is applied to a case study that is uniquely suited to compare the relative effects of strong, weak, and no emphasis on sustainable urban design. We deploy the ecosystem services framework and explore the use of remote sensing and geographic information systems (GIS) analysis to quantify urban designs and associated environmental outcomes at a fine
scale. The ecosystem services framework highlights the contribution of ecosystems to human well-being and aids the planning processes by revealing the tradeoffs associated with alternative land development and management plans and with human-made substitutes that might not deliver the full range of services provided by nature (Millennium Ecosystem Assessment 2003; Foley et al. 2005). The majority of ecosystem service research has been conducted at course scales; however, fine-scale analyses might better inform urban land management occurring at local scales (Alberti 2005). The land change science tradition in geography potentially contributes to this line of inquiry by exploring the applicability of tools, such as remote sensing analysis, to fine-scale urban contexts (Wentz et al. 2011). Increasing the application of remote sensing to
C Copyright 2014 by Association of American Geographers. The Professional Geographer, 0(0) 2014, pages 1–14 Initial submission, July 2013; revised submissions, November and December 2013; final acceptance, December 2013. Published by Taylor & Francis Group, LLC.
Downloaded by [Arizona State University] at 12:57 03 July 2014
2 Volume 0, Number 0, XXXX 2014
Figure 1 Map of Civano I (northernmost community), Civano II (in the middle), and the comparison community (southernmost community). (Color figure available online.)
urban environments is an emerging research priority in geography and ecology (Jenerette et al. 2011; Yang 2011). Civano is a two-phase planned development in Tucson, Arizona, that was intentionally marketed as a sustainable community. The first phase emphasized urban design and green building to achieve this goal, whereas the second phase was solely focused on the latter. It lies adjacent to a conventional suburban development. The proximity and similar size, number of households, and age of the three developments (Figure 1, Table 1) make it uniquely suited to address the following questions: (1) Do qualitative differences in design approach generate quantifiable differences in urban form? and (2) Do those differences lead to measurable changes in ecosystem service delivery? Underlying both questions is the broader challenge of utilizing the traditional tools of land change science, particularly remote sensing, in understanding fine-scale differences
in urban form and environmental function between neighborhoods.
Data and Methods To quantitatively demonstrate differences in urban design across the three communities, we quantified differences in urban design by calculating landscape metrics—percentage composition and patch density
Table 1 Size, households, age, and average building area of study area communities Community
Size (km2)
Households
Age
Building area (m2)
Civano I Civano II Comparison
0.75 1.00 0.75
599 693 613
1999 2007 2001
245.5 309.7 325.9
Downloaded by [Arizona State University] at 12:57 03 July 2014
Do Sustainable Urban Designs Generate More Ecosystem Services? 3 (PD; patches per hectare)—for standard urban land classes: impervious surface, trees and shrubs, low and medium albedo buildings (LMB), high albedo buildings (HAB), exposed soils, grass, pools, and other water bodies (e.g., Myint et al. 2013).We also calculated average building area using parcel data from the City of Tucson Water to determine the total building area per lot. The buildings were further classified as LMB or HAB. We conducted a Kruskal–Wallis analysis of variance (ANOVA) to test whether the means of block-level design factors were significantly different between Civano I, Civano II, and the comparison community. The Kruskal–Wallis is a nonparametric test similar to the parametric ANOVA, but it differs in that the data do not have to be normally distributed. We conducted the test on the three communities using the composition and PD. We then conducted a Mann–Whitney U test to determine whether there was a significant difference between two of the three communities (i.e., comparison–Civano II, comparison–Civano I, and Civano II–Civano I). We made a Bonferroni adjustment to the results of the Mann–Whitney U test to adjust the level of significance so that there was a meaningful difference in means (α = 0.017). All analysis was conducted at the city block level. We compare delivery of three ecosystem service services—microclimate, net primary productivity (NPP), and water provisioning—using indicators for environmental outcomes calculated at the city block scale (Table 2). Microclimate regulation is especially important in arid cities experiencing elevated minimum daily temperatures from the urban heat island (UHI; Brazel et al. 2007). UHI is mitigated through high albedo surfaces and vegetation, the latter of which produces the cobenefit of increasing NPP and its associated benefits (i.e., runoff reduction; Brauman et al. 2007). Increasing NPP in arid environments often requires irrigation from local and nonlocal sources, jeopardizing aquatic ecosystem health and future water availability (Braumen et al. 2007). We represented microclimate regulation by albedo—the ratio of solar short wave radiation reflected from the surface versus the amount absorbed—and daytime temperature. Albedo was estimated from a Quickbird scene, acquired on 13 June 2010 at 18:12 GMT at 2.4-m resolution, by squaring reflectance values calculated from the image. June is considered optimal for studying the summer microclimate conditions because it precedes the
Table 2
monsoon season in the Southwest. We estimated temperature using the thermal band of a Landsat 5 TM scene acquired on 19 June 2010 at 17:48 GMT at 30-m resolution. Desert temperatures are warmer than urban land covers in the morning time because of differences in the heating capacity of urban land covers (Voogt and Oke 2003). We chose a Landsat image to capture thermal conditions in the morning. Even though surrounding desert temperatures are warming faster than some urban land covers at this time, there is tremendous variability for land covers to warm at different rates that are apparent even during the morning time and thus are useful for measuring microclimate regulation (e.g., Jenerette et al. 2007). We used the soil-adjusted vegetation index (SAVI) to indicate vegetative cover as a proxy for primary productivity (Huete 1988; Prince 1991). We selected SAVI, as opposed to the more common normalized difference vegetation index, because it minimizes the influence of soil on vegetation detection (Huete 1988; Qi et al. 1994), which is useful in desert environments where soil can dominate the landscape. We calculated SAVI from the Quickbird scene using the following equation: (Band4 − Band3) ∗ 1.5 . Band4 + Band3 + 0.5 Band 4 covers the near-infrared portion of the electromagnetic spectrum, 760 to 900 nm, and Band 3 covers the red portion at 630 to 690 nm. Vegetation reflects more in the near-infrared part of the spectrum and absorbs more in the red because of photosynthetic activity. To account for the different resolution of the Landsat and Quickbird scenes, SAVI, albedo, and temperature were averaged at the block level. Potable and nonpotable water consumption indicated provisioning of both local (groundwater and recycled) and regional (canal water) water resources. We normalized City of Tucson annual water consumption data by the number of connections per city block. Potable water can be used both indoors and outdoors and is sourced from ground and canal water. Nonpotable water is a substitute used outdoors. We included estimated full cash value (FCV) of single-family homes in our analysis to indicate socioeconomic status (SES) because past research found a positive correlation between wealth and water consumption (Wentz and Gober 2007). Our intention is to link urban design to environmental outcomes and control for SES if
Metrics used in analysis
Measure Microclimate Vegetation Water consumption Affordability
Indicator
Data source
Scale
Year
Temperature Albedo Soil adjusted vegetation index (SAVI) Potable consumption Nonpotable consumption Home full cash value
Landsat Quickbird Quickbird City of Tucson City of Tucson Pinal County Assessor
60 m 2.4 m 2.4 m City block City block Parcel
2011 2011 2011 2010 2010 2011
Downloaded by [Arizona State University] at 12:57 03 July 2014
4 Volume 0, Number 0, XXXX 2014 large, significant differences in wealth among the three communities emerged. We calculated each variable at the city block scale to determine means for Civano I, Civano II, and the comparison community. We used a multinomial logistic regression (MLR; Hosmer and Lemeshow 2000) to differentiate the biophysical (temperature and SAVI) and social (potable water consumption, nonpotable water consumption, and FCV of the plot) covariates between Civano I, Civano II, and the comparison community. The benefit of using an MLR is that the dependent variable can be categorical (e.g., Cao, Kyriakidis, and Goodchild 2011; Tremme and Verburg 2011), including two or more categories. MLR can test which variables are significantly different among the communities by using a probabilistic framework. We used the MLR to estimate coefficients and their significance for the biophysical and social variables as a way of determining how well these variables successfully predict the three development types. We divided the analysis into two MLRs to analyze the social and biophysical covariates separately to keep model development parsimonious (Freedman 1983; Flack and Chang 1987; Hosmer and Lemeshow 2000) and because an initial screening of the variables found that a constant was significant in the model for the biophysical parameters (SAVI and temperature) but not significant for the social (potable water consumption, nonpotable water consumption, and FCV of the plot). In the development of the biophysical model, our goal was to first determine the significant differences in SAVI and temperature. Because we measure temperature and SAVI rather than the services that lead to their outcome, we are in effect measuring the outcomes of microclimate regulation and NPP. We interpret the significance values of SAVI and temperature as indicators of differences in microclimate regulation and NPP. The model for the social variables provides insight into the significant differences related to water provisioning; furthermore, it helps to determine if SES is an additional influence. Finally, ordinary least squares regression models were developed within each community at the block level to correlate land surface temperature (LST; the dependent variable) with albedo and SAVI. We first explored a bivariate comparison of LST with albedo and then with SAVI to show the individual correlation between LST and each of these variables. We also developed multivariate regression models for the individual communities to show how albedo and SAVI simultaneously influence LST differently; this is based on the assumption that each community is designed.
Results Characterizing the Built Environment with Landscape Metrics Average building area (Table 3) was smallest in Civano I (245.5 m2) and largest in the comparison community (325.9 m2), with Civano II in between (309.7 m2).
Table 3 Percentage area and patch density for land-use and land-cover classes in Civano I, Civano II, and the comparison community Class Civano I
Impervious Trees and shrubs Soil LMB MAB Grass Pools Water Civano II Impervious Trees and shrubs Soil LMB HAB Grass Pools Water Comparison Impervious Trees and shrubs Soil LMB HAB Grass Pools Water
Percentage area Patch density 14.9829 29.1785 32.6316 8.7471 12.7938 1.6062 0.0476 0.0123 22.0606 16.0748 37.7255 23.2316 0.1168 0.698 0.067 0.0258 17.3324 22.9375 28.8556 25.8297 1.6521 3.2634 0.1199 0.0093
194.6092 1,176.0555 1,279.6604 1,771.0836 957.6452 238.0112 2.8001 5.6003 337.8262 1,238.219 563.998 344.9836 68.7104 101.6342 4.2944 8.5888 307.9749 1,286.8951 863.4296 537.5812 325.8484 276.3525 48.1211 5.4996
Note: LMB = low and medium albedo buildings; HAB = high albedo buildings.
Civano I had low impervious surface coverage (15.0 percent) and density (194.6). There was a significant difference in both the composition and PD of LMB between Civano I and the other two communities (Table 4). Civano I had 12.1 percent of the land at the block level allocated to HAB, an order of magnitude higher than the comparison community (1.93 percent) and two orders of magnitude higher than Civano II (0.13 percent). The PD of HAB was much higher than the other two communities, too (957.6). Percentage LMB coverage was low (8.7 percent)—much lower than the comparison community and Civano II—and the building size was, on average, the smallest of the three communities (245.3 m2). The LMB PD (1771.1) was high as well. Civano I had a higher percentage of grass (1.6 percent) than Civano II and a lower percentage than the comparison community (3.3 percent) but high PD (238.0). Civano I had the highest percentage of trees and shrubs (29.2 percent). The largest percentage of land was exposed soil (32.6 percent). The percentage of soil included undeveloped land clustered around the western periphery of the community, which partly contributed to the high PD (1279.7). Pools and other water bodies made up less than 1 percent of the area. Civano II had the highest percentage coverage (22.1 percent) and PD (337.8) of impervious surface among the three communities. LMB dominated (23.2 percent), whereas HAB were less than 1 percent of the land cover composition. There was not a significant difference, however, in the composition or PD of LMB between the comparison community and Civano II. The
Do Sustainable Urban Designs Generate More Ecosystem Services? 5 Table 4 Results of the Kruskal–Wallis and Mann–Whitney U test showing the differences in the composition and patch density of select land covers at the block level for each community LMB% LMB PD
Soil%
Soil PD
Tree%
Tree PD
Imp%
Imp PD Grass% Grass PD HAB% HAB PD
0.000∗
0.045∗
0.000∗
0.000∗
0.173
0.003∗
0.000∗
0.098
0.016∗
0.000∗
0.000∗
0.711
0.066
0.059
0.059
0.002∗∗
0.458
0.082
0.073
0.033
0.005∗∗
0.001∗∗
0.001∗∗
0.000∗∗
0.000∗∗
0.992
0.001∗∗ 0.001∗∗
0.374
0.084
0.000∗∗
0.080
0.856
0.000∗∗
0.000∗∗
0.000∗∗
0.000∗∗
0.017∗∗ 0.000∗∗ 0.000∗∗
0.077
0.002∗∗ 0.000∗∗
0.434
0.011∗∗
0.000∗∗
0.000∗∗
Kruskal–Wallis test All three 0.000∗ Mann–Whitney U test Comparison and Civano II Comparison and Civano I Civano II and Civano I
Downloaded by [Arizona State University] at 12:57 03 July 2014
Note: Values are significance levels.% = percentage of landscape; PD = patch density; LMB = low and medium albedo buildings; HAB = high albedo buildings. ∗ Significant at the 0.05 level. ∗∗ Significant at the 0.017 level (using a simple Bonferroni adjustment).
PD in Civano II was 345.0 for LMB and 68.7 for HAB. Civano II had the lowest tree and shrub coverage (16.1 percent) and lower PD (1238.2) than the comparison community. Civano II had the highest percentage of soils (37.7 percent) with low PD (564.0). Less than 1% of Civano II was made up of grass, pools, and other water bodies. The comparison community had slightly higher impervious surface coverage (17.3 percent) and PD (308.0) than Civano I and less than Civano II. Tree and shrub coverage (22.9 percent) was lower than Civano I, whereas PD (1286.9) was slightly higher than Civano I, and both were higher than Civano II. It had the highest percentage of LMB coverage (25.8 percent) and a PD (863.4) higher than Civano II but lower than Civano I. The comparison community had the lowest soil coverage (28.9 percent) with a PD (863.4) lower than Civano I and higher than Civano II. Of all of the communities, it had the highest grass, pool, and water coverage; however, combined they constitute less than 5 percent of the total area. Environmental Variables: Temperature, Albedo, and SAVI We found small differences in surface temperature among the three communities. The mean temperature in Civano I was cooler than in both Civano II and the comparison community, with the comparison community being slightly cooler than Civano II (Figure 2, Table 5). Civano I had a larger standard deviation from the mean temperature across city blocks than both Civano II and the comparison community due to an
Table 5
edge effect from higher temperatures in the surrounding remnant desert. After removing temperature values from city blocks around the edge, the mean temperature in Civano I decreased 0.14◦ C to 31.59◦ C, whereas temperature changes in Civano II and the comparison community were not as pronounced. In Civano II and the comparison community the temperature decreased by 0.02◦ C each to 31.96◦ C and 31.91◦ C, respectively, after negative buffering. Mean albedo was higher in Civano I than in Civano II and the comparison community (Figure 3, Table 6) and Civano I had a larger standard deviation from the mean albedo due to the edge effect from lower albedo roads and desert. Differences in vegetated cover were less pronounced than differences in albedo. Mean SAVI was highest for city blocks in Civano I and the comparison community and lowest in city blocks in Civano II (Figure 4, Table 7). To understand the relative cooling effect of albedo and vegetation, we performed two exploratory bivariate correlation analyses with temperature as the dependent variable—using albedo and vegetation as independent variables—and plotted the results in two scatterplots (Figures 5 and 6). City blocks with high albedo (R2 = 0.328) were a better predictor of city blocks with low temperatures than city blocks with dense vegetation (R2 = 0.258). Low albedo, high temperature city blocks were present in each of the communities; however, high albedo, low-temperature city blocks were predominantly located in Civano I. Similarly, highmean-vegetation, low-mean-temperature city blocks were concentrated in Civano I, and the highest mean temperature, lowest mean SAVI city blocks were located in Civano II.
Mean temperature, ◦ C All blocks
Civano I Civano II Comparison
Negative buffer
M
SD
M
SD
31.73 31.98 31.93
1.70 0.89 0.98
31.59 31.96 31.91
0.97 0.81 0.87
Table 6
Mean albedo
Civano I Civano II Comparison
M
SD
0.130 0.074 0.080
0.164 0.036 0.047
Downloaded by [Arizona State University] at 12:57 03 July 2014
6 Volume 0, Number 0, XXXX 2014
Figure 2 Surface temperature (◦C) derived from a Landsat image acquired in June 2010. The spatial resolution of the thermal image is 30 m. (Color figure available online.)
The results of the block-level regression (Table 8) show that within Civano I, both albedo and SAVI were negatively correlated with temperature, but SAVI was not significant. The significance of albedo and lack of significance of SAVI suggests that albedo has a
Table 7
Mean soil-adjusted vegetation index
Civano I Civano II Comparison
M
SD
0.262 0.178 0.237
0.240 0.151 0.229
stronger correlation to LST for Civano I. Civano II showed a positive correlation between albedo and LST and a negative correlation for SAVI and LST. The positive correlation between albedo and LST is contradictory to what we found in Civano I. Based on the class metrics, however, this positive correlation is somewhat skeptical because the blocks within Civano II are made up of less than 1 percent HAB. In other words, the variance of albedo within Civano II is mostly from the LMB range. The comparison community shows a negative correlation between LST and albedo but a positive correlation between LST and SAVI. Both albedo and SAVI were not significant, however.
Downloaded by [Arizona State University] at 12:57 03 July 2014
Do Sustainable Urban Designs Generate More Ecosystem Services? 7
Figure 3
Albedo derived from a Quickbird image acquired in June 2010 at 2.4-m resolution.
The multinomial regression of temperature and SAVI in the three communities revealed that lower temperature blocks were more likely to be in Civano I rather than Civano II or the comparison community (Table 9). Highly vegetated blocks were more likely to be in Civano I as opposed to Civano II, but vegetation was not significantly different between Civano I and the comparison community.
Social Variables: Potable and Nonpotable Water Consumption and Full Cash Value The mean normalized potable water consumption for city blocks in both Civano II and Civano I was lower than that in the comparison community because the comparison community does not supplement potable water sources with nonpotable supplies (Figure 7, Table 10). Mean normalized nonpotable water
Downloaded by [Arizona State University] at 12:57 03 July 2014
8 Volume 0, Number 0, XXXX 2014
Figure 4 Soil-adjusted vegetation index (SAVI), derived from a Quickbird image acquired in June 2010 at 2.4-m resolution. (Color figure available online.)
consumption in Civano II was greater than in Civano I; however, the standard deviation was high in Civano II, and city blocks using high amounts of nonpotable water were only slightly more likely to be located in Civano II as opposed to Civano I (Figure 8, Table 11). Because the standard deviation from the mean nonpotable water consumption was so large, we also calculated the median. The mean and median were sim-
ilar for Civano I; however, the median for Civano II was much lower, highlighting that a few high values skewed the mean. Civano I property values were slightly higher than values in Civano II and the comparison community (Table 12). Mean FCV of single-family homes was highest in Civano I and lowest in Civano II. The standard deviation was highest in Civano II and Civano
Downloaded by [Arizona State University] at 12:57 03 July 2014
Do Sustainable Urban Designs Generate More Ecosystem Services? 9
Figure 5 The relationship between temperature and albedo at the block level for all three communtities. Civano II and the comparison community show a higher concentration of high mean temperature and low mean albedo blocks, whereas Civano I shows a higher number of blocks with high mean albedo and lower mean temperatures.
Figure 6 The relationship between temperature and soiladjusted vegetation index (SAVI) at the block level for all three communities. The scatterplot shows a negative correlation between SAVI and temperature. Civano I had higher mean SAVI at the block level than Civano II and the comparison community.
Downloaded by [Arizona State University] at 12:57 03 July 2014
10 Volume 0, Number 0, XXXX 2014
Figure 7 Mean normalized potable water consumption per city block (CCF). Potable water consumption was normalized by the number of water connections per city block to correct for differences in concentration of households. City blocks in Civano I and Civano II have lower potable water consumption than the comparison community. (Color figure available online.)
I and lowest in the comparison community, revealing greater diversity in housing price points in the two Civano communities. In the MLR model of potable and nonpotable water consumption and FCV (Table 13), city blocks with less potable water consumption were more likely to be located in Civano I as opposed to the comparison community but not significantly more likely to be located in Civano I than Civano II. Blocks with high nonpotable water use were slightly more likely to be located in Civano II than in Civano I. The comparison lacked nonpotable water connections. Higher FCV blocks were slightly more likely to be located
in Civano I than Civano II. Differences in FCV were not significant between Civano I and the comparison community.
Discussion Our findings reveal differences in environmental outcomes across three different urban designs. Civano I most successfully regulated microclimate by lowering surface temperature through higher prevalence and clustering of albedo building materials and more vegetative cover in the community design than Civano II
Do Sustainable Urban Designs Generate More Ecosystem Services? 11 Table 8 Multivariate regression models for each community that show the correlation between land surface temperature (the dependent variable) and albedo and soil-adjusted vegetation index (SAVI; the independent variables) Civano 1
Downloaded by [Arizona State University] at 12:57 03 July 2014
Constant Albedo SAVI R2
Civano 2
Coefficient
Significance
Coefficient
Significance
Coefficient
Significance
31.931 –1.137 –0.497
0.000 0.023 0.107 0.132
31.757 4.711 –0.832
0.000 0.018 0.004 0.494
31.845 –1.890 1.103
0.000 0.338 0.286 0.236
and the comparison community. These findings are consistent with other studies that find cooling capacity of light-colored roofs and vegetation combined with the clustered spatial arrangement of different land covers appears to generate the lowest microclimate temperatures (Bonan 2000). Civano I potentially generated the highest primary productivity, given previous studies linking irrigated landscapes and vegetative density to increases in arid environments (Haberl et al. 2007). One potential explanation for differences in vegetation is that Civano II was built later and the vegetation has yet to mature. The comparison community, however, also had denser vegetation than Civano II and it was completed last, although much of that vegetative density was turf grass. Another explanation is that Civano I salvaged 80 percent of the native vegetation during construction, whereas Civano II used new plantings that have yet to mature. Furthermore, the urban design of Civano I clusters buildings to maximize open space available for contiguous desert vegetation and common areas. Civano I and Civano II lowered the potable water resource consumption by supplementing that supply with nonpotable water resources. Mean potable water consumption in the Civano communities was almost 40 cent cubic feet (CCF) lower than in the comparison community annually; however, mean nonpotable water consumption in Civano I and even more so in Civano II was highly variable. Both communities contained city blocks with one or few nonpotable service lines and high nonpotable water consumption as well as city blocks with several nonpotable service lines with no or little nonpotable water consumption. One explanation for the differences lies in the implementation of building and design aimed at lowering water provisioning in the two phases of development.
Table 9 Multinomial regression of environmental variables
Comparison
Civano II
Comparison
Intercept SAVI Temperature Intercept SAVI Temperature
Beta
Significance
–812.251 –2.038 25.485 –828.735 –22.638 26.141
0.003 0.812 0.003 0.005 0.038 0.005
Note: Civano I is the reference category. SAVI = soil-adjusted vegetation index.
Civano I required nonpotable water connections on in all residential lots, whereas Civano II made them optional and directed nonpotable resources to common areas and open spaces (Nichols and Laros 2009). Although both Civano I and Civano II used nonpotable resources to irrigate outdoor vegetation, city blocks in Civano I consumed less potable water to achieve denser vegetation. This result suggests that Civano I more efficiently utilized nonpotable resources than Civano II. It is likely that the new plantings in Civano II also required large water inputs compared to the more established vegetation in Civano I. The fact that Civano II directed such large quantities of nonpotable water resources to a handful of city blocks suggests that nonpotable water resources are used to water common areas and open spaces. Guhathakurta and Gober (2007) found that water inputs to irrigated landscapes in Phoenix, Arizona, reach a threshold beyond which point increased water input no longer generates quantifiable increases in vegetative cooling. In light of such findings, the high nonpotable water inputs to open spaces in Civano II might not translate to increases in vegetative density and cooling. The comparison community did not use any nonpotable water for irrigation purposes and the data do not reveal how much potable water might be allocated for outdoor uses. It does show that overall mean potable water consumption was much higher in the comparison community than in both Civano I and Civano II. Our results point to the potential use of high albedo roofing material as a substitute for vegetative cover for microclimate regulation. High albedo roofs are a particularly attractive option in arid environments with limited water resources for irrigated landscapes. The clustered arrangement of buildings in Civano I might amplify the cooling effect through contiguous high albedo area and minimize trade-offs associated with increasing building area (i.e., loss of open space). Furthermore, there was quite a bit of heterogeneity in temperature, albedo, and vegetative cover within each community. This internal variation in temperature
Table 10
Civano I Civano II Comparison
Mean potable water consumption M
SD
65.25 63.94 103.56
22.15 26.99 23.48
Downloaded by [Arizona State University] at 12:57 03 July 2014
12 Volume 0, Number 0, XXXX 2014
Figure 8 Mean normalized nonpotable water consumption per city block (CCF). Nonpotable water consumption was normalized by the number of water connections per city block to correct for differences in the concentration of households. The comparison community did not have any nonpotable water connections. (Color figure available online.)
in Civano I suggests that some portions of the urban design could generate significant cooling, especially if replicable at larger scales. High albedo roofs might be an effective mechanism for microclimate regula-
tion; however, as a substitute for vegetative cover, high albedo roofs cannot replace the full range of services that vegetative cover provides, including habitat, flood regulation, and soil retention.
Table 11 Mean and median nonpotable water consumption
Table 12
Civano I Civano II Comparison
M
SD
Median
68.52 428.53 N/A
78.82 402.72 N/A
48.99 171.13 N/A
Civano I Civano II Comparison
Mean full cash value (FCV) in dollars M
SD
198,964 155,980 157,537
39,135 49,318 26,992
Do Sustainable Urban Designs Generate More Ecosystem Services? 13 Table 13
Multinomial regression of social variables Beta
Comparison
Civano II
Potable norm Nonpotable norm FCV Potable norm Nonpotable norm FCV
0.045 –10.444 –0.227 0.019 0.003 –0.177
Significance 0.026 N/A 0.061 0.135 0.041 0.003
Note: Civano I is the reference category. FCV = full cash value.
Downloaded by [Arizona State University] at 12:57 03 July 2014
Project Scope and Data Limitations In this study we recognize data and analytical constraints. We were unable to capture daily or seasonal variations because we were limited to a single-time snapshot of the Quickbird image. The surface UHI effect is generally more pronounced in cities during the evening when the thermal holding capacity of impervious surfaces stay warmer well into the evening hours. Conversely, the desert is hotter during the late morning and throughout the day (Brazel et al. 2007) but then cools off as the sun sets. Due to data limitations, nighttime temperature could not be evaluated for all three communities from 2007 onward, so we restricted our study and discussion to daytime temperature effects. The differences in temperature during the late morning between the communities gave an indication of sustainable outcomes that are related to energy usage and microclimate regulation. The lack of nighttime temperature data might suppress differences in temperature due to urban design because the dense vegetation and high albedo roofs in Civano I are more likely to cool in the evening, whereas the high impervious surface coverage in Civano II is more likely to retain heat. Future studies should consider nighttime surface temperatures in addition to daytime surface temperatures. We encountered two of the major challenges identified by Yang (2011) in the use of remotely sensed data in urban analysis: (1) integrating remotely sensed data with other social and environmental data sets and (2) capturing interpixel complexity. Our study was constrained by the city block scale for which the water consumption data were available, limiting our use of the 2.4-m resolution Quickbird in the analysis of intraneighborhood heterogeneity. There is also a descriptive gap between the relatively course land-cover classes detected through remotely sensed imagery—even at the 2.4-m resolution—and the finegrain differences in urban and landscape design that might influence environmental performance. For instance, remotely sensed data cannot differentiate between plant species, which has implications for water consumption and biodiversity.
Conclusion Our results show that differences in urban form and urban design contribute to moderate differences in the environmental outcomes at the neighborhood scale
with implications for the provisioning of local ecosystem services. The design of Civano I appears to reinforce the benefits of multiple ecosystem services because it used less nonpotable water to achieve more lush vegetation and lower temperatures than Civano II. These findings begin to address a gap in the literature empirically linking sustainable urban design to environmental outcomes. Future research should continue to expand on these findings through additional case studies unified by the ecosystem services framework. Other ecosystem services—for instance, those tied to biodiversity—merit further investigation as well. This study also explored the potential to use land change science methods that are typically deployed at a landscape scale in a local urban context. Future efforts will require confronting several limitations tied to the use of remotely sensed spatial data that might be too coarse to adequately detect small nuances in urban and landscape design. Combining remotely sensed spatial analysis with locally collected data begins to address some of these challenges and aids in describing fine-scale biophysical differences. Environmental sustainability can be interpreted in many ways, and refining existing frameworks and methods from geography and cognate fields can inform efforts to urbanize sustainably by providing quantifiable metrics for measuring and monitoring a broad suite of environmental outcomes.
Literature Cited Alberti, M. 2005. The effects of urban patterns on ecosystem function. International Regional Science Review 28:168–92. Bonan, G. B. 2000. The microclimates of a suburban Colorado (USA) landscape and implications for planning and design. Landscape and Urban Planning 49:97–114. Brauman, K. A., G. C. Daily, T. K. Duarte, and H. A. Mooney. 2007. The nature and value of ecosystem services: An overview highlighting hydrologic services. Annual Review of Environment and Resources 32 (1): 67–98. Brazel, A., P. Gober, S. Lee, S. Grossman-Clarke, J. Zehnder, B. Hedquist, and E. Comparri. 2007. Determinants of changes in the regional urban heat island (1990–2004) within metropolitan Phoenix. Climate Research 33: 171–82. Cao, G., P. C. Kyriakidis, and M. F. Goodchild. 2011. A multinomial logistic mixed model for the prediction of categorical spatial data. International Journal of Geographical Information Science 25 (12): 2071–86. Conway, T. 2009. Local environmental impacts of alternative forms of residential development. Environment and Planning B: Planning and Design 36 (5): 927–43. Flack, V., and P. Chang. 1987. Frequency of selecting noise variables in subset regression analysis: A simulation study. American Statistician 41 (1): 84–86. Foley, J. A., R. DeFries, G. P. Asner, C. Barford, G. Bonan, S. R. Carpenter, F. S. Chapin, et al. 2005. Global consequences of land use. Science 309:570–74. Freedman, D. 1983. A note on screening regression equations. American Statistician 37 (2): 152–55. Guhathakurta, S., and P. Gober. 2007. The impact of the Phoenix urban heat island on residential water use. Water Resources 73 (3): 317–30.
Downloaded by [Arizona State University] at 12:57 03 July 2014
14 Volume 0, Number 0, XXXX 2014 Haberl, H., K. H. Erb, F. Krausmann, V. Gaube, A. Bondeau, C. Plutzar, S. Gingrich, W. Lucht, and M. FischerKowalski. 2007. Quantifying and mapping the human appropriation of net primary production in earth’s terrestrial ecosystems. Proceedings of the National Academy of Sciences 104 (31): 12942–47. Hosmer, D., and S. Lemeshow. 2000. Applied logistic regression. New York: Wiley. Huete, A. 1988. A soil-adjusted vegetation index (SAVI). Remote Sensing of Environment 25 (3): 295–309. Jenerette, G. D., S. L. Harlan, A. Brazel, N. Jones, L. Larsen, and W. Stefanov. 2007. Regional relationships between surface temperature, vegetation, and human settlement in a rapidly urbanizing ecosystem. Landscape Ecology 22:253–365. Jenerette, G. D., S. L. Harlan, W. L. Stefanov, and C. A. Martin. 2011. Ecosystem services and the urban heat riskscape moderation: Water, green spaces, and social inequality in Phoenix, USA. Ecological Applications 21 (7): 2637–51. Millennium Ecosystem Assessment. 2003. Ecosystems and human well-being: A framework for assessment. Washington, DC: Island Press. Myint, S., C. S. Galletti, S. Kaplan, and W. K. Kim. 2013. Object vs. pixel: A systematic evaluation in urban environments. Geocarto International 28 (7): 657–78. Nichols, A., and J. Laros. 2009. Inside The Civano Project: A case study of large-scale sustainable neighborhood development. New York: McGraw Hill. Prince, S. D. 1991. Satellite remote sensing of primary production: Comparison of results for Sahelian grasslands 1981–1988. International Journal of Remote Sensing 12 (6): 1301–11. Qi, J., A. Chehbouni, A. R. Huete, Y. H. Kerr, and S. Sorooshian. 1994. A modified soil adjusted vegetation index. Remote Sensing of Environment 45 (2): 119–26.
Tremme, A. J. A. M., and P. H. Verburg. 2011. Mapping and modelling of changes in agricultural intensity in Europe. Agriculture, Ecosystems & Environment 140 (1–2): 46–56. Voogt, J. A., and T. R. Oke. 2003. Thermal remote sensing of urban climates. Remote Sensing of Environment 86 (3): 370–84. Wentz, E., and P. Gober. 2007. Determinants of small-area water consumption for the city of Phoenix, Arizona. Water Resources Management 21 (11): 1849–63. Wentz, E. A., K. C. Seto, S. W. Myint, M. Netzband, and M. Fragkias. 2011. Urban remote sensing (URS) and forecasting urban landuse (FORE) workshops: Common ground and targeted opportunities. UGEC Viewpoints 15–17. http://ugec.org/docs/ugec/viewpoints/ugecviewpoints-6.pdf (last accessed 5 November 2011). Yang, X., ed. 2011. Urban remote sensing: Monitoring, synthesis and modeling in the urban environment. Oxford: WileyBlackwell.
V. KELLY TURNER is an Assistant Professor of Geography at Kent State University, 436 McGilvrey Hall, P.O. Box 5190, Kent, OH 44242–0001. E-mail:
[email protected]. Her research interests include urban planning approaches to environmental sustainability and decision making and management of urban ecosystems. CHRISTOPHER S. GALLETTI is a PhD candidate at Arizona State University, School of Geographical Sciences and Urban Planning, Coor Hall, 975 S. Myrtle Ave., Fifth Floor, P.O. Box 875302, Tempe, AZ 85287–5302. E-mail:
[email protected]. His research interests include land change science, remote-sensing, and GIS approaches to understanding global change in paleoenvironments and arid and semiarid landscapes.
