Land system architecture for urban sustainability: new directions for land system science illustrated by application to the urban heat island problem

ABSTRACT

Land system science (LSS) has expanded its research focus from the drivers of land-use and -cover change primarily in rural wildlands to include the social-environmental consequences of this change, urban areas, and sustainability practice. Land system architecture, interacting with the landscape mosaic approach in ecology, offers a special niche for the entry of rural areas and wildlands into urban sustainability research through examinations of the composition and configuration of ‘cityscapes’. Given the fine-grain data requirements of heterogeneous cityscapes, emergent land architecture-mosaic approaches have largely explored the urban heat island (UHI) problem, a topic that links LSS with the interests of urban climatology, engineering, and planning in city morphology or geometry. The subtle distinctions in the treatment of land configuration between land architecture-mosaic approaches and urban morphology-geometry approaches are identified. Several examples of the land architecture-mosaic approach illustrate the understanding gained about the UHI problem as well as its complementarity with morphology-geometry approaches. This understanding provides insights about the design of urban areas at the parcel to neighborhood scales to ameliorate extreme temperatures, an issue of increasing concern for urban areas worldwide and consistent with the sustainability problems identified by such international programs as Future Earth.

KEYWORDS:

• Land system architecture
• landscape mosaic
• land configuration
• urban heat island
• sustainability

1. The expansion of land system science toward urban sustainability

Land system science (LSS) continues to advance in sync with the larger international science agendas addressing global environment change and sustainability. Its contemporary origin was strongly influenced by the need of the environmental sciences, foremost ecology, to improve understanding of land-use and -cover changes in regard to the functioning of Earth system and subsequently social-environmental systems (SESs) at large (Turner, Lambin, & Reenberg, 2007). This charge led to major improvements in earth surface observation and monitoring of land covers (e.g., Gutman et al., 2004), understanding the drivers of land use and the consequences for land-cover change (e.g., Lambin et al., 2001), and modeling these dynamics (National Research Council [NRC], 2014; Parker, Manson, Janssen, Hoffmann, & Deadman, 2003). As LSS matured in concert with its parent international program, the Global Land Project (GLP) (Global Land Project [GLP], 2005), it expanded its reach to such issues as: the history and global reach of human-induced land change and its role in the Anthropocene (e.g., Dearing, Braimoh, Reenberg, Turner, & van der Leeuw, 2010; Ellis, 2011; Ellis, Goldewijk, Siebert, Lightman, & Ramankutty, 2010); land-change impacts on environmental (ecosystem) services (Foley et al., 2005), biodiversity (Bennett et al., 2015), and the vulnerability and resilience of SESs (Turner, 2010); governance impacts on land system performance (Ostrom & Nagendra, 2006); and sustainability (Verburg et al., 2015). More recently LSS has coupled research interests with the Urbanization and Global Environmental Change (UGEC) program (Seto & Reenberg, 2014; Seto et al., 2012). Much of this coupling engages the sustainability research directions of Future Earth (Liverman et al., 2013), the larger, governing program in which both the GLP and UGEC are embedded. Future Earth, in turn, promotes a science that is capable of informing societal decisions regarding sustainability problems.

This commentary melds these two recent research directions emerging within LSS – urban and sustainability. While the range of urban sustainability problems is large, I make the case that LSS has begun and can continue to serve a special research niche important to this problem domain. This niche involves the role of land composition and configuration, or what has been labeled land system architecture (Turner, Janetos, Verburg, & Murray, 2013), on environmental services. A brief illustration of this problem lens is given for the urban heat island (UHI) effect, an environment (dis)service and sustainability issue that confronts urban areas and decision-makers worldwide but especially in hot, arid climates that grow more so.

In making this case, I [1] outline the research links between the land system architecture and the landscape mosaic approaches in ecology (henceforth, land architecture-mosaic) in their focus on the pattern and shape of land covers as they affect SES sustainability problems; [2] assert that these approaches as applied to the UHI problem maintain subtle but important distinctions with those of urban morphology and geometry in urban climatology, engineering and urban planning, foremost involving the shape of multiple, fine-grain land covers; [3] provide several examples of land architecture-mosaic research within LSS that illustrate the kind of insights provided about parcel- to neighborhood-level architecture–UHI relationships and their complementarity with morphology-geometry research; and [4] note that the urban (re)design lessons emerging from this research are of interest to those responsible for urban sustainability and are the kind of LSS products sought by the Future Earth program, that in which the LSS and GLP are housed.

2. Land system architecture and landscape mosaics: linked approaches

The investigative interests of LSS overlaps with various research communities, foremost ecology and landscape ecology (Wu, 2013), resilience (e.g., Folke et al., 2002; Janssen, Schoon, Ke, & Börner, 2006), remote sensing science (e.g., Justice et al., 1998), and political ecology (Turner & Robbins, 2008). Of particular importance at this moment are links to landscape ecology regarding the composition and configuration of land covers and their impacts on SESs and environmental services (Forman, 1995; Rosenzweig, 2003; Wu, 2013), in which configuration refers to the pattern, shape, and connectivity of land covers (also land use). The resulting land architecture-mosaic approaches share configuration metrics developed largely in ecology (e.g., FRAGSTATS: McGarigal, Cushman, Neel, & Ene, 2012) and analytics developed in the spatial and GIS sciences. These links are signaled by shared interest in land or landscape configuration between the GLP and the EcoServices program, both of Future Earth (e.g., Bennett et al., 2015).

Application of land architecture-mosaic approaches to urban areas or ‘cityscapes’ is nascent for both communities, however. LSS has only recently joined UGEC to develop specific research foci on urbanization, land change, and sustainability (above). Urban ecology, in contrast, has grown over the last several decades by way of research labs, centers, and degree programs worldwide (e.g., Alberti et al., 2003). Urban sustainability themes have been fostered by the two urban-based Long-term Ecological Research (LTER Network, National Science Foundation [NSF]) units in the United States (Grimm et al., 2008), while the Urban Long-term Research Area exploratory program of NSF promises more. Despite this research direction, a focus on Foreman’s (2014, 2016) application of the landscape mosaic approach to urban areas, especially in regard to environmental services, has not yet emerged strongly in this research community. It has been recognized (Alberti et al., 2003; Grimm et al., 2008; Pickett & Cadenasso, 2007), along with various calls for studies of urban spatial heterogeneity and patch dynamics that imply attention to land configuration as defined here (Cadenasso & Pickett, 2008), and has been variously applied to landscape patterns (largely distribution) of urban gradients (e.g., Luck & Wu, 2002). Much of the actual research to date, however, tends to focus primarily on the composition and location or pattern of land units (e.g., Troy, Grove, O’Neil-Dunne, Pickett, & Candenasso, 2007) or on ecological and biodiversity issues not necessarily linked explicitly to environmental services (Alberti et al., 2003; Ramalho & Hobbs, 2012).

Land architecture-mosaic approaches ultimately can be employed to address a large range of urban (as well as nonurban) sustainability issues, such as consequences for water use and aquifer recharge, flooding, invasive species, and human health or the trade-offs among them. Many of these issues have important dimensions that require assessment from the parcel to neighborhood levels such as parcel landscaping and its impacts on temperature and water use (e.g., Wentz, Rode, Li, Tellman, & Turner, 2016). Assessments at this level are facilitated by fine-grain satellite and photogrammetric data to generate the heterogeneous, small patches of land covers involved. Such data are increasingly available, if often expensive. Interestingly, fine-grain environmental data for urban areas, such as biota distribution and interactions, surface water movement or soil contamination, are commonly sparse and usually insufficient in kind to examine a large range of trade-offs among environmental services and in spatial reach to cover large metropolitan areas. Even air temperature and precipitation data, which are abundant, tend to be point source in kind, posing problems of matching them to large areas of fine-grain, heterogeneous land covers. Surprisingly, much fine-grain socioeconomic and demographic data are either not public (e.g., household water use in Arizona) or must be reported in aggregate units, such as census blocks, for privacy issues. Such constraints impede full multi-scale analyses, foremost at subblock levels that can prove essential for making adequately informed urban sustainability decisions as illustrated below.

3. Urban heat island: interactions with urban climatology, engineering, and urban planning

To date, exploration of the land architecture-mosaic approaches in urban areas has focused on the UHI problem in part because land-cover classes can be matched to land surface temperature (LST) at multiple spatial scales (1–250 m) across entire ‘cityscapes’ from remotely sensed data. In this exploration of the surface UHI effect, LSS, broadly interpreted to include remote sensing of the environment, and urban ecology, encounter a research issue originally identified and examined by urban climatologists and engineers (e.g., Erell, Pearlmutter, & Williamson, 2012; Gago, Roldán, Pacheco-Torres, & Ordoñez, 2013; Taha, 1997), often with links to urban planning and urban-environment research, such as that found in UGEC (Seto, Solecki, & Griffith, 2015; Solecki et al., 2005).

Urban climate research has long examined urban morphology or geometry. The attributes addressed, however, differ somewhat from those in land architecture-mosaic approaches. Given its focus on air, as opposed to surface, temperature and the biophysical processes involved, climate research referencing morphology-geometry typically treats composition as buildings, including white and green roofs and ‘smart’ structures, impervious surfaces, trees, and bare soil (e.g., Coseo & Larsen, 2014; Giridharan, Lau, Ganesan, & Givoni, 2007; Li, Bou-Zeid, & Oppenheimer, 2014; Li, Wang, Wang, Ma, & Zhang, 2009; Li et al., 2011; Oke, 1981; Oleson, Brown, & Feddema, 2010; Santamouris, 2013). Configuration characteristically involves the vertical dimensions (e.g., building heights, urban canyon height-to-width ratios, and skyviews) and density and patterning of the land-cover composition (e.g., Hart & Sailor, 2009; Smith & Levermore, 2008; Stewart & Oke, 2012; Stone & Rodgers, 2001). The shape (horizontal configuration) of land covers has been addressed much less, and connectivity has tended to focus on street directions relative to urban street canyons (e.g., Bourbia & Boucheriba, 2010). Indeed, recent reviews directed to urban planners regarding the amelioration of the UHI effect basically focus on the vertical dimensions of buildings and distributions of green spaces and impervious surfaces (e.g., Gago et al., 2013; Kleerekoper, van Esch, & Baldiri Salcedo, 2012). In contrast, land architecture-mosaic approaches tend to address more fine-tuned distinctions in the types of land covers examined (e.g., variance in vegetation cover of green spaces), and in addition to their distributional patterns, increasingly address shape, either for individual classes of land cover or the aggregate of multiple land covers. These approaches, however, have not paid as much attention to the connectivity of urban land units or to the vertical dimensions of them, perhaps because of the emphasis to date on LST (see below).

These distinctions are subtle but important. The LSS and urban ecology attention to multiple classes of land covers (or patches) at the fine-grain level increasingly provides insights about their pattern and shape on LST (see below). Importantly, the specificity in land-cover types at fine-grain spatial scales is that which facilitates trade-off assessments among different environmental services, especially identifying appropriate spatial scales for various interventions regarding disservices. Equally important for the UHI problem, this orientation is highly complementary with that emanating from urban climatology and engineering.

4. Land mosaic and architecture approaches to the UHI: examples

A significant amount of the land architecture-mosaic research attention to the UHI has been undertaken in context of the two urban LTERs in the United States as well China’s increasing concern about the environmental conditions of their cities. Various examinations employ 1 m–30 m satellite data (preferably the more fine-grain to capture urban land-cover heterogeneity) to create and spatially overlay land-cover classifications and LST. These data are commonly addressed by FRAGSTAT configuration metrics (McGarigal et al., 2012) and regression analyses. In some cases, especially those associated with the urban LTERs, ecologists, land system scientists, urban climatologists and engineers, and urban planners comprise integrated research teams addressing the UHI effect or work more or less as specialist teams paralleling of one another on the topic.

Several examples illustrate the direction and findings of the research focused on configuration via explicit or implicit land architecture-mosaic approaches, liberally defined. They are arranged to indicate the enlargement of configuration considerations from pattern to shape. Research that loosely fits these approaches but is applied to large segments of urban areas constituting a few general land-cover categories (e.g., dense urban vs. peri-urban cover) or to land cover or its change without attention to the specificities of configuration is not considered (e.g., Buyantuyev & Wu, 2010; Huang, Li, Zhao, & Zhu, 2008; Li et al., 2009; Su, Gu, & Yang, 2010; Zhang et al., 2013).

1. Zheng, Myint, and Fan (2014) employed QuickBird (resampled to 3 m) to derive seven land-cover classes, ASTER (90 m) for surface kinetic temperature data, and local Moran’s I to examine the impact of land-cover patterns on LST in the Phoenix, Arizona metro area. They found that paved surfaces have the largest impact on increasing LST, especially during the nighttime, amplified by clustered patterns of these surfaces, which generated the largest effect when the paved surface fraction exceeded 50%.

2. Fan, Myint, and Zheng (2015) combined QuickBird land-cover and ASTER LST data analyzed by way of Moran’s I and ordinary least squares (OLS) regression to determine the most effective pattern of green spaces (grass and trees) for reducing heat in Phoenix. They found that clustered, as opposed to dispersed, patterns of green spaces proved to be most effective in reducing LST and that a resolution of about 200 m was the most optimal for observing this effect.

3. Myint et al. (2015) used Geoeye-1 (3 m), along with the data and analytical methods of Fan (#2 above), to examine the most effective spatial pattern of multiple land-cover classes on reducing LST in Phoenix and Las Vegas, Nevada. They demonstrated that clustered patterns of green land covers reduce LST, while that of impervious surfaces and exposed soil increase it.

4. Li, Zhou, Ouyang, Xu, and Zheng (2012) employed SPOT (30 m) land-cover data and Landsat TM-derived LST data (30) along with six FRAGSTAT metrics to address the consequences of the pattern of green spaces in Beijing, China, on LST using OLS regressions and spatial autocorrelation methods. They found that increases in the percent area of greenspace cover (composition) to be the principal factor decreasing LST, but decreases in the patch and edge density of this cover (pattern and shape) was also a significant factor, controlling for area covered.

5. Middel, Häb, Brazel, Martin, and Guhathakurta (2014) addressed the impact of urban morphology on near ground temperature in Phoenix. Using the three-dimensional microclimate model ENVI-met, they examined 13 residential forms with various vegetation dimensions, representing 5 neighborhood patterns of the local climate zone classification by Stewart and Oke (2012) using multiple regressions. Daytime cooling was improved by the vegetation class, as well as the form and spatial arrangement of high density land-cover composition. Compactness, in this case of housing, was the most beneficial for daytime cooling, and at the microscale, urban (housing) form may influence temperature more than class of landscaping (e.g., xeric or mesic).

6. Zhou, Huang, and Cadenasso (2011) used Landsat ETM+ data to estimate LST, 0.6 m land-cover data to derive six land-cover classes, a variety of FRAGSTAT metrics, and correlation analyses and multiple linear regressions to address the composition and configuration of the Gwynns Falls watershed of peri-urban and urban Baltimore, Maryland. They found that the percent area covered by buildings and green spaces (composition) was the most significant factor increasing or decreasing LST, respectively. Holding this composition constant, however, various configuration metrics significantly affected LST, indicating that the overall configuration of these land covers was important to UHI mitigation.

7. Connors, Galletti, and Chow (2013) used QuickBird and ASTER to address three types of 240 m land-cover units in Phoenix – mesic and xeric residential, and industrial/commercial units – with Pearson correlation coefficients, analysis of variance, and OLS regression. While each land unit type holds distinctive LST associations, the proportion of green land cover (composition) best explained cooler LST among mesic (abundant vegetation in parcels) residences, whereas the configuration – pattern and shape – of green and impervious covers proved most important for reducing and increasing, respectively, LST in industrial/commercial units.

8. Li et al. (2016) examined residential parcels throughout the Phoenix metropolitan area. They used 1 m National Agricultural Imagery Program (NAIP) and Landsat ETM+ (decomposed to 30 m), cadastral and census data, the normalized moment of inertia (NMI), rather than FRAGSTATS, and regression methods to address the role land-cover composition and configuration on LST. Calculating the compactness of individual land covers and parcel-aggregate land covers by way of the NMI, the configuration of land cover proved more robust than composition in explaining LST, and combined with cadastral and census data, produced outcomes almost as strong as those in which all variables were addressed, including land-cover patterns.

9. Li, Kamarianakis, Ouyang, Turner, and Brazel (under review) addressed the role of land composition and configuration of single family residential parcels on daytime and nighttime LST for the Phoenix metropolitan area. They used five land classes common to residential units derived from NAIP data, LST derived from MASTER data (6.8 m; Hook, Myers, Thome, Fitzgerald, & Kahle, 2001), and applied FRAGSTATS metrics in linear mixed-effects models. Nighttime LST models proved to be the most accurate, and grass and swimming pools (composition) proved to most important for reducing LST. Controlling for composition and socioeconomic factors, however, the land shape index and patch density (pattern) proved significant for reducing LST.

This research informs us that attention to urban land configuration and SES outcomes, consistent with land architecture-mosaic approaches, is increasing. Its application to the UHI effect has focused primarily on LST applied to the neighborhood and parcel levels. This configuration, both pattern and shape, proves to be significant, especially holding land composition constant and accounting for spatial autocorrelation. Clustering (pattern) specific kinds of land covers tends to amplify their positive (e.g., impervious surfaces) or negative (e.g., green spaces) impacts on LST, but the shape of this clustering matters as well. Less compact shapes (i.e., large edge density or small NMI) apparently reduce the amplification effect, whereas more compact shapes (small edge density or large NMI) enhance it. These relationships appear to hold for individual land covers per parcel or parcel-level aggregate land covers and similarly for neighborhoods. Unfortunately, none of the studies in question directly addressed connectivity, so its role, if any, on LST remains unexplored at this time.

These insights are just a few drawn from land architecture-mosaic assessments of LST to date, insights that will surely grow in number and specificity as the spatial scales, environmental settings, specific land covers, new metrics (such as NMI), and analyses that move beyond simple linear regressions and spatial autocorrelations are systematically examined. The expertise of the LSS community, especially remote sensing, land dynamics, and spatial analysis, has an important role to play in this research. In collaboration with ecologists, urban climatologists and engineers, and urban planners – some of which characterizes the selected examples above – this kind of research provides a ‘hybrid vigor’ to LSS and the GLP’s vision of it. Such collaborations will surely lead to the incorporation of the vertical structure of land covers and connectivity among land units, as well as addressing near surface-and canopy-level air temperature as opposed to LST. Addressing the problem dimensions fostered by the architecture-mosaic approaches should prove useful for other urban climate issues such as energy-balance modeling, water demands, human thermal comfort, and for the trade-offs between climate and other environmental services and human well-being.

5. Advancing land system science in sustainability

As LSS enters Future Earth and questions of sustainability in the Anthropocene, it increasingly engages science to inform practice, or what Lubchenco (1998) labeled as science’s ‘new social contract’. This contract does not ask LSS to make sustainability decisions but to provide science-based understanding to inform those who do make them. The land architecture-mosaic approaches to sustainability problem formation and resolution offer one pathway for LSS to contribute to this contract. In the case of the UHI effect, state, county, and city agencies and officials are already engaging efforts to ameliorate extreme temperatures, sea level rise, and other such looming problems (e.g., Solecki et al., 2005). The City of Phoenix, Arizona, for example, has developed a master plan that involves siting green spaces throughout the metro area (City of Phoenix, 2010) to combat the UHI effect, one in which the pattern and shape created by this siting matters in regard to the overall cooling effects and trade-offs with water and energy consumption. Attention in this metro area is also being given to parcel-level vegetation, not only for temperature but for water use implications (e.g., Wentz et al., 2016). The approaches championed here need not be constrained to questions of the UHI, however. The composition and configuration of land units is an essential part of the structure of all land-based SESs, urban or otherwise, maintaining important consequences for their functioning. Attention to the attributes of land cover is an essential part of LSS and the use of them for sustainability problems helps to situate LSS and the GLP firmly within the goals and challenges of Future Earth.

Disclosure statement

No potential conflict of interest was reported by the author.