Introduction
Urban spatial structure is important to understanding social and ecological interactions between the build and natural environment and provides a bridge to sustainable development (Zhou et al., 2017). Characteristics including vegetation and other landcover classes influence, and can be used to estimate ecological functions (Bastian et al., 2014; van Oudenhoven et al., 2012) and forecast changes(Dietze, 2019; Dietze et al., 2018) that are crucial under global change scenarios. Identifying patterns and processes of the structure-function relationship in the urban landscape in the context of environmental and ecological processes is challenging due to variable density and patchy spatial patterns (Pickett & Cadenasso, 2008).
While it is well established that urban areas host ecological communities subject to unique stressors (Jones & Harrison, 2004; Joyner et al., 2019; Reese et al., 2016) absent in natural systems (e.g pollution, high population density), the influence of landscape heterogeneity on the environment is poorly described. Functional classification of urban structure is necessary for understanding the nature of social and ecological relationships in urban areas (Cadenasso et al., 2007; McPhearson et al., 2016; Zhou et al., 2014). Over the last decade, fine-scale landcover classification for urban areas have been developed (MacFaden et al., 2012; Pickard et al., 2015) that allows more nuanced analyses of urban landcover. While some functional classification approaches have been suggested (see for example Cadenasso et al., 2007), major challenges remain in integration of spatial structure and configuration that allows scalable and reproducible analysis of relationships between urban form and process.
A major barrier for understanding the relationship between urban structure and environmental function is the lack of independent measurement of the fine-scale spatial variability of the distribution of environmental and ecological variables. Particularly important is the vertical dimension (e.g. building height) and variation of the three-dimensional landscape that is rarely addressed (Alavipanah et al., 2017) in ecological studies. Where independent measurements exist, such as data from Environmental Protection Agency (EPA) air pollution monitoring stations or United States Geographical Survey (USGS) water monitoring sites, the spatial distribution is not sufficient to allow intra-urban analysis. Surface temperature is one example of a physical property of the urban environment that has been used in research addressing landcover (Zhou et al., 2011), urban heat islands (Rosenzweig et al., 2009; Zhao et al., 2011), and ecosystem services (Schwarz et al., 2011). Likewise, ST structures patterns of taxonomic and functional biodiversity (Scherrer & Körner, 2011; Zogg et al., 1997), hydrology (Reyes et al., 2018), air quality (Li et al., 2018; Sillman & Samson, 1995), and social variables relevant for studies of environmental injustice (Huang & Cadenasso, 2016; Zhang et al., 2017). We employ ST as a proxy for a wide range of potential variables of interest across biotic and abiotic dimensions.
To account for the heterogenous vertical dimension of the built enviroment in urban lanscscape, we employ The STructure of URban Landscapes (STURLA) classification (Hamstead et al., 2016). STURLA is a new classification procedure developed by co-PI Kremer that incorporates the complexity of urban land cover structures, including the vertical dimensions of the built environment. The novelty in the STURLA approach is that it offers a composite functional classification of urban structure, including the vertical dimension, that is automated, and thus can be applied to wide geographic regions systematically. STURLA has has been used to identified patterns of microbial biogeography in the atmosphere of Philadelphia (J. Stewart et al., 2020), and ST in NYC (Hamstead et al., 2016) and Berlin (Kremer et al., 2018).
The objectives of this short study are to identify if STURLA could explain the variation of urban structure in a new model city (Philadelphia), and quantify this variation using a physical property of the environment (ST). Results suggest STURLA identifies common urban structure units that encompass the majority of the variation in the urban landscape strucutre. Moreover, when correlated to surface temperature, these common urban structure classifications exhibit distinct temperature signatures for different urban structure units with temperature trends dramatically similar between Berlin and NYC. Here, we contribute to the developing literature on the urban structure-function relationship using STURLA in Philadelphia.