2.1.1 Understanding impacts of urbanization on spatial variation of land surface phenology requires fine-spatial resolution remote sensing
Understanding the effect of urbanization on land surface phenology (LSP) is a critical step to study the broader influences of urbanization on the environment. Urban vegetation provides crucial ecosystem services, such as reducing noise, absorbing pollutants, serving as habitats for some migratory and local birds. Previous studies confirmed that urban areas experience higher temperature than the surrounding rural regions (Arnfield 2003; Oke 1973, 1982). This
phenomenon is known as the urban heat island (UHI) effect. An accurate knowledge of the impacts of UHI on LSP can help mitigate the vulnerability of urban ecosystem services. For example, quantifying the effects of UHI on LSP can reveal the potential phenological
mismatches between vegetation, insects and birds at higher trophic levels (Miller-Rushing et al. 2010; Thackeray et al. 2016), thus providing clues for biodiversity protection in the urban
1This chapter previously appeared as two articles in the Remote Sensing and Remote Sensing of Environment. The
original citation is as follows:
Qiu, Tong, et al. "Urbanization and climate change jointly shift land surface phenology in the northern mid-latitude large cities." Remote Sensing of Environment 236 (2020): 111477.
Qiu, Tong, Conghe Song, and Junxiang Li. "Impacts of urbanization on vegetation phenology over the past three decades in Shanghai, China." Remote Sensing 9.9 (2017): 970.
ecosystem. Moreover, LSP controls the timing of pollen production, and thus the allergy season in urban areas (Neil and Wu 2006). Understanding the urbanization-induced phenological changes can provide valuable information for public health risk forecasting (Cecchi et al. 2013; Neil and Wu 2006).
Given the significant progress in detecting phenological changes of the natural ecosystems that are generally controlled by temperature (Schwartz et al. 2006; Zhang et al. 2004a; Zhang et al. 2007) and precipitation (Guan et al. 2014; Zhang et al. 2005), it remains less clear how the process of urbanization has altered LSP in the heterogeneous urban environment. Manipulative experiments and ground observations have documented earlier starts of growing seasons (SOS) and later ends of growing seasons (EOS) in the urban center than the surrounding rural areas (Jochner et al. 2012; Mimet et al. 2009). While those studies provide important evidences of effects of urbanization on LSP, site-based observations cannot provide an assemble
understanding of spatially-explicit phenological changes in urban areas due to the lack of
standard data collection protocols and consistent data analysis methods (Gill et al. 2015). Remote sensing observations offer consistent quantitative measurements of land surface properties, making long-term satellite observations ideal resources for monitoring vegetation phenology (de Beurs and Henebry 2004). Many algorithms have been developed to estimate phenological metrics based on time series of vegetation indices derived from Advanced Very High Resolution Radiometer (AVHRR) and Moderate Resolution Imaging Spectroradiometer (MODIS) (Cong et al. 2012; Dannenberg et al. 2015; Hwang et al. 2011; Jonsson and Eklundh 2004; White et al. 2009; Zhang et al. 2003). More specifically, studies have reported an increase of 7.6 days in the length of growing season (LOS) caused by urbanization in the Eastern United States (White et al. 2002). Zhang et al. (2004b) found an increase of LOS by 15 days around urban centers, and the
lengthening of LOS extends up to 10 km beyond urban margin. Zhou et al. (2016) found SOS were 11.9 days earlier and EOS were 5.4 days later around urban centers than their surrounding rural areas in China’s 32 cities. However, our understanding of urban phenology with the coarse spatial resolution images in urban environments is limited due to the complexity of the urban environment. The localized heterogeneity in urban phenology changes as a result of spatial variations in urban land-cover/land-use (LCLU) composition and configuration cannot be revealed using coarse spatial resolution images. The opening of the Landsat archive has enabled the pixel-wise long-term time series analyses at finer spatial resolution (Woodcock et al. 2008). Fisher et al. (2006) demonstrated that the average phenology of New England deciduous forests could be mapped at the Landsat scale using multitemporal Landsat observations that were organized by day of year (DOY). Melaas et al. (2013) and Melaas et al. (2016b) extended the algorithm in a way that allowed the detection of interannual variability in phenology and validated the method in North American temperate and boreal deciduous forest. These approaches have only recently been applied to urban areas (Melaas et al. 2016c; Zipper et al. 2016), and there remain substantially unrealized potential for leveraging them to better
understand how urbanization affects phenological changes. More importantly, landscape patterns not only reflect the urban development and their socioeconomic drivers (Jenerette and Wu 2001; Li et al. 2013; Seto and Fragkias 2005), but also significantly influence UHI (Li et al. 2011). However, the relationship between landscape pattern and vegetation phenology is poorly understood. Therefore, for the first part of chapter 2, we aim to investigate the impacts of urbanization, as well as the urban landscape composition and configuration on vegetation phenology.
2.1.2 Joint effects of climate change and urbanization on temporal shift of land surface phenology remain less studied
The impacts of global climate change on LSP in the natural ecosystem have been well documented in the literature. A general consensus is that global climate warming has caused an advanced SOS in most regions of Northern Hemisphere based on evidence from AVHRR vegetation indices dataset (Cong et al. 2013; Jeong et al. 2011; Myneni et al. 1997; Piao et al. 2006; White et al. 2009; Zhang et al. 2004a; Zhang et al. 2007). This finding has been further confirmed by more recent studies using Moderate Resolution Imaging Spectroradiometer
(MODIS) (Keenan et al. 2014), Satellite Pour I’Observation de la Terre Vegetation (SPOT-VGT) (Cong et al. 2012), and Landsat data (Melaas et al. 2018). In contrast, our understandings on the interannual variation of EOS and its environmental drivers remain relatively weak compared to those of SOS (Gallinat et al. 2015; Richardson et al. 2013). Researchers found an overall positive trend of EOS in response to the global climate warming in mid- and high- latitude of Northern Hemisphere (Garonna et al. 2014; Liu et al. 2016a; Liu et al. 2016b). The factors that most likely impact the shifts of EOS include interannual variabilities of temperature,
precipitation, insolation, and extreme weather events (Liu et al. 2016a; Xie et al. 2015a). Although climate effects on LSP in the natural ecosystems have been extensively explored, the relative contributions of urbanization and climate in influencing temporal variations of LSP in urban ecosystems remain poorly studied. Previous studies mainly focused on understanding the differences of spatial patterns in LSP between urban cores and their rural counterparts. For example, studies using satellite instruments found that SOS advanced by 1-14 days and EOS delayed by 0-20 days in urban centers compared to the corresponding rural surrounding regions in North America and China (Li et al. 2016; Li et al. 2017; Melaas et al. 2016c; Qiu et al. 2017;
White et al. 2002; Zhang et al. 2004a; Zhang et al. 2004b; Zhou et al. 2016; Zipper et al. 2016). However, each of those studies used satellite dataset with relatively short temporal coverage (less than 10 years) or calculated average values from multiple-year phenological metrics or had study period of only one year. The temporal responses of LSP to the combined effects of climate variations and urbanization-induced land cover conversions (i.e., increase of impervious surface areas) thus remains unknown. Zhao et al. (2016) reported that effects of urbanization on
vegetation growth, defined as remotely sensed vegetation greenness, could be divided into direct effects (i.e. replacing naturally vegetated areas with impervious surface areas or different types of vegetation, e.g. lawns) and indirect effects (i.e. altering environmental conditions for plant growth through surface UHI, CO2, nitrogen deposition). They concluded that indirect effects could compensate approximately 40% of the direct effects. They also found that vegetation growth in cities of China was enhanced by urbanization. A recent study in the conterminous United States provided more evidences that supported those conclusions (Jia et al. 2018).
However, the relationship between urban land cover conversions and temporal changes of LSP is still unclear. In addition, most studies attempted to explain the spatial variations of LSP in cities through urbanization-related factors including urban size (Li et al. 2016), percentage of
impervious surface cover (Walker et al. 2015), and urban heat islands (Zhang et al. 2004b). However, these studies did not explicitly incorporate climatic drivers (i.e. temperature, precipitation, and insolation), which influence LSP through time as a result of global climate change. Filling those knowledge gaps advances our understanding of how urbanization and global climate warming jointly impact the urban ecosystem dynamics, and helps policy makers to identify effective mitigation and adaptation strategies to enhance urban ecosystem resilience to future environmental changes. Therefore, for the second part of chapter 2, we aim to investigate
the joint effects of urbanization and climate change on the temporal changes of land surface phenology.