**1. Introduction**

Today, more than half the world's population is living in cities [1]. The continuous growth in urban populations combined with a more extreme urban climate due to global warming are having a detrimental impact on urban ecosystems [2]. In order to maintain the quality of life for the burgeoning numbers of urban residents, it is becoming increasingly important to protect and promote urban ecosystems and their services [3–6]. Concepts such as green and blue infrastructure have been developed in recent years to help tackle the environmental challenges of cities. The strategic planning of urban green structures improves the well-being of inhabitants while simultaneously boosting the resilience of cities to climatic changes [7]. Yet, such strategic planning requires comprehensive insights and information on the multiple functions and services of green infrastructure on different spatial scales. In particular, knowledge and expertise are needed on ecosystem services (ES) on the small spatial scales where planned measures are realized [8–12]. More research into urban ES on small spatial scales will improve our understanding of this planning factor, thereby aiding the integration of the urban ES concept into urban planning as an important factor for sustainable urban development. The correct application of this concept has the potential to better exploit the multiple benefits of

urban ecosystems, so that urban planning can be more closely oriented to natural conditions and resources [13–15]. Furthermore, the development of standards and indicators to assess and describe ES in urban contexts can help politicians, urban planners as well as practitioners create ecological and sustainable cities [16]. While we can already point to a few practical examples of the successful integration of ES-related subjects into diverse planning documents and tools, there are still several unresolved problems limiting a more general implementation [17].

One limiting factor is the lack of information on urban ecosystems on di fferent spatial scales and their services. The poor quality of available spatial or other relevant data on small, local scales often complicates the integration of ES into planning frameworks or decision-making processes [13–20]. A further limitation is the lack of suitable methods to assess ES at such spatial scales; hitherto, most assessment methods have referred to the global, national or regional scale. Clearly, if we wish to promote the inclusion of ES in decision making at the urban level, it is necessary to improve our knowledge of this subject at city-wide but most importantly also on local, and thus site-, scales [9,19]. Over the last few decades, urban ES has become a widely investigated topic in di fferent research fields, with scholars recognizing its importance in mitigating climatic extremes and contributing in diverse ways to human well-being [19,21]. In this study, urban ecosystems are defined as areas largely dominated by the built environment and which comprise gray and green infrastructures [4,22,23]. Of course, urban ecosystems only provide a fraction of the ES used by city dwellers – the larger part of these services are provided by widely distributed ecosystems in the city surroundings. Yet in relation to the size of urban ecosystems, they benefit a large number of citizens [24]. Thus, urban ES have a high anthropogenic impact, representing an explicit type of ES that needs to be considered more closely.

Locally provided ES generally play an important role in promoting the quality of life of urban residents. Yet, the issue of the ES of urban small-scale structures is an underrepresented research field [14]. The small spaces within cities are designed by urban planners in grea<sup>t</sup> detail, and it is exactly this spatial scale and structure that is directly perceived by residents and thus strongly influences the quality of life in the city. Previous studies on the assessment of urban ES have stressed the importance of the spatial scale of investigations [14,25,26]. Hitherto, many assessments have been conducted on larger spatial scales (city, region, nation-wide) with results often presented in a generalized way. To obtain more realistic results, it is necessary to conduct empirical ES studies of smaller urban structures. In order to ensure the practical implementation of the ES concept, we have to focus on spatial structures and scales that are recognized by existing planning tools, e.g., neighborhoods, small single parks, etc. [26]. Furthermore, previous reviews have revealed a large number of di fferent methods used to assess urban ES [14,26,27]. Most of these involve spatial proxy methods, for example, utilizing land use and land cover data to estimate ES supply capacities. Primary data is rarely collected in urban ES assessments [27]. Another approach to the assessment of ES is to consider the complexity of urban structures [14,26]. In this case, it is important to take account not only of built structures but also urban open spaces, for example, the various types of green open space [14,26]. In particular, Haase et al. [14] found that most previous studies assessed regulative ES in cities, with only a few looking at cultural and provisioning ES.

These aspects determine the scope of this review and shape the key questions. We aim to review the current state of knowledge on methods to assess the urban ES of di fferent types of green infrastructures from city to site scales. To this end, we have only considered studies that examine individual spatial structures or forms of land use in cities such as parks, gardens or trees. The review will answer the following three questions: (1) Which urban ES are assessed in relation to green infrastructure types? (2) Which specific spatial structures are the subjects of investigation? (3) Which methods are used to assess ES on larger (city) and smaller (site) spatial scales? Furthermore, we will look at the motivations of studies in assessing the urban ES of di fferent types of green infrastructure types, as well as check which data type (i.e., primary or secondary) has been used by the reviewed studies.

#### **2. Materials and Methods**

#### *2.1. Review Approach*

The first step was to carry out a systematic quantitative literature review after Pickering and Byrne [28]. In comparison to classical meta-analytical reviews, the methodology after Pickering and Byrne [28] aims to determine general aspects of studies (e.g., numbers, types, and geographical aspects), research trends and gaps as well as methodological patterns. To this end, the literature databases "Web of Science" and "Scopus" were searched for relevant peer-reviewed articles published in international scientific journals. This search was conducted from March to April 2019. Several filter criteria were applied to specify the review but still to identify as many relevant articles as possible:


Systematic searches were conducted of the database "Web of Science" for each possible urban green infrastructure type (see Table 1), adding the search terms "urban ecosystem services" and "assessment" or "valuation" (for example, "urban ecosystem services" AND assessment AND park). After completing individual searches for various spatial scales (see examples in the "Scopus" search terms), all results were cross-checked to exclude repeated articles. The search procedure of the database "Web of Science" identified a total of 35 studies.


**Table 1.** Range and definitions of investigated spatial objects.

In the database "Scopus", a general search was conducted for all possible green infrastructure types using the following keyword combination:

TS = (neighborhood\* OR district\* OR estate\* OR meadow\* OR brownfield\* OR allotment\* OR "community garden\*" OR park\* OR woodland\* OR "green space\*" OR "green infrastructure" OR residential OR cemeter\* OR wetland\* OR "urban tree\*" OR "urban forest\*" OR lake\* OR waterbod\* OR river\* OR stream\*) AND ("urban ecosystem service\*" AND assessment OR valuation).

This search identified a total of 31 papers. After cross-checking and combining the results from both databases, two articles were excluded, resulting in a selection of 29 articles. Further articles could be added to this list by screening the bibliographies. This procedure led to a final total of 63 scientific articles.

The second step of the review approach was to search the relevant gray literature, such as reports and documents that were compiled by organizations and institutions that do not belong to the "traditional" academic instances (e.g., governmen<sup>t</sup> departments, non-governmental organizations or civil society). For this purpose, an internet search was conducted to identify some initial potential international funding bodies of projects on urban ecosystems across Europe. Their webpages were then screened for relevant projects, after which the project webpages were studied. This search procedure led to a snowball effect, resulting in the identification of 13 relevant project documents, of which six were additional scientific articles drawn from the bibliographies of the project documents.

From these two review steps, we were able to identify a total of 76 articles. An overview of the assessed studies is presented in Supplementary Materials (S1). Each publication was analyzed and added in an Excel databank, where specific information was extracted and combined in one table.

#### *2.2. Analysis Approach of Included Articles*

The authors of the various articles made use of a range of different terms and expressions for ES. For the purposes of our study, it was first necessary to consolidate these terms to allow us to summarize and compare the investigated ES (cf. key question 1). For this reason, all identified ES from the articles were classified into the corresponding ES sections, groups and classes of the latest version of the Common International Classification of Ecosystem Services (CICES V 5.1).

To provide a comprehensive overview of the investigated differently scaled green infrastructure types (cf. key question 2), the spatial objects in the papers were first assigned to one of two dimensions, i.e,. city- or site-dimension (see Figure 1). This classification was intended to reflect the scope of the investigated objects in each study. Thus, whenever an assessment method was applied to several spatially distributed (yet urban) objects, the dimension of this study was classified as "city" (e.g., ES assessment of various urban parks in a city). Alternatively, if only one object was assessed, e.g., a city park, the dimension was defined as "site". In this case, only situation and location-specific conditions can be said to apply.

**Figure 1.** Schematic illustration of the assignment of studies to city- or site-dimension.

Following this initial classification into city- or site-dimension, the urban green infrastructure types in each dimension were summarized and classified into more precise types, e.g., park, garden, forest, etc. An overview of this classificatory system is given in Table 1.

Furthermore, we identified seven di fferent categories for the method classification of all reviewed articles (cf. key question 3). These categories are as follows: "spatial proxy methods", "samplings/field mapping and observations", "surveys and questionnaires", "economic valuation methods", "model-based methods", "social media-based methods" and "remote sensing and earth observations" (see Supplementary Materials (S2) for a detailed description of the categories).

On the basis of the described classifications, the extracted information from the 76 reviewed articles was then evaluated to answer the key questions.
