*3.5. Site Selection*

Among the 67 reviewed articles, 20 articles included the criterion of site selection. The location of GSI is often regarded as a significant factor affecting the effectiveness of planning [67]. Therefore, identifying high-priority construction areas for various GSI types is always a research hotspot. The appropriate sites contribute to the reduction in the vulnerability of the study area (e.g., floods, climate change), and the acceleration of the production of a wider range of ES [68]. Taylor et al. [69] integrated GIS with e-tools, and identified the potential GSI areas based on the determination of the existing GSI, and the principles of its site selection were as follows: vegetation height < 1.5 m; 10 m buffer zone for cemeteries, playgrounds, and railways; exclusion of impervious surface areas, golf courses, historical sites, water bodies, and wetlands; polygons ≥ 9.29 m2. Martin-Mikle et al. [70] identified hydrologically sensitive areas by extracting land-use types to calculate the topographic index and selected 140 priority GSI sites after the identification of land use, spatial scale, and the applicability of constructing GSI in impervious areas. Li et al. [71] evaluated priority sites for GSI to mitigate floods in Ghent, Belgium, through runoff coefficient, socially sensitive groups, road sensitivity, building sensitivity, and

environmental justice. Langemeyer et al. [72] discussed six types of ESs, i.e., heat regulation, runoff control, habitat, food production, entertainment and leisure, and social cohesion through multi-criteria decision analysis (MCDA) to select priority areas for green roofs in Barcelona, while Song et al. [73] selected eight criteria in three dimensions—social, hydrologic, and physical–geometric—to construct the MCDA framework to evaluate the performance of infiltration trench and permeable pavement in eight sub-catchments in Seoul, South Korea, then ascertained the best location.

#### **4. Discussion**

#### *4.1. Facility Aspect*

#### 4.1.1. Objective Formulation

We affirm the significance of objective formulation, as it affects all subsequent planning steps. If the objective is always described qualitatively, not only will the interest of stakeholders and investors become lower and lower, it will also lead to loopholes in all subsequent planning steps (i.e., type/scenario evaluation, quantity/scale determination, and site selection) and considerable hidden dangers. A feasible solution is according to the ES (i.e., water quantity regulation and water quality regulation services in this review) that GSI can provide, and building an integrated framework of GSI and ES to fully identify the functions of GSI, so as to accomplish the multifunctionality of GSI to obtain maximum benefits. However, the ES concept is rarely used explicitly in planning objectives, which may be caused by the fact that it is not clear how ES provides guidance for decision-making information and whether ES concepts should be introduced in the objective formulation step [74,75], while others argue that the ES concept is not clear enough among planning practitioners and has not reached a broad understanding [76,77]. In fact, urban stormwater management planning is one of the areas that strongly facilitates the integration of ES knowledge [66], and GSI planning affects ES in multiple ways at different decision-making levels [78]. Future GSI planning needs to reduce the complexity of the evaluation process to attract more stakeholders' attention to understand the ES concept. ES assessment has been increasingly conducted as an imperative source of knowledge to support decision making [79]. Meanwhile, incorporating ES assessment results into decision-making processes usually means a significant increase in the amount of information that needs to be considered [80]. In complex decision-making problems, proper knowledge synthesis is a basic step to reduce the burden of information and support evidence-based decision making. Therefore, how to effectively integrate multiple ES assessments is a problem that needs to be solved in the objective formulation step in future GSI planning [66], which means the trade-off among ES should be taken into account, e.g., increase in aquifer storage and groundwater pollution, water purification, and water flow temperature management [20].

In addition, similar or different GSI facility types in different studies perform different functions, which impedes a quantitative objective formulation, as shown in Tables 1 and 2. A viable method is that the authorities summarize various water quantity and water quality regulation capacities of different GSI facility types based on as many existing studies as possible, taking the spatial heterogeneity into account, and then formulate reference values, with reasonable ranges that are based on social, economic, and environmental conditions of specific study areas, which may be more suitable than fixed values, as the latter may affect the rationality of planning objectives. Xu et al. [81] set the water quantity regulation objective according to the Urban Flood Control Engineering Disciplines, China (GB/T50805-2012), while the water quality regulation objective was set based on the water quality volume criterion of BMPs developed by the US Environmental Protection Agency (USEPA); nevertheless, the planning area was Shanghai, China. This method of randomly setting objectives with reference to different standards is obviously flawed. Reference standards should consider regional differences, and provide the best reference range for water quantity and quality regulation objectives such as pollutant reduction rate and total runoff control rate; for example, Zheng et al. [44] integrated 75 GR studies and quantified the average runoff retention rate that reached 62.2%. There are already authoritative references—the USEPA has set an objective for retaining the 95th percentile rainfall event [82]; China's Technical Guide for Sponge City Construction (Trial) [83] proposed 70–90% annual runoff control rate targets for different regions, which sets different runoff control objectives for different subareas and is in line with real-world scenarios to guide the objective formulation in planning areas with different conditions. In line with Roggero [84], highlighting the benefits of policy instruments for GSI planning, these policies have leading roles in planning. However, as shown in the literature we reviewed, there is no clear quantitative objective reference for water quality control or other aspects of water quantity control, and studies only qualitatively proposed factors such as pollutant removal, runoff peak reduction, and time-to-peak delay, without accurate values. Moreover, as relevant studies are increasingly inclined to investigate comprehensive GSI scenarios with multiple facility type combinations, the synergy of facilities in the future should also be taken into consideration.
