*3.1. Obtaining General Information and Factors to Assess the Park Quality*

The quality of urban parks is very important with respect to spatial and environmental justice [34]. The quality-of-life community factors for parks used in this work were selected from the Madrid case (Spain) [35] and Bucaramanga (Colombia) [36]. The model included a total of 20 factors. Some of them were obtained through cartographic analysis using GIS (ArcGIS 10.6.1), others were obtained from fieldwork and, most of them applying both methodologies (checking cartographic results with fieldwork) (Table 1).

**Table 1.** Factor name, description and methodology used to obtain them. Selection based on Canosa, Sáez, Sanabria and Zavala (2003) and Rivera (2015).


Source: own work.

In each urban park in Tarragona, we generated a spatial and theme-based database with information on the urban location, the surface area covered by vegetation, the covered green shadow and the various facilities (Figure 3). This information was digitalized based on the Topographic Map 1:5000 and orthophotography on the same scale; both documents were provided by the Cartographic and Geological Institute of Catalonia (Institut Cartogràfic i Geològic de Catalunya).

**Figure 3.** Example of the mapping database of the Fitolaca park (Tarragona). Source: own work.

Field work was also carried out, through the visits to the parks included in the study, between the spring and summer of 2018. The purpose of these visits was to obtain the direct information needed to establish the PQI (see Table 1), check the digitalized mapping and take photographs to characterize these spaces.

Each of these 20 factors was assessed on a scale of 0 to 3, where, in a standardized way, the value of 0 corresponds to the lowest quality of the factor, and the value of 3 refers to the highest quality of the factor.

## *3.2. Creating Demographic and Socio-Economic Population Indicators*

Data from the Municipal Register of Inhabitants were used to characterize the population of the city of Tarragona on 31 December 2019. This database stores the residents' postal address and, therefore, it was possible to geolocate the registers based on this address and build up a mapping layer. The geolocation was carried out from the Instamaps platform of the Cartographic and Geological Institute of Catalonia (https://www.instamaps.cat/#/. Accessed from 1 December 2020 to 8 January 2021). The resulting layer was imported to ArcGIS and, using the tool "Near", inhabitants living within 300 m of the nearest park access point were selected and assigned to the nearest park, considering the mode of transport to be walking, because it is healthy and not affected by economic conditioning [37].

Some authors [38] use two factors to characterize the population demographics: the level of study and nationality. The level of study collected in the register of inhabitants in Tarragona refers only to the population aged 16 years old and over, and has been grouped into five categories (illiterate, no schooling completed, primary education, secondary education, university education). To compare the different territorial units, a Synthetic Training Index (STI) was created based on the introduction of weighting for the population at each training level. The general formula of the STI is as follows:


In this way, an index is obtained with values between 0 and 1, where 0 would be equivalent to the entire illiterate population and 1 would represent the opposite extreme, with the entire population having a university education. In order to neutralize the influence of the population's aged-based structure in terms of education (an older population tends to have lower education levels than a younger population), a direct standardization was carried out, based on the application of a typical population structure (the whole of the population of the city of Tarragona) to the 14 neighborhoods of the parks under analysis.

The second variable chosen is the origin of the population. In this case, we opted to use the population's place of birth, as opposed to nationality, because this addresses the idea of people from immigrant families who were born in Spain. To compare the different territorial units, the population born abroad was characterized using the average HDI value published by the United Nations Population Division in 2020. In addition, in order to better reflect the diversity of the population born in Spain, the HDI of the autonomous community of birth was taken into consideration. This information was obtained from the Valencian Institute of Economic Research (Instituto Valenciano de Investigaciones Económicas, year 2019).

In third place, due to the lack of disaggregated data at the income level, the population's economic characterization was indirectly analyzed based on housing prices. This information was taken from the property portal, Idealista.com, which allows you to consult the average renting and purchasing prices per square meter for apartments in a specific digitalized area. Thanks to this option, it was possible to define the 300 m area of influence around each park. The information obtained in this way is comparable with that from the Register of Inhabitants.

#### *3.3. Creating the Park Quality Index (PQI)*

The Multi-Criteria Evaluation (CME) encompasses a set of tools aiming to help decision-making [39], in which the various alternatives determined by multiple criteria and objectives are in conflict [40]. This work adopts the multi-criteria evaluation model in order to discover the degree of quality or suitability of urban parks, based on the selection of a series of indicators, subindicators and factors (Figure 4). To do this, the initial 20 factors (first hierarchical level) were grouped into seven subindicators (second hierarchical level) and, in turn, these were joined together in three indicators that correspond to (1) the quality of the vegetation, (2) the quality of the facilities and (3) the quality of the street furnishings (level 3). Finally, the combination of the three indicators leads to the Quality-of-Life Community indicator for parks (PQI) (level 4).

One of the essential characteristics of an MCE is the importance or weights according to the percentage of each factor, subindicator and indicator used in the model. The final result will largely depend on the weight that is assigned to each part of the model. In this case, the weight assignment is related to the established hierarchies and groups, so that they each add up to 100%. If we take the third hierarchical level as an example, and apply the decision formula or rule, vegetation is combined with a 40% weighting and facilities are combined with a 30% weighting, while the weighting for property is 30%. In order to perform the different aggregations of the model, we used the Weighted Overlay command in ArcGIS 10.6.1.

#### *3.4. Statistical Analysis and Environmental Justice*

Finally, once the IQP for each park was calculated, it was correlated with the demographic and socioeconomic characteristics of the population assigned to each park

(populations at a distance < 300 m). To measure the fit of the variables, the Pearson correlation index (R2) was calculated using MS Excel software.

**Figure 4.** MCE of the quality of urban parks. Source: own work.
