Green Infrastructure for Reintegrating Fragmented Urban Fabrics: Multiscale Methodology Using Space Syntax and Hydrologic Modeling

Abstract

Green infrastructure (GI) plays a critical role in addressing urban fragmentation and flood vulnerability, especially in rapidly expanding cities where its optimal placement is essential to maximize social, ecological, and economic benefits. This study presents a multiscale methodology integrating spatial configuration and hydrological modeling to guide GI implementation in Ciudad Juárez, Mexico. The approach applies space syntax theory, fuzzy logic, and geospatial analysis across three spatial levels. At the city scale, the method evaluates street network integration and service accessibility to identify urban centers with potential for regeneration through GI. At the local scale, a 214-hectare area is analyzed using fuzzy multi-criteria decision analysis and Multiscale Geographically Weighted Regression (MGWR) to select the optimal locations for different nature-based solutions. At the microscale, spatiotemporal hydrological simulations of a 25-year return period rainfall event quantify the runoff and infiltration dynamics under different GI configurations, achieving infrastructure layouts that infiltrated over 1000 m³ of stormwater. This framework addresses the research gap on how connectivity and morphology can be combined to prioritize interventions based on flood risk data. The results offer a transferable strategy for integrating Sustainable Urban Drainage Systems (SUDSs) into complex data-scarce urban environments, supporting long-term urban resilience and multifunctional land-use planning.

1. Introduction

Rapid urbanization and climate change are converging to pose unprecedented challenges for cities worldwide. While many cities face extreme fluctuations in their rainfall patterns and hydrological risks, the spatial organization of urban systems often increases these impacts. In response, nature-based solutions such as green infrastructure (GI) have emerged as critical strategies not only for flood mitigation but also for improving connectivity, public space quality, and social cohesion.

In particular, Ciudad Juárez, Chihuahua, Mexico, has experienced significant economic growth driven mainly by industrial activities due to its strategic location near the United States, creating unique cross-border urban dynamics (Figure 1). While rapid urbanization has economically benefited this transnational region, it also brings challenges associated with climate change. Its fragmented spatial structure and socioeconomic disparities make it a representative case for testing spatial models that link mobility, water risks, and urban morphology in rapidly urbanizing border cities.

Like other industrial cities, Ciudad Juárez struggles with morphological issues resulting from rapid and uncontrolled urban sprawl. Its fragmented urban structure and the hydrological challenges common in semi-arid regions of the Chihuahuan desert make it an ideal case to evaluate multi-scalar GI planning. Urban development without proper hydrological planning [1,2] intensifies the disruptive effects of climate change, endangering vital water resources and negatively impacting residents’ quality of life. Addressing these issues requires integrating Sustainable Urban Drainage Systems (SUDSs) into urban design while considering hydrometeorological risks.

Urban planning that prioritizes interaction with nature and interconnected spaces fosters stronger community ties and helps reduce social fragmentation. Enhancing green spaces and creating streets or public areas that encourage interaction with nature through green corridors can significantly improve residents’ physical and mental well-being [3].

Historically, the value of maintaining strong connections with nature has often been underestimated. Current urban planning practices typically prioritize short-term, low-cost solutions. However, long-term infrastructure planning incorporating sustainable solutions is essential to address the effects of climate change effectively [4]. Although initially more expensive, gradually replacing the nearly nonexistent traditional stormwater infrastructure with GI is a practical approach. Long-term urban development strategies using GI to mitigate flooding and promote multifunctional use in public spaces can significantly enhance urban integration.

Ciudad Juárez faces additional challenges due to industrial growth that has resulted in many underutilized urban areas. This research aims to revitalize these Fragmented Urban Fabrics (FUFs) by integrating new green infrastructure and creating multifunctional networks that connect urban centers. Such integration helps combat urban fragmentation, positively influencing the environment, society, and local economy [5].

Comparable spatial strategies have been applied in cities such as México City, Bogotá, Buenos Aires, São Paulo, Santiago de Chile and Lima, where nature-based infrastructure has supported mobility, climate resilience, and social cohesion [6]. However, few studies have linked spatial configuration to hydrological modeling in a multiscale framework, despite similar urban challenges being consistent across other border cities in Mexico like Tijuana, Mexicali, and Reynosa.

While previous studies have explored the role of green infrastructure in stormwater management or pedestrian accessibility independently, few have proposed multiscale models that combine urban morphology and spatiotemporal hydrological performance in the equation. This research addresses that gap by building on existing research focused on optimizing green infrastructure placement considering multiple goals, including urban runoff management [7,8], improving water quality [9], reducing urban heat islands [10], and enhancing urban amenities. Using multi-criteria analysis [11] and geographic information systems (GISs) to prioritize areas for rehabilitation [12], based on spatiotemporal hydrological data [13], land use, topography, and other relevant physical characteristics [14].

City-scale analysis is essential for effective decision-making at smaller scales. However, many studies fail to identify critical urban areas suitable for intervention due to analyses conducted primarily at local scales or microscales. This study emphasizes the synergy achievable through integrating FUF networks based on urban morphology using “Space Syntax” theories [15] and the concept of “Spatial Capital” [16].

Consequently, this research proposes a comprehensive city-scale methodology to prioritize and strategize the rehabilitation of the degraded neighborhoods forming the FUFs, reconnecting them through ecological corridors. This strategy emphasizes the interrelation between urban form and functionality. Identifying strategic locations for interconnected green spaces can fundamentally revitalize these urban areas, enhancing spatial distribution, pedestrian accessibility, and overall urban resilience [17]. Using a multi-scalar methodology overcomes limitations in previous work, where studies often focused only on isolated interventions rather than understanding interactions between city, neighborhood, and site-specific levels.

Additionally, this study conducts detailed local-scale analyses, evaluating various GI options for specific intervention areas using multi-criteria GIS analysis and fuzzy logic to determine feasibility based on morphological and physical constraints [18]. At the microscale level, it evaluates specific sites, analyzing storm behavior and designing customized, nature-based solutions tailored to local conditions [19].

This approach has been widely validated by previous research, demonstrating that optimally placed GI networks using space syntax [20] can effectively address critical objectives for sustainable cities, including environmental equity and justice [21], and strengthen urban resilience by establishing interconnected green spaces to reduce flooding risks [22]. Thus, the methodology adopted here aligns with contemporary best practices and simultaneously addresses multiple Sustainable Development Goals (SDGs), particularly SDG 11 (Sustainable Cities and Communities), SDG 13 (Climate Action), and SDG 6 (Clean Water and Sanitation). By integrating spatial planning with stormwater management, the methodology offers actionable tools for land managers, urban designers, and policy-makers addressing resilience in vulnerable communities.

2. Materials and Methods

This study adopts a multiscale spatial–hydrological methodology to guide the strategic placement of green infrastructure (GI) in fragmented urban fabrics. The approach is designed to bridge the gap between spatial connectivity, environmental performance, and policy scalability. The analysis is structured across three scales: the city scale, the local scale, and the microscale (

Table 1. Multiscale methodology for green infrastructure implementation summarizing the primary activities, required information, and expected outcomes for each phase.

Scale	Process		Main Activity	Required Information	Results
City	1	Identify strategic implementation points in urban centers	Performing spatial analysis: urban accessibility models with Space Syntax and Place Syntax	Road Integration	Spatial Capital analysis for the implementation of new Green Infrastructure
				Population accessibility
				Economic Diversification
				Social Diversity
				Park Accessibility
				Urban environment
Local	2	Describe local objectives, planning and strategy	Conduct Research: Local regulations and projects	Urban Development Master Plans: Historic Center of Ciudad Juárez Recovery of the Asequia Madre	Project objectives and limitations
	3	Identify areas where the implementation of green infrastructure is feasible and its possible restrictions	Analyze public spaces according to the available information: physical restrictions of the type of solution	Land use	Areas with physical conditions necessary for the use of Green Infrastructure Selection Matrix
				Slope
				Water table level
				Soil infiltration
				Building typology
				Existing infrastructure
				Flood depth and runoff velocity
				Micro-basins critical points
	4	Multi-criteria structure for strategic actions	Multi-criteria decision analysis using fuzzy logic: Most reliable option according to physical, social and morphological conditions	Flood susceptibility	Recommended area of implementation: Suitability Level
				Modified Spatial Capital
				Optimal connection Choice analysis
				Public transport routes
				Percentage of different land use analysis
				Multiscale Geographically Weighted Regression (MGWR)
Micro	5	Selection of green infrastructure for proposed areas.	Selection of green infrastructure for proposed areas.	Suitability Level Model	Recommendation of actions to take
				Spatiotemporal Flood Simulations
				Space Time Cube (STC)
				Emerging Hot Spot Analysis (EHSA)
				Time Series Clustering (TSC)

). The methodology enables the identification of GI opportunities from urban system planning to site-specific hydrological performance, selecting systems in harmony with the physical, ecological, and social characteristics of the city.

2.1. City Scale

The main objective of the analysis at this scale was to identify urban centers (Figure 1A) where GI would have the most significant impact based on spatial competitiveness. The methodology is based on space syntax theories, widely used since the early 1980s by architects and urban planners to simulate how spatial configurations of buildings and cities influence human movement and social interaction. Space syntax has proven a powerful theoretical and methodological framework for analyzing urban structures and dynamics, predicting human movement within 55–75%, as well as land-use distribution and other spatial phenomena [23]. Here, it helps identify urban centers where GI can provide the greatest urban regeneration benefits.

The initial phase involved creating a computational spatial analysis model of city roadways, assessing their integration and connectivity through two main metrics: Global Integration and optimal connection (Choice). These analyses are conducted using DepthmapXnet 0.35 and are visualized at different radii highlighting spatial configurations that significantly influence urban mobility. “Integration” measures how interconnected the urban environment is, highlighting more central spaces that encourage activity concentration and support the long-term success of designed or rehabilitated areas. (Figure 2, left).

Another analytical method involves calculating the optimal connection (choice), identifying the most frequently used routes within the city’s road network. Different scales of analysis yield distinct usage patterns; a global city-level analysis highlights the main and secondary avenues (Figure 2, right). Analyses at 2500 m emphasize roads more likely used for short bicycle trips, whereas analyses at 1000 m or 500 m highlight pedestrian-scale streets [17].

Additionally, accessibility models were established following the spatial analysis method developed by Ståhle [24], using Place Syntax to identify the clustering or absence of services, businesses, facilities, or people within a specific radius, thus more effectively representing pedestrian perception than traditional measurements (Figure 3).

Applying Marcus’s [16] spatial capital methodology to spatial data from Ciudad Juárez, this study evaluated street integration levels (Figure 4A), population accessibility (Figure 4B), and economic diversification (Figure 4C) around each block, highlighting spatial configurations favoring pedestrian use. This study also considered social diversity indicators by analyzing the co-presence of individuals with varying daily routines, ages, and social classes (Figure 4D) [25].

Lastly, following Ståhle’s [26] model, which explores the relationship between urban form and user experience, studies emphasized the importance of integrating more green spaces into cities. Strategic urban design can enhance the access, quality, and functionality of these spaces, strengthening community identity and cohesion. These studies evaluate existing urban conditions (Figure 4E) and parks (Figure 4F) to measure pedestrian-friendly spaces.

These city-scale analyses produced a synthesized map highlighting priority areas for GI implementation based on spatial capital criteria (Figure 4G), as defined in previous studies by Granados [17], analyzing data from various urban centers within a 1.5 km radius of their geographic center. A full breakdown of the modeling parameters, maps, and accessibility indicators is provided in Appendix A.

2.2. Local Scale

2.2.1. Available Areas for Green Infrastructure

After identifying the city area with the greatest potential for rehabilitation with green infrastructure (GI) at a global level (entire city), a specific 214.24-hectare area in the Historic Center was selected for detailed local-scale study. Public spaces within the available area were divided into 1 m² hexagons to prioritize them based on their morphological conditions, such as their urban runoff absorption capacities.

Initially, a high-resolution digital elevation model (DEM) was generated using drone photogrammetry, allowing for the assessment of topography, potential water catchment areas for GI development [27], slope percentages, and building typology. An extensive review of international GI implementation manuals and guidelines informed an understanding of the physical restrictions applicable to each space. Typically, restrictions such as slopes, groundwater table depths, infiltration rates, and soil contamination levels determine the feasible types of infrastructure. However, since the study area has an average slope of 3%, a groundwater table depth of 60 m [28], type B soil with an infiltration rate of 1 to 5 cm/h [29], and no severe contamination [30], many typical physical limitations were not applicable.

Areas with physical constraints were thus defined based on proximity to buildings and underpasses, which determine their suitability for implementing GI systems requiring high infiltration capacities. A minimum distance of 6 m was maintained to minimize structural risks. Additionally, proximity within 30 m to existing natural drainage infrastructure, such as the irrigation canals (Acequia Madre) crossing through the area, was considered, since this existing infrastructure would naturally control flooding, making further investment less viable (Figure 5).

2.2.2. Fuzzy Logic

Fuzzy logic provides a modeling process to help make complicated decisions, where input parameters are uncertain or derived from heterogeneous data sources. Initially developed by Zadeh [31], fuzzy set theory enables elements of belonging to a class to be of some degree of membership, but not necessarily binary as in the true/false classification. Using spatial analysis for GI planning, the technique has been useful to transform subjective criteria ranging from physical restrictions to socio-environmental needs to a suitability scale with a range of continuity.

Fuzzy logic has found application in geographic information systems (GISs) for multi-criteria decision analysis (MCDA) wherein diverse layers of information need to be merged [18,32,33]. Unlike Boolean overlays, fuzzy MCDA permits a more realistic evaluation of spatial phenomena by the allowance of gradations in suitability and the incorporation of the expert judgment necessary for the setting of thresholds and membership parameters. The method is particularly useful in urban planning, where spatial decisions must consider both quantifiable environmental conditions and sociomorphological patterns that cannot be measured against hard thresholds.

Fuzzy logic was used here to merge five spatial parameters for GI prioritization in the urban environment with landscape fragmentation. All input variables were transformed into a fuzzy membership raster from an ascending linear or Gaussian function, depending on the relationship of the variable analyzed. The fuzzification transformations were accomplished using ArcGIS Pro’s fuzzy membership tool rasterizing geospatial data and assigning a degree of membership ranging from 0 (completely unsuitable) to 1 (fully suitable). The choice of the membership function depended on the nature of the variable and is described below:

Gaussian Membership Function: This is used when there is an optimal value around which suitability is highest, decreasing symmetrically on either side. The function is as follows:

$$\mu \left(x\right)=e^{{-\left({\displaystyle \frac{x c}{\sigma }}\right)}^2}$$

where there are the following:

• c = the center (optimal value);
• σ = the spread of the function.

Linear Membership Function: This is applied when suitability increases or decreases linearly with the variable. The function is defined as follows:

$$\mu \left(x\right)=\left\{\begin{array}{c}0,\\ {\displaystyle \frac{x-a}{b-a}}\\ 1\end{array}\right., \begin{array}{c}x\le a\\ a<x<b\\ x\ge b\end{array}$$

where there are the following:

• μ(x) is the membership value;
• x is the input variable determined by the raster;
• a and b are the lower and upper values applied here as an ascending function.

The Gamma Function is defined as follows:

$${\mu }_{\mathrm{G}\mathrm{a}\mathrm{m}\mathrm{m}\mathrm{a}}\left(x\right)={\mu }_{\mathrm{P}\mathrm{r}\mathrm{o}\mathrm{d}\mathrm{u}\mathrm{c}\mathrm{t}}{\left(x\right)}^{\gamma }·{\mu }_{\mathrm{S}\mathrm{u}\mathrm{m}}\left(x\right){\left(x\right)}^{1-\gamma }$$

where

$${\mu }_{\mathrm{P}\mathrm{r}\mathrm{u}\mathrm{d}\mathrm{u}\mathrm{c}\mathrm{t}}\left(x\right)={\prod }_{i=1}^n{\mu }_i\left(x\right)$$

$${\mu }_{\mathrm{S}\mathrm{u}\mathrm{m}}\left(x\right)={\prod }_{i=1}^n\left(1-{\mu }_i\left(x\right)\right)$$

γ is the gamma parameter (0 ≤ γ ≤ 1), controlling the degree of compensation between criteria. In this study, a gamma value of 0.8 was selected to allow moderate compensation, ensuring that no single criterion dominated the suitability analysis.

The Kernel density tool in ArcGIS Pro 3.4.3. was used to generate continuous raster surfaces from line and point features (e.g., streets, bus routes, and land-use points), capturing their spatial influence within a specified radius. The tool estimates density based on the following function:

$$f\left(x,y\right)={\displaystyle \sum _{i=1}^n}{\displaystyle \frac{1}{h^2}}·K\left({\displaystyle \frac{d_i}{h}}\right)$$

where there are the following:

• $f(x,y)$ is the estimated density at a location $\left(x,y\right)$;
• $n$ is the number of features (e.g., street vertices or transport routes);
• $h$ is the bandwidth (search radius, set at 200 m in this study);
• $d_i$ is the distance from feature $i$ to location $(x,y)$;
• $K\left(u\right)$ is the kernel function, typically a quartic kernel.

  $$K\left(u\right)={\displaystyle \frac{15}{16}}{(1-u^2)}^2 \mathrm{for} \left|u\right|\le 1; K\left(u\right)=0 \mathrm{otherwise}$$

This ensures that the features closer to a given location have a stronger influence on the resulting density, which decreases smoothly with distance. The Kernel density output raster was then normalized from 0 to 1 and transformed in this case study using a fuzzy ascending linear function to serve as an input in the fuzzy overlay process (Figure 6).

• A fuzzy logic model was then created to integrate the flood depth and runoff velocity in order to generate a flood risk map and assess the absorption capacity of SUDSs at a local scale. The simulation was conducted in ArcGIS Pro using a 25-year return period storm, representing a 6 h rainfall event with a total accumulation of 89 mm, based on a 0.5 m resolution simulation. This scenario was selected as it reflects local hydrological conditions and modeling recommendations [1]. As noted by Muthanna [34], identifying the main runoff flow paths helps pinpoint high-potential areas for SUDS implementation. In this case study, the drag factor was omitted, since the runoff velocities did not exceed the low-impact threshold identified by Yamanaka et al. [35] and were therefore considered a positive factor for intervention.
  Both maps were combined using fuzzy logic, first applying a Gaussian function with a mean of 0.75 m and dispersion of 0.5 to the flooding depth raster (Figure 7A), gradually reducing values in areas with no flooding or extreme flooding conditions requiring other hydrological solutions. Similarly, a linear function was applied to the maximum runoff velocity values of 0.27 m³/s (Figure 7B). Both layers were then combined using a fuzzy gamma function at 0.8 through the fuzzy overlay tool, balancing the influence of both factors and flexibly integrating zones with high flood and runoff susceptibility (Figure 8A).
• For morphological constraints, a modified spatial capital analysis identified the most urbanistically sustainable pedestrian-level spaces lacking green areas [17]. This value was calculated by summing and averaging five accessibility models used to create the spatial capital (integration, population accessibility, economic diversification, urban environment, and social diversity). The values obtained from the park accessibility model were subtracted from the modified spatial competitiveness result, normalizing values from 0 to 1 using a linear ascending function for fuzzy logic analysis (Figure 8B).
• The optimal connection (choice) analysis of the streets was used at an influence distance of 2500 m to determine the circulation preference at the local level, thus helping to connect the urban centers most likely to create pedestrian dynamics in the area, [17]. This was conducted by converting the streets into a raster using the ArcGIS Pro “Kernel Densit” tool by setting a search radius at 200 m. This tool is normally used in urban studies to capture the influence effects of urban elements [36]; in this case the choice value at 2500 m from each road was able to be used in the fuzzy logic model using an ascending linear function (Figure 8C).
• To evaluate the relationship of the 88 public transport bus routes in our study area with the traffic they would generate, the Kernel density technique was used at 200 m, generating a raster that identifies the areas with the greatest accumulation of public transport routes, highlighting the southern area where buses avoid entering the historic center. Thus, an ascending linear function was used to relate the flow of pedestrians who would walk towards them, acting as an indirect indicator of pedestrian movement [23], since most routes do not have established stops. Therefore, integrating the unique characteristics of the density of public transport lines is considered relevant since they cross strategic areas near the international bridge, serving as attraction nodes for non-residential pedestrian flow, which includes tourists or people transiting to the US (Figure 8D).
• The distribution of pedestrian movement can also be explained by the number of commercial fronts that are established in the streets of the study area [23]. This is another of the main factors influencing future activity in the area, regardless of what small changes the municipality may make in trying to create new commercial fronts around them [37]. Therefore, a methodology has been used to explore the m2 of the different land uses within a 25 m buffer of the 1047 vertices of the streets analyzed in the study area. This value is reasonable to capture the urban typology of each street since the smallest blocks in the study area measure 50 m in width. The land-use variables obtained from the Historic Center Urban Development Plan were grouped into seven different categories (commerce, services–amenities, housing, parks and recreation, vacant and unused land, industrial, and under construction) to obtain the percentage of land-use types for each street segment. A kernel density raster was then created at 200 m with the percentage of commerce–service–amenity land use for the street segments, thus measuring their level of attractiveness in the study area using an ascending linear function (Figure 8E).
• The integration of the five fuzzified layers was carried out using the fuzzy overlay tool with the gamma operator, which integrates restrictive and compensatory logic. A gamma value of 0.8 was chosen in order to offer moderate compensation between factors so that no one factor would be able to overshadow the composite suitability measure (Figure 8F).

In this case, the model identifies flood-prone lands as being highly suitable for GI measures, reversing the traditional way of considering such lands as constraints. This decision follows the aim of tackling hydrological weaknesses using SUDSs. Additionally, spatial configuration indicators based on space syntax theory [15] were used in the analysis to quantify the underlying morphological rationality of the street network. This addresses an essential methodological gap in the literature, as urban connectivity is rarely incorporated in flood resilience and GI planning models. By combining network analysis with environmental, accessibility, and land-use factors, this fuzzy logic approach introduces a reproducible and adaptable system for guiding GI implementation in data-poor, rapidly urbanizing contexts.

2.2.3. Automated GI Selection Matrix

Green infrastructure (GI) strategic placement can be effectively guided by automated GIS-based models that apply rule-based logic to align GI typologies with site-specific physical and hydrological conditions. In this study, the most appropriate GI type for each 1 m² hexagon in the intervention area was selected using a geospatial decision matrix implemented through Attribute Selection and spatial analysis tools in ArcGIS Pro. This matrix evaluates the flood depth, runoff velocity, infiltration restrictions (due to proximity to sensitive infrastructure), and area typology, based on spatial layers generated in previous stages.

The decision-making logic is structured using Boolean classification rules, applying an if/then hierarchy that encodes criteria thresholds following the matrix detailed in

Table 2. Selection matrix of green infrastructure typologies [41,42,43,44,45,46].

Green Infrastructure (GI)	Flood Level (m)	Runoff Velocity (m3/s)	Infiltration Restrictions	Area Type	Infiltration Rate (mm/h)	Maximum Infiltration (mm)	Description
I-DIP (Permeable Pavement with Geo-cellular Deposits)	<1	N/A	Yes	Sidewalks, Streets, Parking lots, Impervious recreational areas	100	1000	Mixed system with permeable pavement and underground geo-cell storage. High storage capacity.
D-PAV (Mixed Permeable Pavements)	0.3−1	N/A	No	Sidewalks, Streets, Parking lots, Impervious Recreational Areas	50	200	Permeable pavement that combines tree pits and concrete tiles with gravel and clay. It facilitates water infiltration and allows tree roots to grow. Ideal for urban areas with vegetation.
I-PAV (Simple Permeable Pavement)	>0.3	N/A	No	Sidewalks, Streets, Parking lots, Impervious Recreational Areas	30	100	Permeable pavement used near buildings. It infiltrates directly into the soil without a deep layer of gravel and clay mixture, ideal for urban areas where foundation damage must be avoided.
I-ZAN (Infiltration Trenches)	>1.5	<0.05	Yes	Green Areas, Road Medians, Vacant lots	100	500	Infrastructure with a high infiltration capacity that controls runoff speed, preventing erosion and serving as the first line of defense in integrated systems.
I-BIO (Bio-retention Strips)	0.3−1.5	>0.05	Yes	Green Areas, Road Medians, Vacant lots	50	200	They use a specifically designed soil layer where native vegetation grows and that helps maximize the storage and filtration of storm water into underground drainage systems.
I-JAR (Rain Garden)	>0.3	>0.05	No	Green Areas, Road Medians, Vacant lots	50	100	They use a specifically designed soil layer where native vegetation grows to maximize the infiltration, storage and filtration of storm water into underground drainage systems.
F-Est (Infiltration Ponds)	<1.5	N/A	Yes	Green Areas, Road Medians, Vacant lots	50	1000	Designed with a highly permeable substrate, they have a high water storage and filtration capacity. They are ideal for recharging aquifers. They can be used in areas larger than 45 m2 with a 1:2 ratio.
I-CUN (Vegetated Swale)	0.3−1.5	N/A	No	Green Areas, Road Medians, Vacant lots	30	150	Designed with containment barriers to channel water towards high infiltration systems.
I-POU (Infiltration Wells)	<1.5	N/A	Yes	Green Areas, Road Medians, Vacant lots	100	1000	Structures excavated in the ground, designed to infiltrate water into aquifers. In areas smaller than 10 m2, the ratio is 1:2.
Semi-permeable Urban Areas (spaces with some vegetation, compacted soil)	Base Model	Base Model	Base Model	Green Areas, Road Medians, Vacant lots	5	20	Ciudad Juárez has an arid climate, with predominantly sandy and clayey soils, assigned the lowest value of a type B soil (NRCS) because urbanized areas tend to compact over time.

. This approach enables a streamlined, scalable, and replicable method for matching GI types to localized urban conditions. It offers a comprehensive and systematic framework for flood mitigation through nature-based solutions, validated by its growing application in contemporary research. Several studies have implemented similar GIS-based decision logic to optimize GI placement, demonstrating its practicality and scientific grounding [38,39,40].

2.2.4. Multiscale Geographically Weighted Regression (MGWR) Models

Multiscale Geographically Weighted Regression (MGWR) is a spatial statistical technique used to find how relationships between variables change across a geographic space. Unlike traditional regression models [23] that assume relationships are constant across the study area, MGWR allows each explanatory variable to operate at its own spatial scale, capturing spatial heterogeneity more accurately [47]. This is particularly useful in urban morphology and land-use studies, where local variations in accessibility, connectivity, or socio-environmental context can strongly affect outcomes.

In the latest urban research, MGWR has been successfully combined with space syntax measures to analyze how network connectivity influences urban outcomes. For example, Wang et al. [48] used integration values to explore how street connectivity contributed more strongly to land quality in some districts than in others. Similarly, Zhang et al. [49] used MGWR with space syntax metrics set at a 3 km radius to analyze how well-connected green corridors have a positive effect inciting people to use them for jogging more often in Shenzhen. Cao et al. [50] also used MGWR to map spatial variation in housing prices and its correlation to proximity to parks and well-integrated streets, confirming that network morphology impacts urban value differently across a city.

In this study, MGWR is applied in ArcGIS Pro to confirm the relationship between optimal connection (choice) and various land-use typologies in Ciudad Juárez’s historic center, joining spatial configuration theory into GI planning. Following earlier studies that used configurational measures to predict land use [51], we tested the relationship between street connectivity and land-use intensities at three different scales.

The percentages of five land-use categories (commerce, services–amenities, housing, parks and recreation, vacant and unused land) were calculated for each street segment. Three MGWR models were then created using choice values at 2500 m, 1000 m, and 500 m radii as the dependent variable. Then, inverting the process generated nine different MGWR models using the land-use categories as the dependent variables and the “Choice” values of the street as the explanatory variables. All the variables were normalized from 0 to 1 prior to modeling. Industrial and under-construction land-use categories were excluded from the choice dependent analysis, and so were the parks and commerce percentages for the land-use dependent analysis due to irregular spatial distribution and a lack of statistical significance, which introduced instability in preliminary models.

2.3. Microscale

In this third approach, the Zone II.2 Centro micro-watershed was used as a case study. The area was identified by the Ciudad Juárez Risk Atlas [52] as one of the most vulnerable and at-risk areas for flooding, due to its location being within the hydraulic system that forms the Mariano Escobedo Stream. This creates a large floodplain and the accumulation of storm water in the overpass located on Insurgentes Avenue. This can be mitigated by implementing flood control systems, taking advantage of the available spaces in the Guadalupe Mission Plaza and surrounding areas (Figure 9A). Similar spatially constrained drainage contexts have been addressed with decentralized GI retrofits in Latin American cities such as Barranquilla, where SUDSs were proposed instead of storm drains [7].

For this purpose, the design of a sustainable urban drainage system similar to that of the case originally defined by Schutze et al. [53] was considered, which proposes controlling floods using GI, which are connected to the mixed drainage system, since there is no functional storm drainage network in the study area. These types of hybrid infrastructure systems are the most effective in developing countries considering the resilience they demonstrate in technical implementation, economic, and environmental aspects [54].

To evaluate the influence of the SUDSs at this scale, we relied on the suitability map obtained from the fuzzy logic analysis. We selected the areas that could be modified with slight topographic modifications, increasing their flood retention capacity. A high-resolution digital elevation model (DEM) was created by performing a photogrammetric survey of the study area, which clearly captured the terrain’s topography at a 10 cm resolution. High-resolution DEMs are widely used in SUDSs siting to identify microtopography and flow paths for runoff interception, especially where flow monitoring is unavailable [35].

This allows us to analyze the main runoffs within the micro-watershed and nearby watersheds with more precision than the local-level analysis. In this way, we proceeded to evaluate how water could be channeled, using speed bumps or other infrastructure, to improve the performance of the proposed GI (Figure 9B).

A hydrological analysis was then performed with ArcGIS Pro´s flood simulator to measure flood reduction over time, subjecting a base model and a model with GI to a designed storm with a maximum intensity of 14.83 mm/h for 6 h. This design-storm approach is common in data-scarce regions, offering a valid basis for modeling runoff and GI performance in the absence of calibration [55].

The infiltration rate and maximum infiltration of each type of green infrastructure proposed in the study area are used and compared to the base simulation, which considers all permeable areas as semipermeable urban spaces. The results of the two multidimensional flood level models are subtracted, and zonal statistics are calculated within each hexagon in the study area. This allows us to analyze flood dynamics over time and measure how GI reduces runoff peaks.

To analyze the efficiency of GI, three Space–Time Cubes (STC) (Figure 10A) models were created in ArcGIS Pro (base model, GI model, and flood-level difference model). This allows us to represent how the flood level varies over time in each space available for GI implementation. The Emerging Hot Spot Analysis (EHSA) (Figure 10B) tool was then used on the STC models, detecting where GI successfully reduces water accumulation, as well as patterns in absorption over time. Spatiotemporal flood tools like STCs and EHSA are increasingly used to detect trends and identify vulnerable zones in flood-prone urban areas [13,56].

Finally, the Time Series Clustering (TSC) tool was used in the STCs, which allows us to group the GI into cluster classifications with similar infiltration behaviors or abrupt changes in their floodwater infiltration performance. This helps us evaluate which systems need to be modified to absorb new floodwater volumes after the terrain has been modified, showing how long the water can infiltrate before reaching its saturation capacity.

3. Results

The development of a multiscale, multi-criteria analysis is necessary to effectively and holistically integrate GI into the city. To present a methodology that would identify the areas that would best contribute to managing the natural water cycle in an established urban environment, its pre-existing urban dynamics had to be taken into account, along with an analysis of its spatial configuration at different levels to address both environmental and spatial planning challenges simultaneously [57].

This section presents the results of applying this methodology at the city scale, local scale, and microscale, combining spatial analysis tools (such as space syntax, fuzzy logic, and MGWR) with hydrological modeling and geospatial simulation techniques. By integrating land-use patterns, urban forms, and runoff behavior, the results provide a robust framework for locating, prioritizing, and adapting green infrastructure to specific urban conditions. Moreover, spatiotemporal tools like Space–Time Cubes (STCs), Emerging Hot Spot Analysis (EHSA), and Time Series Clustering (TSC) allow the assessment of GI performance over time, offering insights into which systems remain effective and which require redesigning. The findings are organized by scale, demonstrating how each level contributes to a comprehensive urban resilience strategy.

3.1. City Scale Results

The initial phase of the process involved identifying spaces with potential for implementing GI at the city level, integrating space syntax theories as a diagnostic tool to assess spatial accessibility and connectivity patterns. These theories, developed by Hillier [15], revealed that the centralized structures underlying the city’s road network correlate with morphological transformations throughout its history. The city’s initial urban growth spurt took place until the 1970s, when tourism was the predominant economic activity in the city. This caused these neighborhoods to present a grid-like layout with narrow, short streets, and the main objective was to create a link between the historic center, the PRONAF area, and the international bridges to the United States.

Due to their spatial configuration, their roads today continue to have a high level of integration with the rest of the city. This is reflected in the high level of pedestrian accessibility in all studies, except for the study of park accessibility due to the lacking urban regulations at the time [17], as shown in the spider chart in Figure 11.

Analyzing the main challenges facing these urban centers and potential ways to address them is essential for sustainable urban development. The results indicate that the first step could be the strategic implementation of green spaces, which would counteract the current problem of unused lots in the area connecting the historic center and the PRONAF area. These vacant spaces generate deterioration and insecurity in the neighborhood, which in turn diminishes the desirability and value of surrounding properties in the event of an urban reform project [58].

Transforming these areas into green spaces could have the opposite effect [48], paving the way for future plans toward a more pedestrian-friendly city by creating green corridors linking the two urban centers (Figure 12). Given the short distance between them (approximately 2500 m), there is robust potential to encourage active mobility and reduce dependence on motorized transport [49]. Moreover, the presence of a human-scale urban grid, characterized by permeable street patterns and mixed land uses, enhances walkability and supports densification and rehabilitation efforts in these key corridors [51]. Therefore, these areas represent strategic zones where the city should prioritize investment in sustainable infrastructure and spatial reintegration [36,48].

3.2. Local Scale Results

At the local scale, green GI implementation opportunities in the historic center were first identified using hydrological modeling and spatial suitability analysis. By combining fuzzy logic with a geospatial decision matrix, this study evaluated the flood risk, runoff dynamics, and land-use conditions to determine the best GI types for each location. This process generated an initial proposal for GI distribution based on terrain, drainage needs, and proximity to urban infrastructure (Figure 13).

The most implementable type of GI in the study area was identified as the infiltration permeable pavement (I-PAV), which is possible to implement in 265,000 m² of sidewalks since it allows controlled infiltration close to buildings, followed by its implementation in 54,672 m² of parking areas. Also, the implementation of 43,519 m² of permeable pavement with a retention system is recommended, which integrates tree pits and interlocking concrete with a sublayer of gravel and clay, which has the highest infiltration rate and resistance in conditions with a high level of sediment accumulation [59]. This, together with its multifunctionality of infiltrating, storing, and channeling excess rainwater through pipes towards the mixed drainage, makes it a priority system since it will also improve mobility and appearance in the area [38].

Currently, there are a total of 127,428 m² of permeable surfaces for implementing GI in the study area, which are neglected, giving a poor urban image to the area. Most of these are only used as passageways leading to commercial centers or to crossings into the US. These could be rehabilitated, improving the city’s aesthetics and biodiversity, using a combination of GI according to its infiltration and runoff detention needs. It was estimated that a total of 123,799 m² of rain gardens and 11,207 m² of bioretention strips could be implemented in the study area. And in this way only using more technical infrastructure such as infiltration wells as a last option due to their implementation and maintenance costs (Figure 14).

To strengthen the spatial logic of GI placement and to prove if space syntax theories functioned according to previous studies, MGWR analysis was conducted. This allowed us to assess how different land uses correlate with street connectivity at multiple spatial scales (choice 500 m, 1000 m, 2500 m), validating the relevance of key corridors for strategic GI intervention.

The MGWR results confirm that streets with high choice values at 2500 m and 1000 m are key to commercial activity in the area. The model also explains approximately 54.23% of the variability in the vacant-lot percentage values. This is a concrete result for this type of analysis and suggests a significant correlation between streets with an optimal connection (choice) at 500 m and the distribution of vacant lots. Thus, this validates the hypothesis that spatial configuration plays a central role in urban functionality in the historic center, since low-traffic areas are more likely to end up abandoned. Thus, highlighting the importance of integrating this information into GI planning for strategic investment, the relationship between land uses, and the optimal connection values allows for prioritizing the most relevant streets for urban connectivity and sustainable regeneration.

• Areas with high accessibility within 500 m (choice 500 m) would be ideal for community-oriented green infrastructure (small urban parks or community gardens).
• Areas with high accessibility within 1000 m (choice 1000 m) can serve as green infrastructure that complements commerce, such as tree-lined streets, wide pedestrian areas, and market spaces.
• Areas with high accessibility within 2500 m (choice 2500 m) could have more iconic green infrastructure, marking urban landmarks to connect the two urban centers (large urban parks, vegetated civic plazas, green corridors).

3.3. Microscale Results

It was shown that the geomorphology within the micro-watersheds in the study area follows a dendritic drainage pattern, where runoff follows short, irregular courses, partially following the reticular structure of the urban fabric. This results in the underutilization of runoff storage capacity in the nearest catchment area, since the few available areas of permeable surfaces where GI implementation is possible are at a higher level than the sidewalks. Therefore, another type of ground-level infrastructure must be proposed in these spaces, thus diverting runoff inward and increasing their infiltration capacity. This type of microscale terrain-based retrofit is consistent with the SUDS practices applied in other semi-arid cities, where minor topographic changes have been shown to improve runoff interception without major infrastructure investment [44].

Thus, the terrain was modified by simulating three 30 cm high physical barriers in strategic locations, channeling the main runoff from the study area to the catchment areas. A spatial statistical analysis was then performed, allowing us to quantify the influence of GI with high precision. The micro-topography was altered using a 10 cm resolution DEM, and depressions were modeled 20 cm below the sidewalks based on runoff flow directions, as recommended in SUDS implementation guidelines [46]. By modifying the relief of the possible catchment areas and using the GI typology proposed at a local scale, 1085.72 m³ of water can be infiltrated into the micro-basin until the catchment areas overflow and the storm water runoff ends up at the critical flooding point in the Insurgentes Avenue (Figure 15).

The GI works in much of the study area, reducing flooding and creating a safer and more hospitable environment. However, some specific areas become saturated and redirect water, causing new flooding hot spots, since terrain modification had not been considered when proposing the GI at the local scale. Therefore, using a combination of the EHSA and TSC analyses reveals which GI is functioning correctly and which areas require system improvements.

In this study, the EHSA tool was configured to identify statistically significant hot spots using a fixed time-step of 30 min in a 1 m² hexagonal grid, following best practices in space–time flood modeling [13,60]. This allowed for the identification of both persistent and newly emerging hot spots, offering a detailed understanding of temporospatial GI effectiveness.

The results from TSC show that in the Functional Cluster of GI, the percentage of hot spots in critical flooding is generally lower, indicating that these GIs are successful in reducing flooding. In contrast, the Dysfunctional Cluster category shows a higher percentage of time when the infrastructure is flooded, suggesting that these GIs are unable to efficiently manage the amount of runoff or become saturated quickly. Time Series Clustering has been used in flood resilience studies to evaluate GI performance over storm events [60]. In this case, the clustering patterns clearly differentiate zones that require structural upgrades from those where GI is working as intended.

Therefore, in impermeable spaces, 1953 m² of mixed permeable pavement (D-PAV) and 4796 m² of simple permeable pavement (I-PAV) present flooding issues. It is suggested to change these GI systems by implementing more cellular storage systems (I-DIP) or infiltration wells. Similarly, it is recommended to replace the flooded 3751 m² of rain gardens (I-JAR) with bioretention strips (I-BIO), which have a higher runoff absorption capacity (Figure 16).

Using the EHSA allowed us to evaluate the impact of GI in two scenarios. On the Base model (Figure 17B), areas with recurrent flooding were identified as “Intensifying Hot Spots”, since they remained flooded during 90% of the simulation. These areas are surrounded by zones where water accumulation fluctuated until it overflowed the catchment area, which are labeled as “Oscillating Hot Spots”. Following the implementation of green infrastructure (Figure 17C), a 38% reduction was observed in areas with intensifying flooding, transforming them into zones of oscillating hot spots. This led to a 318% increase in areas classified as new hot spots.

This means that while flooding in these areas did not disappear entirely, storm water accumulated more gradually and to a lesser extent, only peaking during the final 30 min of the 6 h storm simulation. Such a change indicates that the GI systems successfully delayed peak flooding and increased infiltration time, buying critical time during intense rainfall events, which is an important outcome for urban flood resilience planning.

In summary, the results across the three scales exhibit the usefulness of joining spatial connectivity analysis with hydrological simulations. The findings confirm that GI can be optimized by spatial context: from macro level connectivity interventions (green corridors), to targeted GI typologies (permeable pavements, rain gardens), to precise site-scale retrofits that reduce flood tenacity over time.

4. Discussion

Overall, the methodology employed in this study successfully identified areas with the highest potential for integrating green infrastructure (GI) within fragmented urban fabrics, maximizing the social, ecological, and economic benefits of GI as a tool for mitigating hydrometeorological risks. By combining spatial configuration analysis and hydrological modeling across multiple scales, this approach effectively addresses a critical gap in current urban planning research. Although the integration of spatial morphology and storm water modeling has been widely recommended in the recent literature as a strategy to enhance multifunctionality and urban resilience [13,57,60], its application in practice remains limited. This study demonstrates that such a combined methodology is not only feasible but offers significant advantages for planning adaptive, site-responsive GI strategies in complex data-scarce urban environments.

Based on the morphological analysis carried out on the city-scale study [15,16], two main urban centers (historic center and PRONAF) were identified as the best areas to focus efforts to evaluate the potential implementation of GI. These centers were selected due to their spatial predisposition to create a sustainable environment and the potential benefits of creating flood risk mitigation strategies to help rehabilitate them. This supports similar findings by Cao et al. [50] and Zhang et al. [49], who used MGWR to show how green corridors and well-connected areas increase land value and promote active transportation and demonstrate that choice-based network connectivity at 500 m and 2500 m radii aligns with local commercial and pedestrian dynamics in the historic center [51].

Additionally, studies by Li et al. [36] demonstrate that routes between two centers are not only influenced by high-demand areas, such as transportation or commercial hubs, but that users prefer areas with green corridors and will often choose them over other routes. This highlights the importance of GI promoting connectivity and comfort for cyclists and pedestrians, since creating more accessible routes will attract more users and reactivate the economy. Considering that, the municipal government of Ciudad Juárez is planning the city’s first bike-sharing system in the study area. Therefore, the choice-type analysis helps predict cyclist traffic in this zone [61], with assisting in validating a strategic rehabilitation plan being crucial for certain roads and facilitating sustainable urban mobility.

At the local level, different spaces within the historic center were evaluated to determine their potential for implementing different types of GI. Considering fuzzy MCDA to assess the benefits provided by GI [18,32,33], it is possible to identify where to focus rehabilitation efforts by conducting a sustainable urban analysis using geospatial tools. This approach is aligned with methods used by Failache et al. [39] and Gulshad et al. [40], where spatial and performance-based criteria were combined with rule-based logic to prioritize GI placement, demonstrating the scientific advantages of automated matrix selection models.

On the other hand, a viable and high-potential alternative is implementing strategies for renaturalizing the canal system within the study area. Creating adaptation plans for areas adjacent to these canals using riparian restructuring techniques [62] would provide essential goods and services to resolve hydrometeorological risks in the urban center. Despite having less than the recommended 30 m distance between the river channel and first constructions in some sections, these solutions could be implemented in 22.64 hectares adjacent to the canals, contributing to the functional beautification of these zones. Thus, this creates new recreational areas that mitigate pollution effects in the streams and adapts them as biological corridors within the urban fabric.

While leaving the analysis of private spaces for future studies, the following points are determined: Buildings that can implement green roof solutions could be analyzed. The presence of roofs with a slope of less than 10% could be prioritized for implementation in public, commercial, and industrial buildings. Residential buildings, on the other hand, could implement roofs with reflective materials and rainwater collection cisterns, reducing flooding and heat islands in the area. Making these nature-based retrofits is particularly effective in semi-arid regions [44], where surface-level GI is limited.

Finally, at the micro level, the results from the previous scales were used to identify the best possible solution, based on the specific geospatial constraints of each area, in terms of topography, hydrology, land use, and mobility patterns. This generated multiple benefits, such as flood risk mitigation, improved mobility and public spaces, as well as improvements in the quality of life and social well-being of these communities. By using spatiotemporal tools like STCs and EHSA, this analysis not only mapped outcomes but revealed the evolution of flood hot spots in time, a method increasingly advocated for adaptive GI planning [13,56] and which shows promise for future disaster planning due to its high spatial and temporal precision.

5. Conclusions

This study demonstrated that a multi criteria, multiscale approach integrating spatial configuration and hydrological modeling can be effective in informing the implementation of green infrastructure (GI) in fragmented urban settings. By integrating space syntax, fuzzy logic, and storm water simulation, the proposed framework facilitates data-driven, context-sensitive planning at the city scale, local scale, and microscale.

The model was successful in not only identifying where GI must be implemented but also in what typologies are most appropriate based on urban form and hydrological function, re-establishing the underlying research objective of improving connectivity and flood resilience in semi-arid, data-poor urban areas.

The proposed methodology is implementable in other Mexican border cities (Figure 18) and transferable to similar international settings that are experiencing rapid urbanization, spatial disarticulation, and climate exposure [63]. It aligns with global sustainability agendas, particularly SDG 6 (Clean Water and Sanitation), SDG 11 (Sustainable Cities and Communities), and SDG 13 (Climate Action).

Yet, given the scarce availability of high-resolution hydrological data, synthetic storms and USDA soil infiltration rates were employed, though these are sufficient for simulating runoff and would be significantly improved through calibration against empirical field data. Likewise, the inclusion of community engagement and sociospatial feedback loops would be able to improve social equity in GI planning.

Future applications need to expand the model by integrating additional environmental and social criteria, such as urban heat island effects, long-term maintenance costs, and community preference, with the application of tools like the Fuzzy Analytic Hierarchy Process. The continued integration of spatiotemporal tools (e.g., STCs, EHSA) and spatial network measures will further promote adaptive and evidence-based GI planning in the face of evolving urban and climatic stressors.