1. Introduction
Green Infrastructure (GI) is a foundational network of natural and semi-natural elements crucial for delivering ecosystem services, enhancing biodiversity, and improving urban well-being. In semi-arid and arid regions like Iran, GI faces a paradoxical threat: its necessity is greatest due to harsh climatic conditions, yet its viability is most compromised by rapid, unplanned urban expansion and intense water stress. While the dynamics of GI are well-studied in temperate climates, a significant research gap exists regarding its long-term spatiotemporal changes and ecological consequences in water-scarce urban settings.
This study addresses this critical gap by conducting a systematic, metric-based analysis of GI transformations in Borujerd, Iran, a representative mid-sized city in the semi-arid Zagros region. Over a 35-year period (1990–2025), this research moves beyond simple land cover change detection to quantify the qualitative degradation and fragmentation of the urban ecological matrix, providing insights critical for sustainable urban planning in arid environments.
2. Methodology
The study employed a robust multi-temporal remote sensing and spatial analysis framework. The study area encompassed Borujerd city and its ecologically connected peripheral villages to capture the complete urban-rural ecosystem. The methodology was structured in three primary phases:
2.1. Data Acquisition and Pre-processing: Four Landsat satellite images (Landsat 5/TM for 1990 and 2000; Landsat 8/OLI for 2018; Landsat 9/OLI-2 for 2025) were acquired for the summer season. All images were pre-processed in TerrSet (v18.31) to Level 2 Surface Reflectance, with additional atmospheric correction using the Dark Object Subtraction (DOS) algorithm and geometric standardization to ensure cross-sensor comparability.
2.2. Land Cover Classification and Accuracy Assessment: Land cover was classified into three primary classes: (1) Built-up areas (urban fabric, infrastructure), (2) Green Infrastructure (parks, gardens, farms, natural vegetation), and (3) Barren land (bare soil, desertified areas). A supervised classification was performed using the Maximum Likelihood Classifier (MLC) algorithm. Accuracy was rigorously assessed using independent ground control points, resulting in high accuracy levels (Overall Accuracy >90%, Kappa Coefficient >0.82 for all years), meeting international standards for landscape studies.
2.3. Landscape Metric Analysis: The classified maps were exported to Fragstats (v4.2) to calculate 10 key landscape metrics at both the class and landscape levels. These metrics were selected to capture different dimensions of landscape change:
Quantitative/Area Metrics: Class Area (CA), Percentage of Landscape (PLAND).
Structural/Configuration Metrics: Number of Patches (NP), Largest Patch Index (LPI), Edge Density (ED), Landscape Shape Index (LSI), Effective Mesh Size (MESH).
Functional/Pattern Metrics: Contagion (CONTAG), Shannon's Diversity Index (SHDI), Shannon's Evenness Index (SHEI).
This multi-metric approach allowed for a comprehensive assessment of fragmentation, connectivity, and spatial heterogeneity.
3. Results and Discussion
The results reveal a dramatic transformation of Borujerd's landscape, characterized by extensive urbanization and the severe fragmentation of its green spaces.
3.1. Land Cover Change: Built-up areas expanded explosively by 128% (from 1,206.54 ha in 1990 to 2,748.6 ha in 2025), primarily at the expense of barren land, which decreased by 25%. The GI area showed a fluctuating trend, decreasing from 4,296.6 ha in 1990 to a low of 3,849.84 ha in 2000, before a partial recovery to 4,388.94 ha by 2025. This late recovery is likely due to afforestation efforts or agricultural expansion but masks a more profound qualitative degradation.
3.2. Green Infrastructure Fragmentation: Despite the net recovery in area by 2025, the landscape metrics revealed severe structural fragmentation:
The Number of Patches (NP) for GI increased by 74% (from 651 to 1132), indicating that once-continuous green areas were subdivided into smaller, isolated pieces.
The Largest Patch Index (LPI) decreased by 17%, signifying the loss of core habitat areas essential for supporting biodiversity.
Edge Density (ED) increased, indicating more complex and irregular boundaries between green patches and urban areas, increasing edge effects.
Effective Mesh Size (MESH) nearly halved, demonstrating a significant increase in the isolation of GI patches and a collapse of ecological connectivity.

3.3. Overall Landscape Degradation: At the landscape level, connectivity (CONTAG) decreased significantly from 84.48 to 73.9, indicating a breakdown in the spatial cohesion of the entire landscape. Increased diversity (SHDI) and evenness (SHEI) reflected a more heterogeneous and intermixed landscape, which in this context is a negative consequence of fragmentation and urban sprawl rather than increased ecological richness.
3.4. Discussion of Implications: The findings align with global studies on urban ecological fragmentation but highlight acute challenges specific to semi-arid regions. The intense fragmentation increases water stress, as smaller, isolated green patches have higher edge-to-area ratios, leading to greater evapotranspiration and increased irrigation demands—a critical issue in water-scarce Iran. This creates a negative feedback loop where maintaining ecological aesthetics exacerbates water resource depletion.
The drivers of this change are linked to rapid population growth, rural-urban migration, and municipal policies that prioritized industrial and commercial development over ecological conservation, leading to low-density, sprawling expansion (confirmed by the rising LSI of built-up areas). The study underscores that in arid environments, the spatial configuration of GI is a more critical determinant of its sustainability and ecological functionality than its total area alone.

4. Conclusion and Policy Recommendations
This study concludes that rapid urbanization has driven the severe fragmentation of GI in Borujerd, significantly reducing ecological connectivity and functionality despite a recent increase in green space area. The quantitative and qualitative metrics provided a much more nuanced understanding than area change alone could offer.
The findings necessitate a paradigm shift in urban planning for semi-arid Iranian cities. Key policy recommendations include:
1. Adopting a Hierarchical GI Network: Planning should focus on creating a connected network of core habitats (protecting large existing patches), linear corridors (along rivers, roads), and small integrated green spaces in dense urban fabrics.
2. Water-Sensitive Design: Mandating the use of native, drought-resistant plant species to reduce irrigation needs and integrating blue-green infrastructure for rainwater harvesting and stormwater management.
3. Adaptive Zoning and Governance: Implementing adaptive zoning codes that limit impervious surfaces and promote infill development over urban sprawl. Establishing interdisciplinary "green councils" to unify urban planning, water management, and environmental agencies.
4. Strategic Conservation: Prioritizing the conservation of remaining barren lands not as vacant lots, but as crucial areas for groundwater recharge and future green space expansion.
This research establishes landscape metrics as vital diagnostic tools for arid cities navigating planetary boundaries. Future studies should integrate hyperspectral imagery to distinguish native from non-native vegetation and employ hydrological models to quantify the water trade-offs associated with different GI planning scenarios.