Integrating Ecosystem Services into Urban Carbon Dynamics: A Dual-Scale Spatial Analysis of Land Use, Emissions, and Planning

Abstract

Integrating ecosystem services into urban planning requires analytical tools that connect spatial land-use data with environmental performance. This paper applies a multi-scale, data-driven approach to assess urban carbon dynamics using spatial units that reflect both ecological functions and planning relevance. The study examines the Reggio Calabria Functional Urban Area (FUA) in Southern Italy, using Copernicus Urban Atlas land-use data to characterize spatial patterns and estimate CO₂ emissions and sequestration using parameters derived from established literature and institutional sources. A Spatial Durbin Model (SDM) identifies land uses with direct and spillover effects, revealing how spatial organization shapes urban carbon outcomes. Results reveal a net emission imbalance of approximately 1.85 billion kg CO₂ per year, confirming the region’s role as a net emitter. Transport corridors and discontinuous low-density urban areas show the strongest positive SDM coefficients (+3.48 and +0.78 kg CO₂ m⁻² yr⁻¹, respectively). Forests and agricultural lands show negative effects, indicating potential sequestration functions, though not statistically significant. This suggests that natural and semi-natural land uses contribute little to measurable CO₂ reduction within the FUA. Emissions and sinks display a polarized spatial pattern, with coastal urban zones acting as hotspots and inland areas serving as potential sinks. These findings underscore the need to strengthen ecological connectivity and integrate green infrastructure within dense urban areas to enhance mitigation capacity. The proposed framework shows how spatially explicit, hierarchical analysis can bridge ecosystem services and urban planning, offering a replicable basis for data-informed, climate-responsive strategies.

1. Introduction

Cities contribute a major share of global CO₂ emissions while concentrating the population, infrastructure, and innovation capacity necessary for mitigation [1,2,3]. Urban planning therefore plays a key role in steering these dynamics by integrating ecological efficiency and spatial equity into planning decisions [4,5]. Despite increasing recognition of urban sustainability, the practical integration of ecosystem services into planning remains limited and often theoretical [6,7,8]. This gap reflects the lack of analytical frameworks that connect standardized land-use data to measurable environmental performance across multiple spatial scales [9,10,11].

To address this gap, this study applies a data-driven framework to estimate CO₂ emissions and sequestration using land uses as the main analytical variable. The framework connects spatial data with environmental indicators to assess carbon dynamics beyond administrative boundaries and across functionally defined urban areas. This choice serves two purposes: land use reflects planning outcomes and standardized classifications allow a consistent assessment.

The framework adopts a hierarchical system of spatial units that reflects the nested structure of urban systems and supports multi-scale analysis. The analysis adopts three nested units: eco-cell, eco-community, and eco-district to capture the multi-scalar nature of ecosystem services. While conceptually inspired by the Sino-Singapore Tianjin Eco-City [12], its use here is scientifically motivated by the need to represent urban systems as nested ecological and functional entities. This design reflects principles of landscape ecology and urban systems theory, where processes are scale-dependent yet interconnected. It enables the quantification of carbon dynamics across multiple spatial levels, supporting both local analysis and broader policy assessment.

The Reggio Calabria Functional Urban Area (FUA) in Southern Italy was selected as the case study within the ECO-SET (ECOsystem SErvices for cities in Transition) project. The project aims to develop multidisciplinary approaches for integrating ecosystem services into urban planning processes. Reggio Calabria was chosen as a representative Mediterranean medium-sized city characterized by fragmented urban growth, heterogeneous land uses, and high exposure to climate-related risks.

Within the broader ECO-SET project, this paper contributes to one of two complementary research lines. The first focuses on developing structural and functional models of ecosystem services through hierarchical and network-based representations of urban systems [13]. The second research line, of which the present study is part, advances a quantitative, spatially explicit assessment of carbon-related ecosystem services using harmonized land-use data and spatial econometric models. A central feature of this line is the identification of spatial spillover effects, which reveal how emissions and sequestration dynamics propagate across adjacent areas. By quantifying these cross-boundary externalities, the analysis not only improves the understanding of urban carbon processes but also provides a scientific basis for compensatory planning measures, enabling targeted interventions and the balancing of environmental impacts within the FUA. Together, these two lines form the conceptual and analytical foundation of ECO-SET, combining ecosystem service theory with operational tools for planning and governance.

This research provides a spatially integrated estimate of carbon performance. Land-use classes were derived from the 2018 Urban Atlas, which was used to characterize spatial development patterns due to its standardized classification and wide coverage across European urban areas. However, its minimum mapping unit (0.25 ha for urban classes and 1 ha for rural classes) limits its suitability for analysing ecosystem services at functional scales.

The analysis operationalizes this hierarchical system through two functional scales: eco-cells (400 × 400 m) and eco-districts (1600 × 1600 m) to capture both local carbon dynamics and broader urban patterns. This approach, adapted from multi-scale planning frameworks, ensures analytical precision while maintaining planning relevance.

Building on this spatial structure, the research proposes a replicable method for integrating urban ecosystem services (UES) into urban planning through spatially explicit and data-driven analysis.

Accordingly, the study examines how multi-scale, land-use-based spatial analysis can quantify and interpret urban CO₂ emissions and sequestration to support ecosystem-based and mitigation-oriented urban planning.

The study pursues three objectives:

• First, the study adopts a dual-scale spatial analysis, fine-grained eco-cells to broader eco-districts, as the analytical foundation for examining carbon dynamics. Within this structure, standardized emission and sequestration coefficients are applied to each land-use class, enabling consistent quantification of urban carbon performance;
• Analyse spatial patterns, hotspots, and cross-boundary spillovers;
• Assess the planning implications of this dual-scale approach, highlighting how UES can inform zoning and climate-responsive strategies.

The paper is structured as follows: Section 2 outlines the conceptual and methodological framework; Section 3 describes data and methods; Section 4 presents the findings; Section 5 discusses planning implications; and Section 6 concludes with implications and future directions.

2. Background and Conceptual Framework

Cities face growing pressures from climate change, environmental degradation, and socio-economic disparities, making it essential to reorient urban planning around environmental limits and regenerative potential. The Natural Capital Approach provides a framework for recognizing how natural assets support human well-being, resilience, and long-term sustainability. Linking ecological processes with economic and social systems helps integrate biodiversity, ecosystem function, and service flows into planning and land-use decisions [14,15].

Within this framework, ecosystem services improve the supply of goods and services necessary for societal well-being by enhancing adaptability to current and future climate risks, reducing ecological footprints, and supporting resilience, health, and quality of life.

To implement the principles of the Natural Capital Approach effectively, ecosystem-based approaches have become central to policy and planning. They apply ecological functions and natural resources to address societal challenges such as climate adaptation, disaster risk reduction, and resource efficiency. By leveraging ecosystems’ inherent resilience rather than relying solely on engineered solutions, this approach aligns the preservation of ecological integrity with human development goals.

In urban planning, ecosystem-based approaches promote incorporating natural elements into built environments and enhancing access to ecosystem services. They also support a more geographically and service-oriented land-use strategies that influence ecological outcomes. Within this context, UES translate broader ecosystem processes into tangible benefits [16].

UES represent the benefits derived from ecological functions within cities, ranging from air purification and stormwater regulation to carbon sequestration and recreational opportunities [17]. In high-density areas, UES provide measurable contributions to public health, environmental quality, and climate change adaptation [18]. Bolund and Hunhammar (1999) first conceptualized ecosystem services in an urban context, defining locally produced services like air filtration, microclimate regulation, and recreational space provision [19]. The authors emphasized the importance of planners and policymakers valuing and appreciating such services, thereby laying the groundwork for incorporating ecosystem functions into land-use planning in urban areas. Despite their significance, UES remains underrepresented in planning frameworks due to limited spatial integration [7,8].

A key challenge is the lack of analytical tools that can effectively connect UES with land-use data in a spatially explicit and operationally relevant way for urban planners. Traditional urban planning, operating within rigid administrative boundaries, often fails to capture the cross-boundary ecological flows and spillover effects of urban processes. This spatial mismatch between data availability and planning interventions limits the ability of local authorities to make informed decisions about climate mitigation.

To integrate ecosystem services into practice, several classification systems have been established, with the most popular being the Common International Classification of Ecosystem Services (CICES). CICES provides a detailed typology that categorizes services as provisioning, regulating, and cultural. Spatial planning must also account for dynamic urban boundaries and the increasing importance of functional spatial units beyond administrative definitions.

In this study, ecosystem services were classified through a comprehensive review of scientific literature, European policy documents, the Italian Institute for Environmental Protection and Research (ISPRA) reports, and the Common International Classification of Ecosystem Services (CICES) framework. The objective was to combine CICES international standards with ISPRA’s regional expertise to create an integrated and context-specific classification applicable to the Italian urban landscape. This harmonized framework supports the identification of functional criteria for ecosystem services, linking European directives with local environmental and planning contexts. The resulting classification provides planners and policymakers with a practical tool to evaluate ecosystem health, sustainability, and resilience within different urban and regional settings (Figure 1).

Among the categories of ecosystem services, regulating services play a critical role in mitigating the effects of urbanization. These include air purification, climate regulation, and carbon sequestration, which are particularly urgent in rapidly urbanizing areas where natural ecosystems face increasing pressure [20,21,22].

Over the past decade, the integration of UES into urban planning has increasingly emphasized quantitative assessment and spatial integration [23,24]. While earlier studies were mainly conceptual or qualitative, recent research demonstrates how UES can inform zoning, infrastructure development, and green space design [25,26]. Nevertheless, challenges persist, particularly limited data availability and the absence of standardized methods for assessing ecosystem functioning in urban systems [27].

Because ecosystem service distribution is connected to land-use structure, urban form, density, and land cover types influence the quantity and quality of services produced [28,29]. Green areas such as urban forests and parks contribute high regulating and cultural services, whereas dense built-up areas often function as ecological stressors.

Beyond ecosystem service provision, land use exerts a dual influence on urban carbon dynamics [30,31]. It both reflects and shapes patterns of energy consumption, mobility, and ecological capacity. Consequently, land-use decisions influence the spatial pattern of carbon sources and sinks and, therefore, are critical to emission-reduction efforts. Yet land use is rarely assessed in functional terms, in terms of its environmental performance, despite being one of the most direct instruments available to urban planners [32].

Urban land-use categories vary widely in emission intensity. Residential density, industrial areas, and transport infrastructure are commonly associated with high rates of emissions, due to their direct and indirect energy use [33,34]. However, this relationship is not linear; dense mixed-use areas can have lower per-capita emissions than low-density sprawl [35,36]. This complexity demands a land-use-based emission modelling approach, calculating emission intensity per unit area according to land-use function and form [37].

To support such analysis, this study incorporates functional criteria for ecosystem services, translating broad ecosystem service categories, such as provisioning, regulating, cultural, and supporting services, into measurable, location-specific indicators. Each criterion is designed to capture how ecosystem functions respond to local environmental pressures and land-use patterns. Focusing on regulating services linked to vegetation cover and filtration efficiency, this approach links ecological performance with spatial planning, helping identify areas where ecosystem functions are most critical for urban sustainability.

To operationalize this framework, the Urban Atlas dataset developed by the European Environment Agency under the Copernicus Land Monitoring Service was used as a harmonized land-use baseline. Its standardized classification for over 700 FUAs enables consistent comparison and replication [38]. Previous studies have successfully applied Urban Atlas data to UES mapping, urban sprawl assessment, and green infrastructure planning [39,40,41,42]. Its fine spatial resolution makes it particularly suitable for modelling carbon emissions and sequestration across multiple scales.

The hierarchical spatial framework adopted in this study comprises three nested units: eco-cell, eco-community, and eco-district (see Figure 2). This three-tier structure was originally developed to promote mixed-use, compact, and accessible urban environments within a 15-min walking radius in Sino–Singapore Tianjin Eco-City [12].

The eco-cell represents the smallest unit of analysis and functions as the building block of the framework. Comprising individual parcels such as small parks, street segments, or building plots, eco-cells capture micro-level ecological capacities and reflect the diversity of urban functions at fine spatial resolution. This scale highlights localized ecosystem service processes such as carbon sequestration, biodiversity support, pollutant filtration, stormwater regulation, and recreational accessibility and illustrates how targeted interventions at the parcel level can reinforce broader urban sustainability objectives.

While the full hierarchy in Singapore included the intermediate eco-community scale, this level was not applied in the present analysis. Preliminary tests showed that aggregating eco-cells (400 × 400 m) to eco-communities (800 × 800 m) produced results highly correlated with the eco-district scale (1600 × 1600 m), offering limited additional insight while increasing computational demand. For this reason, the analysis focused on two scales, eco-cells for fine-grained patterns and eco-districts for strategic interpretation, while retaining the three-tier framework conceptually to ensure scalability and comparability in future studies.

At the broadest scale, the eco-district integrates clusters of eco-communities into strategic zones aligned with planning and governance. It enables assessment of aggregated carbon balances by combining the sequestration potential of natural areas with emissions from built-up and infrastructural land uses. This scale supports spatial strategies such as ecological zoning, compensation mechanisms, and large-scale green infrastructure planning. In doing so, it bridges neighbourhood dynamics with policy-oriented frameworks for urban climate governance. Conceptually, the framework can be envisioned as a mapping process: the eco-cell functions as a pixel in a high-resolution image, while the eco-district represents a cluster of these pixels forming a coherent picture of urban ecological performance. This analogy highlights the framework’s ability to bridge fine-grained analysis with strategic urban planning.

Together, these scales form a cohesive framework linking fine-grained ecological processes with higher-level planning and governance. This nested approach reduces the spatial mismatch common in environmental assessments, where coarse emission data contrast with the fine scale required for planning. Integrating ecosystem service assessment with standardized land-use data provides a replicable, data-driven foundation for climate-responsive urban planning.

The following sections apply this approach to Reggio Calabria FUA to assess the spatial pattern of CO₂ emissions and identify planning priorities aligned with ecosystem-based, mitigation-oriented interventions.

3. Methodology

This study adopts a multi-scale, data-driven framework designed to quantify urban CO₂ emissions and sequestration and examine their spatial dependencies within Reggio Calabria FUA. The methodology integrates geospatial data processing, parametric emission modeling, and spatial econometric analysis into a coherent workflow. The process begins with the acquiring and harmonizing Copernicus Urban Atlas land-use data, which form the spatial basis for defining analytical units. CO₂ emission and sequestration coefficients are then assigned to each land-use class using parameters from national and international datasets and literature. These values are spatially aggregated through a hierarchical grid system composed of eco-cells (400 × 400 m) and eco-districts (1600 × 1600 m), capturing both localized and citywide carbon dynamics. Finally, a Spatial Durbin Model (SDM) evaluates direct and spillover effects of land-use categories on CO₂ emissions, producing a spatially explicit understanding of urban carbon interdependence.

Figure 3 summarizes the methodological workflow, from data collection to modelling and spatial interpretation.

3.1. Description of Workflow Steps

The methodological workflow shown in Figure 3 consists of five main steps:

• Land-use data preparation: Urban Atlas data are harmonized for the Reggio Calabria FUA, and land-use classes are mapped to the 400 × 400 m eco-cell grid (Section 3.2).
• Compilation of CO₂ coefficients: Emission and sequestration parameters are compiled from institutional datasets and literature, converted to kg CO₂ m⁻² yr⁻¹, and assigned to each land-use class (Section 3.3).
• Cell-level emission and sequestration calculation: For each eco-cell, land-use shares are multiplied by the corresponding CO₂ coefficients to obtain total emissions, sequestration, and net balance (Section 3.4.1).
• Spatial econometric modelling (SDM): SDM quantifies both direct and spillover effects of land-use shares on CO₂ emissions using a 5-nearest-neighbour spatial weights matrix (Section 3.4.2).
• Aggregation to eco-districts: Eco-cell values are aggregated into 1600 × 1600 m eco-districts to support planning interpretation and zoning analysis (Section 3.4.3).

The selection of the spatial scales (400 m and 1600 m) and their planning relevance is detailed in Section 3.3.4.

3.2. The Study Area

The study was conducted in the Reggio Calabria FUA. The FUA is located at the southernmost tip of Calabria region, Italy (see Figure 4). It extends from the coastal plain toward the hilly and mountainous areas of Aspromonte, encompassing dense coastal settlements with extensive natural and agricultural hinterlands (Figure 5). The area hosts approximately 550,000 inhabitants distributed across Reggio Calabria’s urban core and several smaller municipalities. Its urban structure reflects typical Mediterranean patterns: compact coastal centers, expanding suburban belts, and fragmented peri-urban zones. The economy is largely service-based, with limited industrial activity and growing logistics functions linked to the port, airport, and transport corridors. These features make Reggio Calabria FUA a meaningful case for studying carbon dynamics and functional urban areas in Southern Europe. Its combination of dense urban fabric, transitional rural zones, and ecological areas provides a representative setting to test how spatial structure and land-use organization influence urban carbon balance.

3.3. Compilation of Land Use–Based CO₂ Emission and Sequestration Parameters

Before assigning coefficients, all emission and sequestration values collected from institutional datasets and literature were harmonized to a common unit (kg CO₂ m⁻² yr⁻¹). When sources reported a value range, we selected either (i) the value most commonly used in European or Italian studies or (ii) a central value to avoid over- or underestimation. In cases where multiple studies existed, preference was given to those with similar climatic and land-use conditions to Southern Italy. The final coefficients used in the calculations are reported in Table 1. Additionally, Supplementary Material S1 provides the full description of the sources and rationale for the CO₂ parameters assigned to each land-use type.

The project utilizes land-use data obtained from Urban Atlas, which classifies the area into 29 land-use types (Figure 6). For the Reggio Calabria FUA, only 25 categories were available; two were labeled No Data, and Orchards and Wetlands were not present in the area.

Because land-use types differ in ecological function, standardized parameters were selected to ensure consistency in calculations. We compiled emission and sequestration factors from literature and official databases, selecting values already expressed per square meter (m²) (

Table 2. Descriptive statistics of total carbon emissions, sequestration, and net emissions per 400 × 400 m eco-cell in Reggio Calabria FUA (values in kg CO₂ yr⁻¹).

Metric	Min	Max	Mean	Median	Std Dev
Total CO2 Emissions (kg)	81.69	29,082,155	593,813.8	128,104.03	1,454,353.68
Total CO2 Sequestration (kg)	−4573.84	2,768,441	−12,405.2	−4527.59	108,490.38
Net CO2 (kg)	−4451.58	28,132,719	581,408.6	123,558.36	1,413,734.35

). This approach enables comparable calculations across land uses, improving analysis of their relative emissions and sequestration contributions. These values denote annual CO₂ emissions (positive) or sequestration (negative) per square meter. The following subsections summarize the adopted parameters, outlining their sources and the rationale behind their selection.

3.3.1. Artificial Areas

Residential emissions were calculated using parameters from the Italian Information System on Energy Performance Certificates (Sistema Informativo sugli Attestati di Prestazione Energetica, SIAPE), managed by ENEA [44]. SIAPE offers an overview of the Italian building energy efficiency.

Urban Atlas residential land-use includes six density-based subclasses distinguished by their density and surface sealing. Emission factors were applied to the whole surface of each residential class and weighted by predefined percentages representing built-up coverages. The base emission coefficient for urban fabric (37.5 kg CO₂ m⁻² yr⁻¹), derived from ENEA, represents the mean emission intensity of highly sealed residential surfaces (>80% imperviousness).

To reflect differences in built density, the coefficient was adjusted according to the impervious surface fractions defined in the Urban Atlas. Accordingly, percentage adjustment factors were introduced to distinguish between the residential subclasses:

• ➢ Continuous Urban Fabric (S.L. > 80%) → × 100%
• ➢ Discontinuous Dense Urban Fabric (S.L. 50–80%) → × 65%
• ➢ Discontinuous Medium-Density Urban Fabric (S.L. 30–50%) → × 40%
• ➢ Discontinuous Low-Density Urban Fabric (S.L. 10–30%) → × 20%
• ➢ Discontinuous Very Low-Density Urban Fabric (S.L. < 10%) → × 5%
• ➢ Isolated Structures → × 1%

These percentage factors weight the base coefficient according to the impervious fraction of each residential subclass.

These adjustment factors were reported in the CityChem Supplementary Material [45] and reflect the relative impervious surface of each subclass.

In this way, the emissions are projected in line with the level of urbanization, recognizing that high-density residential districts make a greater contribution to CO₂ emissions because of higher energy usage, whereas low-density districts have a relatively lower contribution.

For land use classified as Industrial, Commercial, Public, Military & Private Units in Urban Atlas dataset, distinguishing subcategories was not feasible; therefore, SIAPE’s non-residential emission factor (62.8 kg CO₂ m⁻² yr⁻¹) was applied.

Before adopting this parameter, industrial activity was assessed using the Industrial Emissions Directive and the European Pollutant Release and Transfer Register Regulation dataset [46], and OpenStreetMap (OSM) data, confirming the absence of large-scale industrial facilities in the FUA. OSM identified smaller commercial and industrial sites, but their heterogeneity prevented separate treatment. Consequently, a single coefficient was applied to represent the heterogeneous but generally low-intensity nature of non-residential urban land uses.

For Fast Transit Roads and Other Roads, no normalized surface-based emission factors exist. Therefore, regional transport emissions were downscaled to the FUA and allocated proportionally to road surface area.

Railways used the standard emission coefficient of 50 tCO₂ m⁻² yr⁻¹ from the International Union of Railways (UIC) Carbon Footprint of Railway Infrastructure report [47].

Port emissions were derived from site-specific studies of the Reggio Calabria and Villa San Giovanni ports [48,49], converted from CO₂eq to CO₂ using IPCC factors.

For Reggio Calabria Airport, the emissions were calculated using actual flight schedules, aircraft types, passenger loads, and route distances, totalling approximately 72,104 tCO₂ per year.

Landfill emissions used factors from Case Passerini and Giugliano studies [50], converted from daily to annual values, yielding 0.88–2.08 kg CO₂ m⁻² yr⁻¹.

For Construction Sites, a coefficient of 9.31 kg CO₂ m⁻² yr⁻¹ was applied, based on the study Carbon Emissions of Construction Processes on Urban Construction Sites [51], which quantified emissions from on-site operations, material transport, and machinery use.

CO₂ emissions for Sports and Leisure Facilities land use category were estimated using SIAPE-derived parameter of 75.9 kg CO₂ m⁻² yr⁻¹.

3.3.2. Agricultural Areas

For Green Urban Areas land use, a parameter of −1.06 kg CO₂ m⁻² yr⁻¹ has been derived from the report “Quantifying the Greenhouse Gas Benefits of Urban Parks” [52].

Agricultural coefficients were based on European Commission reports on agricultural soil carbon dynamics [53]. Arable land ranged between −0.1 and −0.3 kg CO₂ m⁻² yr⁻¹; permanent crops (vineyards, olive groves) ranged between −0.22 and −2.17 kg CO₂ m⁻² yr⁻¹ [54,55]. Pastures were −0.2 to −0.7, and complex cultivation systems −0.15 to −0.5 kg CO₂ m⁻² yr⁻¹.

3.3.3. Natural Areas

Broad-leaved, coniferous, and mixed forests sequester −1.72, −0.99, and −1.36 kg CO₂ m⁻² yr⁻¹, respectively; an average of −1.36 kg CO₂ m⁻² yr⁻¹ was applied [56,57,58]. Shrublands and herbaceous vegetation were assigned −2.2 kg CO₂ m⁻² yr⁻¹ [59]. Open spaces and wetlands were excluded due to negligible presence in the FUA.

3.3.4. Other Types

For water bodies, fluxes ranged from −0.007 kg CO₂ m⁻² yr⁻¹ (coastal waters) to −0.65 kg CO₂ m⁻² yr⁻¹ (rivers) [60,61,62].

Detailed derivations of the parameters and their literature sources are provided in Supplementary Material S1, while Table 1 summarizes all land-use categories and their CO₂ values.

Table 1. Urban Atlas land use classification, including corresponding CO₂ emission and sequestration parameters.

Code	Legend	Classification Name	CO2 Emissions (kg CO2 m−2 yr−1)	Source
Artificial Areas
11100		Continuous urban fabric (S.L. > 80%)	37.5 kg CO2 m−2 yr−1 × 100%	[44]
11210		Discontinuous dense urban fabric (S.L. 50–80%)	37.5 kg CO2 m−2 yr−1 × 65%
11220		Discontinuous medium density urban fabric (S.L. 30–50%)	37.5 kg CO2 m−2 yr−1 × 40%
11230		Discontinuous low dens. Urban fabric (S.L. 10–30%)	37.5 kg CO2 m−2 yr−1 × 20%
11240		Discontinuous very low-density urban fabric (S.L. < 10%)	37.5 kg CO2 m−2 yr−1 × 5%
11300		Isolated structures	37.5 kg CO2 m−2 yr−1 × 1%
12100		Industrial, commercial, public, military, and private units	62.8 kg CO2 m−2 yr−1
12210		Fast transit roads and associated land	Regional road transport CO2 emissions downscaled to Reggio Calabria FUA, distributed per capita, then allocated to roads based on area.
12220		Other roads and associated land
12230		Railways and associated land	50 tCO2 m−2 yr−1	[47]
12300		Port Areas	Villa San Giovanni port emissions ~10,640 tCO2/yr Reggio Calabria port emissions ~950 tCO2/yr	[48]
12400		Airports	72,104 tCO2/yr	Author’s calculations based on data retrieved from the airport’s official website (flight schedules and destinations), Flightradar24 (air traffic monitoring), and emission factors by destination/travel distance.
13100		Mineral extraction and dump sites	Landfills: 0.88–2.08 kg CO2 m−2 yr−1	There are no active mining activities in Reggio Calabria [63,64]; therefore, the parameter applies only to landfills.
13300		Construction sites	9.31 kg CO2 m−2	[51]
13400		Land without current use	0
14100		Green urban areas	−1.06 kg CO2 m−2 yr−1	[65]
14200		Sports and leisure facilities	75.9 kg CO2 m−2 yr−1	[44]
Agricultural Areas
21000		Arable land (annual crops)	−0.1 to −0.3 kg CO2 m−2 yr−1	[66]
22000		Permanent crops	−0.22 to −2.17 kg CO2 m−2 yr−1	[54,55]
23000		Pastures	−0.2 to −0.7 kg CO2 m−2 yr−1	[66]
24000		Complex and mixed cultivation	−0.15 to −0.5 kg CO2 m−2 yr−1
25000		Orchards	There are no Orchards in Reggio Calabria FUA
Natural Areas
31000		Forests	−1.36 kg CO2 m−2 yr−1	[56,57,58]
32000		Herbaceous vegetation associations	−0.22 kg CO2 m−2 yr−1	[59]
33000		Open spaces with little or no vegetation	0
Other Types
40000		Wetlands	There are no wetlands in Reggio Calabria FUA.
50000		Water	Beaches: −0.007 kg CO2 m−2 yr−1 Small lakes & ponds: −0.52 kg CO2 m−2 yr−1 Rivers: −0.65 kg CO2 m−2 yr−1	[60,61,62]
91000		No data	0
92000

Table 1. Urban Atlas land use classification, including corresponding CO₂ emission and sequestration parameters.

Code	Legend	Classification Name	CO2 Emissions (kg CO2 m−2 yr−1)	Source
Artificial Areas
11100		Continuous urban fabric (S.L. > 80%)	37.5 kg CO2 m−2 yr−1 × 100%	[44]
11210		Discontinuous dense urban fabric (S.L. 50–80%)	37.5 kg CO2 m−2 yr−1 × 65%
11220		Discontinuous medium density urban fabric (S.L. 30–50%)	37.5 kg CO2 m−2 yr−1 × 40%
11230		Discontinuous low dens. Urban fabric (S.L. 10–30%)	37.5 kg CO2 m−2 yr−1 × 20%
11240		Discontinuous very low-density urban fabric (S.L. < 10%)	37.5 kg CO2 m−2 yr−1 × 5%
11300		Isolated structures	37.5 kg CO2 m−2 yr−1 × 1%
12100		Industrial, commercial, public, military, and private units	62.8 kg CO2 m−2 yr−1
12210		Fast transit roads and associated land	Regional road transport CO2 emissions downscaled to Reggio Calabria FUA, distributed per capita, then allocated to roads based on area.
12220		Other roads and associated land
12230		Railways and associated land	50 tCO2 m−2 yr−1	[47]
12300		Port Areas	Villa San Giovanni port emissions ~10,640 tCO2/yr Reggio Calabria port emissions ~950 tCO2/yr	[48]
12400		Airports	72,104 tCO2/yr	Author’s calculations based on data retrieved from the airport’s official website (flight schedules and destinations), Flightradar24 (air traffic monitoring), and emission factors by destination/travel distance.
13100		Mineral extraction and dump sites	Landfills: 0.88–2.08 kg CO2 m−2 yr−1	There are no active mining activities in Reggio Calabria [63,64]; therefore, the parameter applies only to landfills.
13300		Construction sites	9.31 kg CO2 m−2	[51]
13400		Land without current use	0
14100		Green urban areas	−1.06 kg CO2 m−2 yr−1	[65]
14200		Sports and leisure facilities	75.9 kg CO2 m−2 yr−1	[44]
Agricultural Areas
21000		Arable land (annual crops)	−0.1 to −0.3 kg CO2 m−2 yr−1	[66]
22000		Permanent crops	−0.22 to −2.17 kg CO2 m−2 yr−1	[54,55]
23000		Pastures	−0.2 to −0.7 kg CO2 m−2 yr−1	[66]
24000		Complex and mixed cultivation	−0.15 to −0.5 kg CO2 m−2 yr−1
25000		Orchards	There are no Orchards in Reggio Calabria FUA
Natural Areas
31000		Forests	−1.36 kg CO2 m−2 yr−1	[56,57,58]
32000		Herbaceous vegetation associations	−0.22 kg CO2 m−2 yr−1	[59]
33000		Open spaces with little or no vegetation	0
Other Types
40000		Wetlands	There are no wetlands in Reggio Calabria FUA.
50000		Water	Beaches: −0.007 kg CO2 m−2 yr−1 Small lakes & ponds: −0.52 kg CO2 m−2 yr−1 Rivers: −0.65 kg CO2 m−2 yr−1	[60,61,62]
91000		No data	0
92000

To conduct fine-grained yet scalable analysis, two of the three spatial units were used:

• Eco-cell: A 400 × 400 m spatial grid used as the base unit for land use classification and carbon estimation.
• Eco-district: A 1600 × 1600 m grid formed by aggregating 16 eco-cells, used to summarize and visualize regional trends.

The intermediate eco-community level (800 × 800 m) was conceptually retained but excluded from the analysis, as preliminary tests showed strong correlation with the eco-district scale and limited additional insight. Urban Atlas data were harmonized to 400 m resolution, assigned to eco-cells and then aggregated into eco-districts for regional comparison and planning interpretation.

3.4. Analytical Framework

3.4.1. CO₂ Emission and Sequestration Calculation

This analytical framework integrates carbon accounting with spatial econometric modeling. Each eco-cell and eco-district was assigned CO₂ values per square meter using land-use-based emission and sequestration parameters. Combined with SDM, these emission surfaces allow quantification of direct (local) and spillover (neighboring) effects of land uses on carbon outcomes, supporting a holistic interpretation of carbon dynamics. The SDM was selected for its ability to model spatial dependence and feedback, which are common in urban ecological systems. This setup captures how land-use patterns in one cell influence both local emissions and those of neighboring areas.

Net annual carbon outcomes were estimated for each eco-cell using a parametric approach based on land-use categories. The following formula was applied:

CO_2nxa = ∑(Area_n × Coef_n)

(1)

where:

• Area_n: Area of land-use category i in square meters;
• Coef_n: Annual CO₂ coefficient (kg/m²) for category i.

Each of the 3188 eco-cells was assigned land-use shares that were multiplied by the standardized emission/sequestration coefficients to calculate net carbon outcomes.

3.4.2. Spatial Durbin Model (SDM)

To analyze the spatial influence of land uses on emissions, a Spatial Durbin Model was used:

CO_2nxa = α + βX + γWX + λWCO₂ + ε

(2)

where:

• X: Local predictors (land use shares, population);
• WX: Spatial lags of predictors;
• WCO₂: Spatial lag of the dependent variable;
• λ: spillover coefficient;
• W: Spatial weights matrix (5-nearest neighbors).

The spatial-weights matrix was constructed as a row-standardized 5-nearest-neighbour matrix using eco-cell centroids.

The SDM was chosen over alternative models (e.g., Spatial Lag or Spatial Error Models) because it captures both direct and spillover effects. This is especially important in urban environments, where land-use processes generate cross-boundary externalities [67].

In the context of spatial econometrics [68,69], spatial spillover effects refer to the indirect impacts that originate in one spatial unit and propagate to others due to spatial dependence in the data. They capture the extent to which changes in an explanatory variable in one location affect the dependent variable in neighboring locations, either through physical proximity, functional connectivity, or shared socio-environmental processes. Within SDM, these effects arise from the inclusion of spatially lagged independent variables and represent the mechanism through which local phenomena generate broader spatial feedback.

Model parameters were estimated using the GMM Combo estimator (Generalized Method of Moments—Combination estimator), which is suitable for spatial models with endogenous regressors and spatial autocorrelation. GMM Combo combines fixed effects and instrumental variable approaches to enhance efficiency and minimize bias.

To ensure model robustness, several validation procedures were carried out. First, Moran’s I test was applied to the model residuals, confirming the presence of significant spatial autocorrelation and thereby justifying the use of spatial econometric techniques. Second, SDM was compared with alternative specifications, the Spatial Lag Model (SLM) and the Spatial Error Model (SEM), using Akaike Information Criterion (AIC) and log-likelihood values. The SDM consistently outperformed these alternatives, indicating a better overall model fit. Finally, sensitivity analyses were conducted on both the emission coefficients and the spatial resolution of the grid system. These tests demonstrated that the model remained stable in its identification of emission hotspots and significant land-use predictors.

3.4.3. Aggregation to Eco-Districts

Eco-cell results were aggregated into larger eco-districts (1600 × 1600 m) to translate micro-level findings into planning-scale units.

Aggregated emissions were calculated as:

ΔCO_2x = CO_2x(before) − CO_2x(after)

(3)

This aggregation facilitates the identification of priority zones for emissions mitigation, conservation strategies, and land use reallocation and facilitates integration with existing planning frameworks.

3.4.4. Justification for Spatial Scale Selection

The 400 × 400 m eco-cell and 1600 × 1600 m eco-districts were selected based on planning relevance, data-resolution constraints, and precedents from sustainable-city frameworks. The 400 m eco-cell corresponds to eight times the Urban Atlas minimum mapping unit (50 × 50 m), preserving land-use heterogeneity without producing excessive fragmentation. It also approximates the neighborhood block and a five-minute walking radius commonly used in urban planning.

The 1600 m eco-district scale aggregates 16 eco-cells and represents a mid-scale unit suitable for zoning, infrastructure planning, and emission clustering analysis. A 1600 m span reflects the scale of urban districts used in European emissions inventories and planning assessments. It is large enough to incorporate diverse land-use combinations yet compact enough to identify actionable regional patterns.

An intermediate 800 × 800 m scale was considered but excluded because it offered limited additional interpretive value between the fine and coarse resolutions. The selected scales provide a practical balance between spatial granularity and analytical readability. SDM further mitigates potential aggregation effects by accounting for spillovers across units. Together, these scales offer meaningful resolution for both local diagnostics and district-level planning.

4. Results

4.1. Eco-Cell Analysis (400 × 400 m)

The eco-cell analysis (400 × 400 m) reveals significant heterogeneity in carbon dynamics across the Reggio Calabria FUA. Each cell was evaluated for annual CO₂ emissions, sequestration, and net carbon balance based on land use composition and standardized carbon coefficients (Figure 7a–c). Table 2 summarizes descriptive statistics for total emissions, sequestration, and net CO₂ per eco-cell.

Total CO₂ emissions per eco-cell range from 81.69 kg in peripheral zones to over 29 million kg in dense urban cores. CO₂ sequestration values are consistently negative but much smaller, reflecting the limited distribution of green and agricultural land uses. The average net CO₂ per eco-cell is 581,408.6 kg, underscoring the region’s dominant status as a net emitter.

This disparity points to the critical role of landscape planning in achieving regional emission mitigation. Although natural and semi-natural land uses offer modest per-cell sequestration, their spatial distribution provides an essential buffer against high-emission zones. Targeted greening and land-use interventions in these hotspots could significantly reduce Reggio Calabria’s overall carbon footprint.

Population density was integrated at the eco-cell level to contextualize emission drivers.

Table 3. Descriptive statistics of population distribution across 400 × 400-m eco-cells in the Reggio Calabria FUA. These values provide context for assessing the influence of population density on spatial CO₂ emission patterns.

Metric	Value
Min Population	0
Max Population	621
Mean Population	92.3
Median Population	64
Std. Deviation	115.7
Total Eco-Cells	3188

summarizes population distribution across the grid, supporting subsequent SDM interpretation.

Figure 7 illustrates the spatial distribution of CO₂ dynamics across the FUA at the eco-cell scale. Figure 7a shows the highest emission hotspots along the coastal corridor. Figure 7b shows the highest sequestration in the northern and eastern parts of the FUA, corresponding to forested and agricultural zones. These areas contribute to regulating ecosystem services and help mitigate overall emissions. Figure 7c reveals a clear spatial asymmetry: coastal and central eco-cells act as net emitters, while inland and peri-urban eco-cells perform as carbon sinks. This spatial differentiation establishes the empirical foundation for the multi-scale analysis that follows.

4.2. Spatial Durbin Model Results

The SDM was used to identify direct and spillover effects of land use and population on CO₂ emissions at the eco-cell scale.

Table 4. The results of SDM estimate the relationship between land-use categories and net CO₂ emissions across 3188 eco-cells in the Reggio Calabria FUA. The coefficients illustrate the change in net CO₂ emissions (kg) associated with an area unit increase in land use area (or population-adjusted road categories), while accounting for spatial spillover effects. p < 0.05 = significant.

Land-Use Category	Coefficient (kg CO2 m−2 yr−1)	p-Value	Cell ID	% of Total Area	Interpretation
Continuous Urban Fabric (>80%)	−0.309	0.000	487	10.5%	Compact form reduces emissions
Discontinuous Dense Urban (50–80%)	+0.157	0.000	732	14.8%	Dense sprawl increases emissions
Discontinuous Medium Urban (30–50%)	+0.197	0.0002	489	9.7%	Moderate sprawl effect
Discontinuous Low-Density Urban (10–30%)	+0.782	0.000	222	4.4%	Heavy contributor due to sprawl
Other Roads and Associated Land	+3.482	0.026	119	2.3%	Strong transport-related emissions
Industrial/Commercial Units	+0.0180	0.061	98	3.8%	Moderate impact; interpret with care
Isolated Structures 1	−58.982	0.000	3	0.3%	Model anomaly: minimal sample
Forests	0.0	1.000	95	6.2%	No effect detected, likely due to low variation
Herbaceous Vegetation	−0.00010	0.9992	64	4.1%	Not detectable effect
Green Urban Areas	+0.0187	0.986	71	1.9%	Weak correlation with emissions
Arable Land (Annual Crops)	−0.00680	0.992	103	5.5%	No measurable mitigation effect
Pastures	−0.00160	0.9958	59	3.4%	No significant impact
Permanent Crops	−0.00150	0.9958	37	2.7%	Not significant
Complex/Mixed Cultivation	−1.946	0.942	22	1.4%	Limited sample
Airports	−0.000003	0.9999	1	0.2%	No impact, tiny sample
Port Areas	−0.00038	0.996	2	0.4%	Insufficient representation
Mineral Extraction/Dump Sites	−0.00285	0.997	2	0.3%	Very sparse
Construction Sites	−0.00080	0.998	7	0.6%	Weak statistical signal

¹ Very small sample size (n = 3); result is not reliable and excluded from interpretation.

reports the model coefficients and statistical significance for all land-use categories.

The results indicate that urban form strongly shapes the spatial pattern of carbon emissions. Continuous urban fabric shows a significant negative relationship with emissions (−0.309 kg CO₂ m⁻² yr⁻¹, p < 0.001), suggesting potential efficiencies from shared infrastructure, shorter travel distances, and optimized energy use. In contracts, discontinuous low-density urban fabric is a strong positive contributor (+0.157 kg CO₂ m⁻² yr⁻¹, p < 0.001), reinforcing the emissions cost of urban sprawl.

Infrastructure-related land uses, such as other roads, show a significant positive effect (+3.482 kg CO₂ m⁻² yr⁻¹, p = 0.026), reflecting transport-related emissions. Industrial and commercial areas are marginally significant (p = 0.061), while isolated structures, appearing in only three cells (<0.1% of the study area), produce outlier results due to low spatial representation. The coefficient for Isolated Structures land use appears highly negative due to its presence in only three cells (0.1% of the study area) and is treated as an outlier without interpretation.

Most of the natural and semi-natural land uses, including forests, herbaceous vegetation, arable land, and pastures, did not yield significant results (p > 0.95), likely due to limited variability and peripheral clustering. Airports, ports, and construction areas show no significant effect, primarily because they appear in very few eco-cells (often only 1 or 2), limiting statistical power.

To understand how land-use categories and population density influence carbon outcomes, the SDM was applied to eco-cell data.

Table 5. SDM results for selected land-use categories and population. Coefficients represent direct and spatial spillover effects (kg CO₂ per m² per year) on net annual CO₂ emissions across 400 × 400 m eco-cells. All values are based on GMM Combo estimation with a 5-nearest neighbour spatial weights matrix. Statistically significant values (p < 0.05) are considered robust.

Land-Use Category	Direct Effect (ß)	Spillover Effect (W·X)	p-Value (Direct)	p-Value (Spillover)
Industrial units	2.35	1.14	0.001	0.02
Continuous urban fabric	1.88	0.97	0.005	0.04
Airports	3.42	1.89	0	0.001
Forests	−1.92	−0.82	0.002	0.05
Arable land	−1.35	−0.65	0.01	0.07
Herbaceous vegetation	−1.11	−0.52	0.015	0.08
Population 2	0.98	0.45	0.03	0.1

² Population is modeled as persons per eco-cell; effects are reported per m² by normalizing to the cell area (160,000 m²). Magnitudes reflect the marginal effect of adding residents within the fixed cell area.

summarizes SDM results for selected land-use categories and population density (kg CO₂ m⁻² yr⁻¹), showing expected changes in annual CO₂ emissions per additional square meter. It should be noted that Table 5 presents SDM impact measures, which account for spatial feedback and therefore differ from the coefficients in Table 4.

Table 5 provides insight into how land-use categories and population influence net CO₂ emissions on a per-square-meter, per-year basis. It reports the impact measures derived from SDM, representing the estimated marginal effects of each variable, including both direct and spatial spillover effects. These impact measures offer a more accurate interpretation than the raw coefficients in Table 4, as they account for spatial feedback and interdependence. Consequently, raw coefficients and impact measures may differ in sign or magnitude. For example, Continuous Urban Fabric has a negative coefficient in Table 4 but a positive direct impact in Table 5 (+1.88 kg CO₂ m⁻² yr⁻¹), indicating that spatial feedback results in net positive emissions. Similarly, an increase of one square meter in airport infrastructure corresponds to an additional 3.42 kg CO₂ m⁻² yr⁻¹ within that eco-cell, and 1.89 kg in neighboring cells.

Industrial land contributes 2.35 kg CO₂ m⁻² yr⁻¹ locally and 1.14 kg CO₂ m⁻² yr⁻¹ to surrounding cells, consistent with emissions patterns from manufacturing and heavy transport. Continuous Urban Fabric still shows positive emissions (+1.88 direct, +0.97 spillover), likely due to energy inefficiency in buildings and private vehicle usage.

In contrast, natural and agricultural land uses act as carbon sinks. One square meter of forest reduces emissions by 1.92 kg CO₂ m⁻² yr⁻¹, and arable land contributes a negative 1.35 kg CO₂ m⁻² yr⁻¹, with minor spillover effects.

Population density shows a moderate yet significant direct effect of +0.98 kg CO₂ m⁻² yr⁻¹, suggesting higher density increases energy use and mobility demand. Spillover effects are weak and not statistically robust, meaning population-driven emissions are confined mainly to the local context.

4.3. Eco-District Aggregation (1600 × 1600 m)

To stress environmental performance at a broader scale, CO₂ emissions and sequestration values from 400 × 400 m eco-cells were aggregated into 1600 × 1600 m eco-districts. Each eco-district was formed by grouping every 2 × 2 block of adjacent eco-cells, starting from the top-left (northwestern) corner of the study grid, into one larger unit. This yielded 232 eco-districts (1.6 km² each), ensuring consistent spatial coverage across the FUA.

Table 6. Summary statistics of CO₂ emissions, sequestration, and net CO₂ values across 232 eco-districts (1600 × 1600 m) in the Reggio Calabria FUA.

Metric	Min	Max	Mean	Median	Std. Deviation
Total CO2 Emissions (Million kg)	0.00	152.00	9.96	1.40	21.06
Total CO2 Sequestration (Million kg)	−4.32	−1.65	−1.95	−1.71	0.94
Net CO2 (Million kg)	−3.81	152.9	581,408.6	8.31	21.03

presents summary statistics of total CO₂ emissions, sequestration, and net carbon values per eco-district. The results show considerable spatial heterogeneity in carbon dynamics. On average, each district emits approximately 9.96 million kg of CO₂ per year, while sequestering around 1.95 million kg, leading to an average net emission of 8.31 million kg. However, the high standard deviation for net emissions (±21.03 million kg) reveals that while many districts emit modestly, others act as extreme hotspots.

Maximum emissions reach approximately 152 million kg per district, reflecting dense urban areas. Conversely, districts composed predominantly of natural land cover show near-zero emissions.

Sequestration ranges from −4.32 million kg to −1.65 million kg, producing net values from −3.81 to +152.9 million kg. This variation underscores the differing ecological roles of each district.

Figure 8 presents the spatial aggregation of emissions, sequestration, and the net CO₂ at the eco-district scale. Figure 8 shows the emissions concentrated along the coastal belt and in districts. Figure 8b illustrates that sequestration is highest in northern and eastern eco-districts characterized by forests and agricultural land. These zones partially offset emissions from the urban core, though sequestration capacity remains uneven. Figure 8c shows that only a few eco-districts act as net carbon sinks, while most central and coastal districts remain net emitters. Overall, these results suggest a spatial imbalance between emission sources and natural absorption capacity within the Functional Urban Area, emphasizing the importance of maintaining and expanding vegetated and low-intensity land uses in proximity to high-emission zones.

Eco-district results support scenario evaluation and planning optimization. For instance, high-emissions and low-sequestration districts are candidates for targeted land use change, green infrastructure, or stricter emission regulations. Conversely, districts showing substantial sequestration could serve as carbon sinks or conservation zones. Figure 8a–c reinforce these spatial patterns by visualizing emissions, sequestration, and net carbon performance.

The eco-district aggregation highlights spatial asymmetries between emission sources and sequestration zones, forming the basis for subsequent planning interpretation.

5. Discussion

This research contributes to urban carbon governance by demonstrating how a spatially disaggregated, land-use-based model can inform localized mitigation strategies. Drawing on ecosystem-based planning and natural capital frameworks, the findings confirm that land-use structure influences both emissions and sequestration. In the case of Reggio Calabria FUA, high-resolution modeling at the eco-cell level revealed that dense urban forms, particularly continuous urban fabric, produce lower net emissions than sprawling, discontinuous settlements. This supports existing consensus that compact forms promote more efficient energy use [35,36]. However, the benefits of density depend on spatial organization, ecological structures, and functional context. The results add nuance by showing that compactness may reduce emissions even in infrastructure-constrained or socioeconomically diverse areas.

In Reggio Calabria FUA, emissions and sequestration reflect the city’s linear coastal form and sharp topographic transition toward the Aspromonte foothills. High-emission hotspots coincide with the concentration of port-related logistics, transport corridors, and dense residential belts facing the Strait of Messina, while the inland areas characterized by forests, shrublands, and agricultural mosaics act as the main carbon sinks. This spatial duality underscores the imbalance between ecological capacity inland and urban pressure along the coast, a pattern typical of Mediterranean cities with morphology constrained.

The SDM results added a spatial econometric dimension, revealing both direct and spillover effects of land-use types on CO₂ outcomes. Industrial units, airports, and continuous urban fabric emerged as statistically significant positive contributors to emissions. Natural and semi-natural land uses showed negative effects, suggesting a carbon-sequestration function, although these were not statistically significant. This finding likely reflects the fragmented and peripheral distribution of Reggio Calabria’s natural areas, which, while ecologically valuable, have limited spatial influence on neighboring emission hotspots. Strengthening the connectivity between these zones and the urban core could therefore enhance their regulatory function.

Beyond their statistical significance, the identified spillover effects have important conceptual implications for urban planning. Spatial spillover effects capture the interconnectedness of urban systems, a cornerstone of contemporary planning theory. They show that actions or transformations in one area (e.g., densification, infrastructure development, land-use change) produce secondary effects in adjacent areas. These effects can be positive (e.g., improved accessibility, diffusion of green infrastructure benefits) or negative (e.g., pollution, congestion, displacement).

Recognizing spillovers helps planners understand the functional, not just administrative, structure of cities, how neighborhoods, districts, interact as part of a broader spatial network. This aligns with theories such as “Polycentric urban development and functional urban areas (ESPON, OECD), which emphasize flows and interdependencies rather than static boundaries. Landscape ecology and metropolitan governance both view the city as a spatial continuum where environmental processes and human activities propagate through networks.

The ECO-SET framework illustrates how spatial econometrics can support ecosystem-service compensation and balancing mechanisms. Spillover effects can highlight zones that benefit from or suffer due to changes elsewhere. For example, emissions from dense areas may diffuse into adjacent green or residential zones, which can be prioritized for mitigation, while positive spillovers from green areas can be integrated into local carbon accounting.

Nevertheless, these patterns align with the recognized role of vegetation in carbon cycling and suggest areas where reinforcing green infrastructure could enhance ecological performance. The spatial dependence observed in the SDM confirms that land use influences not only local ecological outcomes but also nearby areas, emphasizing the importance of considering cross-boundary interactions and functional linkages in urban planning [11,70]. Aggregating these dynamics to the eco-district level further revealed zones of ecological fragility and resilience, offering critical governance aligned with local zoning and regulatory intervention.

Spatial carbon studies in both European and non-European contexts further support the interpretation of spatial interdependence in CO₂ dynamics. Studies such as Xia et al. [71] in Zhejiang Province, China, and Sporkmann et al. [72] in Europe similarly found spatial autocorrelation in emissions and strong links between land-use intensity and carbon outcomes. However, the hierarchical framework applied here extends these insights by linking ecosystem services with functional urban zoning, providing a transferable analytical approach for cities with similar morphological and environmental conditions.

Furthermore, the spatial results reveal that Reggio Calabria’s most emission-intensive districts coincide with areas of economic activity and limited vegetation cover, while the most effective carbon sinks are confined to peripheral areas with low population density. This contrast highlights the need for integrated planning that combines densification with ecological restoration, particularly in peri-urban zones.

Examples such as IBA Hamburg climate-neutral district and the Vilnius regeneration strategy, and C40 Climate show how compact forms and coordinated governance reduce emissions at district scale [35,73]. Likewise, ecosystem-service valuation studies in cities such as Berlin, London, and Helsinki illustrate how linking land-use typologies with ecological performance offers a consistent framework for assessing regulating services such as carbon sequestration [74]. Collectively, these examples show that the spatial configuration of land use, alongside technological measures, plays a significant role in shaping urban carbon dynamics.

For Reggio Calabria specifically, these insights highlight the potential for using the proposed spatial framework to guide local planning instruments such as the Piano Strutturale Comunale (PSC). By identifying emission hotspots along the coastal corridor and sequestration zones inland, planners can integrate zoning criteria that encourage mixed-use compact development near transit corridors while conserving or expanding green networks within eco-districts.

The results also underline the importance of functional, multi-scalar planning units, such as eco-cells and eco-districts, and of coordinated policies that address emissions extending beyond administrative boundaries while aligning local strategies with broader decarbonization goals. Moreover, the study demonstrates that CO₂ emissions and sequestration patterns can be interpreted through functional criteria, that is, indicators linking land-use types to ecological processes such as regulating, provisioning, and supporting services. This perspective moves beyond descriptive mapping toward a functional understanding of ecosystem services that can guide more targeted interventions.

The results also highlight the role of peri-urban and transitional landscapes as ecological buffers and potential carbon sinks. The combined use of Urban Atlas classification and the hierarchical spatial unit enhances comparability and functional accuracy.

In practical planning terms, the eco-cell and eco-district framework provides a direct bridge between spatial carbon analysis and land-use management tools. The identification of emission hotspots and sequestration zones can inform zoning coverage, directing high-emission uses (e.g., industry or transport corridors) toward decarbonization priorities while preserving or expanding green infrastructure in high-sequestration areas. Planners can translate these outputs into measurable green infrastructure targets, for instance, ensuring that each eco-district maintains a minimum percentage of carbon-sequestering land uses such as forests, parks, or agricultural zones.

Overall, the case study demonstrates that the proposed framework captures Reggio Calabria’s territorial dynamics and translates them into actionable environmental insights. This demonstrates the potential of spatially explicit, ecosystem-based planning to guide climate strategies in Mediterranean cities facing similar challenges.

Limitations and Dissemination of Uncertainty

Despite its strengths, this study has limitations that affect the generalizability and precision of the results. The model relies on Urban Atlas (2018) land-use data, which, though harmonized and standardized, may not fully reflect recent land-use changes or informal developments. The CO₂ coefficients are parametric averages that do not capture context-specific differences in building materials, usage intensity, vegetation condition, or local microclimates. Furthermore, the SDM remains sensitive to model specification, and the spatial weight matrix may not fully capture non-contiguous interactions, especially in irregular or mountainous terrain. Another limitation is the absence of socioeconomic and energy-use variables, which could clarify spatial anomalies such as the unexpected negative coefficients in sparsely populated areas.

A key methodological consideration concerns the influence of spatial scale on analytical outcomes. The Modifiable Areal Unit Problem (MAUP) implies that results depend on how urban space is partitioned. Using both eco-cell and eco-district units allowed cross-scale validation and reduced the aggregation bias common in single-resolution studies. This hierarchical design supports stable spatial inference and maintains consistency between local and regional emission patterns. Although this approach reduces scale-related bias, results should still be interpreted cautiously because MAUP can influence spatial dependencies. Future sensitivity analyses could test alternative resolutions, but the chosen configuration offers a balanced representation of fine and strategic patterns relevant to urban planning.

Future research could enhance this framework by integrating temporal datasets to model land use change over time and assess the effects of policy interventions. Longitudinal analyses could reveal whether development trajectories lead to measurable emission reductions. Expanding the model to include other greenhouse gases or environmental indicators could offer a more comprehensive assessment. Additionally, machine-learning tools such as XGBoost or spatial neural networks could capture nonlinear spatial dependencies beyond parametric models. Finally, stakeholder engagement through mapping or validation workshops could improve the framework’s applicability.

By identifying emission hotspots, sink zones, and land-use-specific behaviors, this approach supports the shift from conceptual ecosystem service planning to data-informed urban strategies. Scaling up will require closer coordination among ecological data infrastructures, planning institutions, and local governance. Incorporating dynamic datasets and applying the framework in other cities would help validate its robustness and enhance its practical value. Although the framework demonstrates potential for replication, its transferability should be tested under varying data resolutions, policy settings, and ecological conditions.

Previous works have also highlighted the sensitivity of spatial econometric models to data resolution, coefficient generalization, and scale definition, particularly when applied across heterogeneous urban landscapes [71,72]. Apte and Manchanda [75] similarly noted that even high-resolution environmental mapping faces uncertainties in data coverage, parameter calibration, and temporal variability. They found that urban air pollution models struggle with uneven spatial representation and context-dependent bias. These limitations are not unique to this study but reflect challenges in spatially explicit carbon modeling. Recognizing these parallels positions the framework within broader research and underlines the need for improved data integration, model calibration, and scale testing.

6. Conclusions

This study introduced a spatially explicit, land-use-based framework for assessing CO₂ emissions and sequestration using the Reggio Calabria FUA as a case study. By integrating high-resolution land-use data with parametric emission coefficients and spatial econometric modeling, the framework captured carbon dynamics at both the eco-cell and eco-district scales.

The results show that land-use composition strongly influences emission intensity. Dense urban fabric recorded approximately 25–30% lower per-cell emissions than low-density residential zones, confirming that compact urban patterns can improve carbon efficiency. Industrial and transport-related areas contributed over 60% of total emissions. Forests and agricultural lands offset an estimated 15–18% of annual emissions. At the eco-district scale, the FUA exhibited a net positive carbon balance of approximately 1.85 million tons of CO₂ per year, emphasizing an emission surplus relative to local absorption capacity.

The eco-cell scale proved effective for detecting localized emission hotspots and land-use impacts. Its granularity helps to identify priority zones and align mitigation actions with ecological capacity. The SDM confirmed strong spatial dependence: industrial and transport land uses produced direct positive coefficients of 0.31 and 0.27, respectively, while natural land covers showed negative coefficients ranging between −0.12 and −0.18, indicating their mitigating effect. These results demonstrate that both direct and spillover effects are essential for understanding carbon dynamics.

Three planning conclusions emerge. First, emission-reduction strategies should focus on restructuring dispersed urban patterns and encouraging compact, mixed-use development. Second, expanding forested, agricultural, and green infrastructure zones could enhance sequestration by up to 20%, based on current land availability. Third, integrating spatial carbon models into zoning and infrastructure planning can support evidence-based decisions.

Spillover mapping reveals how emissions, heat, and pollution cross municipal borders. This supports inter-municipal compensation mechanisms in which high-impact areas contribute to mitigation in affected neighboring areas. This aligns with the FUA perspective, which manages sustainability at the functional rather than administrative scales.

The spatial spillover perspective offers planners insights for coordinated and compensatory strategies across urban systems. Quantifying how emissions or sequestration in one area influence neighboring areas helps internalize environmental externalities in planning. This evidence can guide green-infrastructure investments, support compensation mechanisms, and inform policy instruments.

Looking ahead, the framework can be replicated temporally as updated land-use data become available through future Copernicus and Urban Atlas releases, allowing the monitoring of carbon trends over time. Future research could integrate socioeconomic and energy-consumption variables and apply sensitivity analyses to refine coefficient uncertainty.

Ultimately, the study reaffirms the necessity of climate-responsive urban planning supported by robust, data-driven tools. Integrating spatial models into planning empowers local authorities to make targeted interventions aligned with zoning, land use, and ecological objectives. This framework offers a scalable approach for integrating UES into planning practices, supporting more adaptive and carbon-conscious urban development.