Assessing Rooftop Solar Potential for Photovoltaic Installations in Genoa: A GIS-Based Approach for Urban Energy Planning

Over the past decade, energy systems have undergone a notable transformation towards greater sustainability, marked by enhancements in efficiency and the decentralisation of renewable energy production. This evolution has catalysed a sharp rise in solar energy deployment, spurred by policy interventions from both national governments and international bodies seeking to address the escalating climate emergency. Additionally, the strategic imperative to diversify energy supply sources and mitigate reliance on external providers has reinforced the momentum of this transition. Within the European Union, photovoltaic (PV) technology has played a central role in this expansion, with the cumulative installed capacity reaching approximately 339.4 GW by the close of 2024, including an estimated 62.6 GW added in that year alone [1].

Over recent years, PV technology has evolved from a solution largely reserved for affluent single-family households into a widely accessible and economically viable option across diverse residential settings. In parallel, many European countries have progressively introduced regulatory frameworks that facilitate the development of PV sharing schemes and energy communities. Within this evolving landscape, digital solar platforms play an increasingly important role in promoting knowledge dissemination and fostering collaboration among stakeholders [2].

This type of sharing community has been formally recognised in Italy through the establishment of Renewable Energy Communities (RECs), i.e., legal entities composed of members such as citizens, small and medium enterprises, local authorities, associations, educational and research institutions, religious organisations, and environmental groups. These members collectively produce and share electricity from renewable sources for their own consumption, thus supporting a decentralised energy model by encouraging the deployment of small-scale renewable installations. The establishment of RECs was initiated with the implementation of the RED II Directive and was further formalized with the approval of the Configurazioni di autoconsumo per la condivisione dell’energia rinnovabile (CACER) and Testo Integrato per l’Autoconsumo Diffuso (TIAD) Decrees, by Autorità di Regolazione per Energia Reti e Ambiente (ARERA) and Italian Ministry of Environment and Energy Security (MASE).

Given that approximately 39% of the European population resides in urban centres and 36% in towns and suburbs [3], these densely populated areas present considerable opportunities for the integration of PV systems and the advancement of distributed energy generation. It is worth noting that, by 2024, over one-third of PV systems installed globally are located on rooftops [1], underscoring the relevance of urban buildings as key assets for solar energy exploitation.

Within the framework of RECs, which are supported by dedicated regulatory and operational mechanisms, identifying suitable urban surfaces for solar deployment becomes a critical step. In this regard, solar assessment platforms, commonly referred to as solar cadasters, serve as key tools. These platforms integrate high-resolution spatial and meteorological data to provide accurate evaluations of solar potential across cityscapes, enabling stakeholders to identify rooftops that are viable for PV or solar thermal installations [4]. The effectiveness of such assessments relies heavily on accounting for critical factors such as surface tilt, azimuth, and shading from surrounding structures, which are known to influence annual insolation and energy yield across both thermal [5] and PV applications [6].

Although detailed city-scale solar platforms are still relatively scarce, increasing interest in assessing urban solar potential has led to a growing body of research aimed at developing solar cadasters and urban planning strategies. Among the notable initiatives are the regional solar cadastre of Geneva [7], as well as urban platforms in various French cities (e.g., Montpellier, Béziers, Nantes, Bordeaux) and German cities (e.g., Bremen, Hamburg, Kaiserslautern). Similar efforts have also been implemented in parts of Belgium, Luxembourg, the Netherlands, Norway, Poland, Spain, Sweden, and the United Kingdom. A comprehensive review and classification of these platforms, categorised into atlases, mapping, and cadasters, can be found in the study [8].

In this context, the present study introduces a methodology for estimating annual rooftop solar irradiation in a selected district of Genoa (44.4056°N, 8.9463°E), a Mediterranean city located in northern Italy. The approach employs Digital Surface Model (DSM) and LiDAR datasets, processed within a QGIS environment, to accurately account for the effects of slope and orientation of urban surfaces relevant for solar technology deployment. Beyond the evaluation of solar exposure, the study includes a series of statistical metrics to develop a composite interest index for PV potential, evaluated at a spatial resolution of 0.5 × 0.5 m². This index combines rooftop area, mean annual insolation, and intra-surface insolation variability to identify rooftops that receive uniform solar exposure throughout the year. This is essential, as spatial variability in sunlight, often caused by intermittent shading, can have a considerable impact on solar system performance.

This chapter outlines the methodology adopted to compute annual solar insolation within the QGIS environment, using the Solar Energy on Building Envelopes (SEBE) model algorithms implemented through the Urban Multi-scale Environmental Predictor (UMEP) plugin. The following sections provide a detailed description of the input resources, the solar radiation modelling approach, and the formulation of profitability indices used for the comparative assessment of rooftop suitability (Figure 1).

qq20251126-113113.png

Figure 1. Schematic representation of the proposed methodology for assessing rooftop solar potential

3.1 Input resources

This chapter introduces the primary input resources employed in the present analysis, including raster datasets, vector layers, and Typical Meteorological Year (TMY) data, which together form the foundation for the spatial and climatic characterisation of the study area.

Raster data are structured as matrices composed of regularly sized pixels, each representing a specific spatial area. These data types are particularly well-suited for representing variables that change continuously across space, such as temperature, elevation, or land cover. The spatial resolution of a raster dataset is determined by the size of its individual pixels, with smaller pixel sizes providing finer spatial detail. In this study, the analysis is based on two types of raster files: the Digital Elevation Model (DEM) and the DSM. These raster datasets represent the Earth's surface in image form, where each pixel corresponds to a specific elevation value at a defined geographic location, referenced by latitude and longitude. A DEM provides a three-dimensional digital representation of the terrain surface, capturing detailed information on topographic features such as elevation and surface variation. They are generated by dividing the surface into a grid of cells or points, each assigned an elevation attribute. The accuracy and spatial resolution of DEMs depend on the quality and source of the input data. The DSM is a three-dimensional digital representation of the Earth's surface: unlike the DEM, the DSM includes all features present above the bare ground, such as buildings, vegetation, and both natural and artificial structures, thereby representing the topmost surface layer. In this analysis, DSM files are derived from recent LiDAR surveys and consist of 0.5 × 0.5 m² raster data.

The second type of data employed in this study consists of vector files. Unlike raster data, which are pixel-based, vector files represent real-world objects using geometric elements such as points, lines, and curves, defined by coordinates and mathematical instructions that allow for rendering shapes accurately. These files are widely used for applications requiring high resolution and scalability, as their mathematical basis allows for resizing without loss of quality. From the Geoportal of the Municipality of Genoa (https://mappe.comune.genova.it/MapStore2/#/), not only DEM and DSM raster layers are available, but also building footprints, as well as vectors representing several infrastructures. The building polygons include height attributes, which can be accessed through QGIS by selecting individual features and the attribute table (Figure 2).

qq20251126-113126.png

Figure 2. Vector layer of buildings overlaid on the orthophoto of the Albaro district (DSM 3930), Genoa, Italy

Finally, TMY data were obtained from the PVGIS platform [14] and are based on the ERA5 database, chosen for compatibility with the UMEP plugin. The database encompasses data for the period 2005-2023 and results in an annual global horizontal insolation (H_tot,H,y) of approximately 1445 kWh/m²; the diffuse component (H_diff,H,y) accounts for 490 kWh/m².

3.2 Insolation model description

The SEBE plugin enables the estimation of annual solar exposure on urban surfaces while considering the 3D morphology of the city. Initially, raster files representing the height and azimuthal orientation of each pixel are obtained, which are then used to generate slope rasters (Figure 3). This raster becomes one of the inputs for the Daily Shadow Pattern tool, which calculates the shadowing effects.

qq20251126-113134.png

Figure 3. Slope raster used for shadow computation

Using the well-established method introduced by Ratti and Richens [15], the sun’s path through the sky is simulated throughout the year at regular intervals and a shadow factor S for each pixel is defined according to Eq. (2). The shadow factor is calculated based on binary shadow values from buildings (S_B) and vegetation (S_V), and by considering vegetation transmissivity ($\tau$).

$S=S_B-\left(1-S_V\right) \cdot(1-\tau)$                       (2)

The shadow factor assumes values between 0 (i.e., pixel completely shaded) and 1 (i.e., pixel not shaded). Thus, the visibility of the sun for each DSM cell is assessed for each time step, accounting for all the available obstructions. In this analysis, shadow calculations are performed at 30-minute intervals and the shadow map is created by generating a Boolean image.

For the calculation of solar irradiance, the meteorological data is pre-processed and redistributed into 145 patches of similar solid angles throughout the sky vault according to the well-known approach presented by Lindberg et al. [16]; this approach has been adopted for numerous analyses of solar potential and the effects of tilt and orientation, as the recent paper by Ferry et al. [17]. Through this cumulative sky approach, the sky dome is represented by aggregating solar radiation data over time for each patch, allowing for a detailed assessment of solar exposure on surfaces with varying orientations and inclinations (Figure 4).

qq20251126-113156.png

Figure 4. Cumulative sky model generated for the city of Genoa based on PVGIS–ERA5 data, processed using MATLAB

To redistribute the direct component of solar radiation, the position of the sun is computed for each time step using the meteorological dataset. The solar position is then projected onto the sky dome and assigned to the centroid of the nearest sky patch among the 145 discrete elements defined by the default SEBE sky subdivision scheme. The model assumes that the direct irradiance originates from the solar disk center as it traverses the sky vault. In parallel, the diffuse component is spatially distributed over the sky dome using the Perez AWM, which provides a relative luminance distribution lv for the i^th sky element depending on the position of the sun and the meteorological conditions at the given moment, as in Eq. (3).

$\begin{gathered}l v_i= {\left[1+a \cdot \exp \left(\frac{b}{\cos \theta_i}\right)\right] \cdot\left[1+c \cdot \exp \left(d \cdot \chi_i\right)+e \cdot \cos ^2 \chi_i\right]}\end{gathered}$                   (3)

Relative irradiance is defined based on the zenith angle of each sky element (θ_i), the angle between the sky element and the position of the sun (χ_i) and a set of coefficients dependent on the insolation conditions: a is related to the horizon brightening or darkening, b to the horizontal gradient of luminance, c to the circumsolar intensity, d to the circumsolar width and e to the backscattering light on earth surface.

Absolute irradiance values (Lv_i) for each of the N sky patches, defined by their respective solid angles $\Omega_i$, are computed such that their cumulative contribution, when projected onto a horizontal surface, matches the total diffuse irradiance at ground level (G_diff), as expressed in Eq. (4). The model then incorporates the direct irradiance component by adding an additional element corresponding to the solar disk's position on the sky vault.

$L v_i=l v_i \cdot \frac{G_{d i f f}}{\sum_{i=1}^N l v_i \cdot \Omega_i \cdot \cos \theta_i}$                       (4)

Finally, the calculation of the global irradiance G_tot,T is based on the summation of the beam, diffuse and reflected contributions associated with each sky element i, as expressed in Eq. (5).

$\begin{gathered}G_{t o t, T}= \sum_{i=1}^N\left[G_{b, H} \cdot \cos \theta_w \cdot S+G_{d i f f, H} \cdot S+G_{t o t, H} \cdot(1-S) \cdot \rho\right]\end{gathered}$                 (5)

In detail:

• Subscripts tot, b, and diff refer to total, beam and diffuse irradiance;
• Subscripts T and H distinguish between tilted (T) and horizontal (H) surfaces;
• S is the previously introduced shadow factor;
• θ_w is the incidence angle, i.e. the zenith angle of the sun with respect to a tilted surface;
• ρ is the albedo of the surroundings, here assumed equal to 0.15.

3.3 Indicators for PV roof suitability

The output generated by the SEBE algorithm consists of a detailed dataset in which each rooftop is identified along with a set of attributes, including:

• Roof ID;
• Total roof area;
• Average insolation across the roof pixels;
• Minimum and maximum pixel insolation values per roof;
• Standard deviation of insolation among the pixels representing each roof;
• Median insolation value across the roof pixels.

To refine the analysis and filter non-relevant surfaces, specific thresholds were established. Only rooftops with an area exceeding 50 m² were considered, and minimum annual insolation thresholds ranging from 1000 to 1500 kWh/m² were applied, as further detailed in the results section.

Based on the insolation output data, this study proposes an overall index for rooftop solar utilisation potential (Φ_r), with the aim of supporting decision-makers, policymakers, private citizens, and businesses in identifying the most promising surfaces for decentralized renewable energy production, particularly within the framework of RECs. This unified indicator, developed in reference to the case study of the Albaro district in Genoa, enables comparative assessments at broader urban or regional scales for evaluating solar potential across different parts of the city. Specifically, as detailed in the following equation, the indicator is derived from the normalized linear combination of three distinct variables, each rescaled using min-max normalization as shown in Eq. (6). In the following, an asterisk superscript will be used to denote variables that have been rescaled using min–max normalization.

$x *=\frac{x-x_{\min }}{x_{\max }-x_{\min }}$                 (6)

The three variables used to compute the index Φ_r are described in detail as follows:

• The annual average insolation received by rooftop r ($\overline{\mathrm{H}}_{\mathrm{y}, \mathrm{r}}$);
• The effective rooftop area of r receiving annual insolation values above a predefined threshold th [kWh/m²/yr] ($\mathrm{A}_{\mathrm{r}}^{\text {th }}$);
• A solar uniformity index (U_r), expressed by Eq. (7) and derived from the ratio of the standard deviation to the mean insolation: in this case, the standard deviation is calculated based on the annual insolation values of each pixel p within the set of pixels P_r corresponding to rooftop r, with reference to the average insolation of the same roof.

$U_r=1-\frac{\sqrt{\sum_{p=1}^{P_r}\left(H_{y, p}-\bar{H}_{y, r}\right)^2}}{\bar{H}_{y, r} \cdot \sqrt{P_r}}$                  (7)

Finally, the index for rooftop solar utilisation potential Φ_r is determined according to Eq. (8) by weighing the three variables mentioned using numerical constants w₁, w₂, and w₃, which in this study take the values provided in Table 1.

$\Phi_r=w_1 \cdot\left(\bar{H}_{y, r}\right)^*+w_2 \cdot\left(A_r^{t h}\right)^*+w_3 \cdot\left(U_r\right)^*$                   (8)

Table 1. Weight coefficients for Φ_r calculation

The present study aims to provide a comprehensive understanding of the algorithms implemented in the UMEP plugin for QGIS, with the goal of evaluating the potential for PV installations at the urban scale. The analysis focuses on the Albaro district, a selected neighbourhood in the city of Genoa, Italy, with particular attention to the relevance of PV solutions within the framework of newly established RECs, where the spatial relationship between production and consumption significantly influences incentive allocation.

In the initial phase, filtering thresholds were applied, including minimum values for rooftop surface receiving sufficient solar exposure and a minimum annual solar irradiance per unit area, as derived from cumulative sky modeling based on TMY data. This pre-processing step enabled the identification of approximately 34% of the rooftops included in the study area’s building vector layer as suitable candidates. The primary outcome was the generation of a high-resolution (0.5 × 0.5 m²) solar potential map of the neighborhood, capable of capturing the impact of adjacent buildings, urban vegetation, and distant obstructions on annual insolation. Although beyond the main scope of this work, the map also allows the identification of urban canyons, i.e. areas where vertical PV installations would be particularly unfeasible due to persistent shading.

The overall solar potential of the filtered rooftops was found to be comparable to the total installed PV capacity across the entire province of Genoa (approximately 1,800 km²), underscoring the significance of urban PV deployment. A statistical analysis was subsequently conducted to examine the influence of the applied thresholds on the distribution of suitable roof surfaces and the uniformity of insolation, revealing clear patterns depending on the irradiance criteria.

Finally, a composite index of solar utilisation potential was computed for each rooftop, based on a linear combination of annual insolation, usable area, and irradiance uniformity. While 703 of the 1,229 qualified rooftops showed index values between 0.4 and 0.5, a smaller subset demonstrated high suitability for PV deployment, thus enabling a prioritisation of rooftops for future distributed solar energy projects.