Characterization of the proximity to urban areas of the global energy potential of solar and wind energies

This study estimates the global energy potential of solar photovoltaics and onshore wind power and characterizes it with respect to the proximity to urban areas. Solar and wind power are centerpieces of a decarbonized energy system, and that different to other energy resources are disperse and widely available across the world. Therefore, characterizing how close or far these resources can improve the estimation of their availability. The analysis is based on a model using geo-referenced data and parameters related to the energy resources, technologies and land features. Results showed that the energy potential of solar (409 PWh yr−1) and wind (354 PWh yr−1) energies concentrates in the vicinity of urban areas, demonstrating the value of resources close to urban areas for covering current electricity needs. For example, current electricity consumption can be covered with high-grade solar resources (capacity factor >24%) within 30 km away from urban areas, or with middle-grade onshore wind resources (capacity factor >20%) within 20 km away from urban areas. Thus, it suggests that constraining the use of solar and wind energy in the proximity of urban areas due to social acceptability concerns, may significantly impact the deployment of high to mid-quality resources. The study is a starting point to evaluate the effect of restrictions and costs related to the proximity on the availability of renewable resources and their penetration in long-term decarbonization scenarios.


Introduction
Solar and wind energy are fundamental components of a decarbonized energy system. Solar and wind resources are widely available across the globe but dispersed and in low densities. The dominant technologies for producing electricity from these resources, solar photovoltaic (PV) panels and horizontal axis wind turbines, can be installed in a wide range of conditions. Their modular nature has enabled a high penetration rate in many countries. The current global capacity of solar PV and onshore wind power were 710 GW and 593 GW in 2019, respectively, and contributed altogether to 7% of global electricity supply [1]. The available solar and wind resources are abundant and many times larger than the current levels of energy supply [2]. However, the feasibility of harnessing these resources with the dominant technologies is constrained by geographic, technical, economic, and social factors. Therefore, it is necessary to quantify the availability of resources considering as many factors as possible, to clarify the potential to deploy these resources and contribute to the decarbonization of the energy supply. One of these factors is the spatial distribution of solar and wind resources with respect to the location of the energy consumers. However, this factor remains unexplored in global studies.
This study improves previous assessments by introducing the distance to urban areas to characterize the energy potential at global scale. Past studies have estimated the availability of solar and wind resources at global scale, basically using geo-referenced data, and focusing on different methodological aspects. Some only estimate the theoretical energy potential, such as for wind [3,4] and for solar PV [5]. Others, for the case of wind, put emphasis on limits to the capacity factor [6], sub-annual resolution [7], technology features as well as economic valuation of the resource potential [8][9][10]. For the case of solar there are techno-economic assessments [11,12]. There are also combined studies of solar and wind [13][14][15][16]. Although some studies account for the distance to urban areas in the case of wind power [17,18], they lack any explicit characterization of this factor. Overall, studies agree on the fact that solar and wind resources are disperse, larger than current energy supply at the global scale, and that their practical deployment across regions differs according to the climate, the land suitability, and the features of the power plants considered. Although these studies acknowledge the disperse nature of these resources, they lack any explicit characterization of the proximity or remoteness with respect to the location of energy demand. Distance has important implications for the practical deployment of solar and wind resources, such as the need for extension of electricity transmission lines, and the acceptability of local communities towards renewable energy facilities.
This paper estimates the availability of solar PV and onshore wind resources at global scale, and characterizes explicitly the proximity/remoteness of the resources based on the distance to urban areas. The analysis is based on a model using a database of geo-referenced data developed for the study. The novelty of the paper is the explicit characterization of the distance of the global energy potential with respect to the location of urban areas, by means of a method that avoids unsuitable locations along the shortest path between a grid cell and an urban area. By means of this, it is possible to arrive at a more realistic estimate of the energy potential, in contrast to previous studies that regarded as feasible the energy potential even in locations very far from urban areas.

Methodologies for estimating the energy potential and the distance to urban areas
The main inputs in the analysis are the technical potential from a model using a geo-referenced database, and the distance to energy demand [19]. The latter assumes energy demand centers correspond to urban areas. Outputs include the annual technical energy potential for grid cells with spatial resolution of 0.5 arc-minute (1 km horizontal grid), and aggregated into 17 world regions. Technical potential's outcomes are classified by grades according to the annual average capacity factor (CF) of electricity supply, and by distance to the closest urban area (i.e. urban proximity). The outcomes represent the annual average technical potential, reflecting historical conditions of the availability of solar and wind resource, large scale solar and wind technologies, and land cover distribution. Representation of the energy demand location is based only on the spatial distribution of urban areas, disregarding current electricity transmission infrastructure and energy demand densities. Details on the methodologies, the assumptions and the data used for the estimations are presented in the subsections below. The model documented and characterized in this study, has been applied in previous studies [20,21] and provided input data to an integrated assessment model used to assess emission scenarios at global [22] and national scale [23][24][25].

Technical energy potential
Technical energy potential is defined as the quantity of energy available for practical use after considering geographic, resource specific, and technology constraints. The methodology to estimate the global energy potential is based on a model, including data of the spatial distribution of solar and wind resources, and of geographical features (e.g. elevation, land cover, etc), along with parameters representing assumptions on land suitability and performance of technologies [19,26]. The estimated energy potential represents the annual average for 17 major regions. The global datasets used in the model are listed in table S-1 in the supplement. Solar PV potential is estimated from monthly average solar insolation on a horizontal plane [27], accounting for optimal inclination angle [26] of and shadowing among PV panels. The assumed technology is multicrystalline panels mounted as arrays in solar farms, with a conversion efficiency of 20% [28]. Wind energy potential is calculated from monthly average wind speed at 50 meters [29], and corrected to the turbine hub height (90 m) using a wind shear exponential formula. A Rayleigh distribution is assumed to derive full load operation hours from monthly average wind speed. Wind farms consisting of horizontal axis turbines of 2 MW rated capacity and 90 m hub height are assumed, with availability factor of 97%, cut-in and cut-out wind speeds of 3 m s −1 and 25 m s −1 respectively, and a maximum output corresponding to wind speed of 12 m s −1 . Wind farms consider a spacing of turbines equivalent to a density of capacity of 10 MW km −2 , and an array efficiency of 90%.
Geographic restrictions include a maximum slope of 3% for PV and 20% for wind [30], elevation below 2000 meters [31], and land suitability factors as listed in table 1. The latter represent limitations for installing conversion technologies due to competition from other land uses, and are set based on previous estimations of the global energy potential [9,13]. Values for solar PV are smaller than those for wind, given that wind turbines occupy a small surface area and are installed with considerable space among them, leaving empty areas suitable for other uses. Areas excluded from the estimation are forests, areas protected for nature conservation reasons, areas difficult to access (referred to as wilderness areas), wetlands, ice-covered areas and water bodies. Restrictions specific to each type of resource include the reduction in Sunshine hours due to landscape elevation angle, and losses in wind speed due to surface roughness. Restrictions specific to conversion technologies include the conversion efficiency, and the reduction in available installation area due to shadowing effects for PV panels and wind speed losses for wind turbines. Other restrictions, such as those related to economic, social and political aspects, are out of the scope of the assessment. Variability of the RE resource, and future scenarios for land cover, land use and population are not considered. Main equations for estimating energy potential per grid cell are presented below. Additional equations are provided in the supplement. Solar PV energy technical potential [MWh/yr]

Opt angle
Wind energy technical potential [MWh/yr]

Estimation of distance to demand centers
Distance to energy demand centers is defined for each grid cell as the weighted distance to the closest urban area. Urban areas are grid cells corresponding to 'urban and built-up' according to International Geosphere-Biosphere Programme (IGBP) land cover data [32]. As shown in figure 1, for a given source cell, the distance to a urban area (shaded cells in figure 1) is calculated as the cumulative distance between the center points of contiguous grid cells along the shortest path (dotted line in the figure) while avoiding cells unsuitable for installing transmission lines (gray cells in figure 1). Different to a straight line distance, this approach (usually referred to as 'cost-distance' in the field of GIS analysis) can represent distances that account for physical obstacles and avoid unsuitable locations (oceans and protected areas). Distance is converted from latitude-longitude units (e.g. arc-degrees) to length units applying a set of GIS tools supported by the GIS software (ESRI-ArcGIS ®). This conversion combines data for the size of each grid cell in meters in the vertical (north-south) and horizontal (east-west) directions, and the moving direction along the shortest path in each grid cell, to estimate the cumulative distance from the grid cell to the corresponding closest urban area. The flow of calculations is presented in figure 2.

Global results
The technical potential of solar PV is 409 PWh yr −1 , which is around 2.5 times the current global total primary energy supply (TPES) (160 PWh yr −1 as of 2018). Half of the energy potential corresponds to high-quality resources (i.e. CF above 24%), which are equivalent to 1.2 times the global TPES. For onshore wind power, the technical potential amounts 354 PWh yr −1 , which is equivalent to 2.3 times the global TPES. Most of the potential concentrates at low CF (below 30%). The wind resource of high quality, i.e. with an CF above 30%, is around 28% of the total technical potential. This is equivalent to more than 4 times the current global electricity demand (22 PWh yr −1 as of 2018). Numerical results of the energy potential are in the supplement (table S2).  The figure 3 shows the distance profiles, which represents the distribution of the energy potential across the 'distance to the closest urban area' for three grades representing different levels of resource quality (based on thresholds of annual average capacity factor). The figure 4 shows the cumulative distance profiles, which indicate the cumulative energy potentials summed by increasing 'distance to urban area' for the same three resource grades. The figure with the distance profiles shows the way in which the energy potential is distributed with respect to their proximity/remoteness to urban areas. The second figure describes how much of the energy potential is available within a given distance threshold.
The distance profile of solar PV resources indicates that the availability of resources decreases sharply after peaking at 20 km from the closest urban area. This trend is observed in the three grades (i.e. levels of CF) analyzed, with a less pronounced peak for the highest grade (CF >24%). Around 2.5% of these resources are in the closest proximity to urban areas. The distance profile of the technical potential of onshore wind power also decreases sharply with distance. The middle-grade resources (CF >20%) located closest to urban areas represent around 2.5% of the total technical potential. These outcomes are driven mainly by the spatial distribution of croplands and grasslands, which are the land cover types with the largest suitability factors (5% for solar PV and 30%/50% for onshore wind) among the suitable land cover types assumed in the estimation. These land cover types tend to locate in the vicinity of urban areas, in contrast to other suitable land cover types (such as barren/ sparsely vegetated areas).
The cumulative distance profile of solar PV ( figure 4) indicates that around half of the total potential is available within 60 km off urban areas. The current global TPES can be covered with middle-grade resources (CF >20%) within 50 km off urban areas. Depending only on high-grade resources (CF >24%) to supply the same amount of energy would require reaching resources up to 200 km off urban areas. The global electricity demand could be covered with high-grade resources just 30 km away from urban areas. The cumulative distance profile  of onshore wind shows a contrasting picture, were global TPES can only be covered including the whole range of resources within 50 km off urban areas. The whole middle-grade resources (CF >20%) are enough to cover only around half of the TPES. In contrast, global electricity consumption could be covered with middle-grade resources around 20 km away from urban areas.

Regional results
The regional cumulative profiles of the energy potential are shown in figure 5 for solar PV, and figure 6 for onshore wind. The figures also display the electricity demand and TPES as of 2018 for each region (regional distance profiles are in the supplement). The solar PV potential of high-grade resources (CF >24%) within 100 km off urban areas is enough to cover TPES in Brazil, Africa (XNF, XAF), Latin America (XLM), Middle East (XME), Oceania (XOC), and South Asia (XSA). Compared to electricity demand, high-grade resources close to urban areas are significant in India and USA.
The onshore wind potential of high-grade resources (CF >30%) across regions is considerable compared against the electricity demand in Brazil, Former Soviet Union (CIS), Africa, Other Europe (XER), Latin America (XLM), Middle East, Oceania and South Asia. In all regions except India, Japan, Turkey, and Southeast Asia, middle-grade resources (CF >20%) close to urban areas will suffice to cover electricity demand. These resources reach TPES levels within less than 100 km away from urban areas in Latin America, Middle East, Oceania and South Asia.

Discussion points
In a decarbonized global society dominated by renewable energy supply, the importance of solar and wind resources in remote areas will depend on each region's context. Those regions with deficit of renewable resources compared to their energy consumption levels, will have to rely more on remote resources, which will bring along more needs for investments in electricity transmission infrastructure, and/or dependance on imports of renewable electricity (or other energy carriers derived from renewable resources) from other regions. In regions with high amounts of surplus resources, remote resources can serve as reserves that could be exploited to fulfill future domestic demand or for export to resource scarce regions.
This study is a first quantitative and explicit assessment of the distance of solar and wind energy resources with respect to energy demand centers (i.e. urban areas) at global scale. It represents a quantitative proof of the spatial distribution of the energy potential, and of its applicability to specific regions and capacity factor ranges based on geo-referenced data. The outcomes demonstrated that there is a clear tradeoff between the availability of high-quality resources, which are attractive in terms of performance (capacity factor), and their accessibility (proximity to urban areas), which translate into larger expenditures for transmission infrastructure as well as increased difficulty (and likely also costs) for the installation, operation and maintenance of the technology. In addition to the needs for transmission infrastructure, another important implication of the analysis is the acceptability of local population towards large scale solar and wind farms. There are several studies reporting the lack of acceptability of actual installations in several countries, driven for example by concerns over the impacts on health (due to noise and flickering in the case of wind farms), on the local environment (due to interference with natural habitat and migration routes of certain species), or on the landscape [34][35][36]. Therefore, part of the energy potential in the vicinity of urban areas and protected areas might be unsuitable from the acceptability point of view. A previous study suggests that accounting for visibility restrictions for siting wind farms close to urban areas can decrease the global energy potential of wind power by around 20% [18].
Compared to other studies estimating the global energy potential, listed in table 2, this study's outcomes for solar PV (409 PWh yr −1 ) are in the middle range of those previously reported. For onshore wind, the outcomes (354 PWh yr −1 ) are also in the middle range of values reported in the literature. Differences in outcomes come from the use of different data sources, and parameters' assumptions, as well as different definitions for the ranges of quality of resources (capacity factors or unit supply costs). In terms of source data, this study uses monthly average data summarized into annual values, which is in contrast with recent studies which use hourly data [7,11,[15][16][17]. Another difference comes from the difference in spatial resolution of the solar and wind resource datasets, which in this study is 1 arc-degree in contrast to most studies which used 0.3-0.5 degrees resolution. In terms of the parameters' assumptions, those related to the land suitability can produce large differences, as suitability factors vary within a wide range among studies. For example, the range of values for croplands for solar PV and onshore wind in the literature is 0%-5% and 20%-80%, respectively. The values used in this study were set within the range of values commonly found in the literature, although the land cover dataset used is not the same and, thus, may likely use different definitions and categories of land cover. Technology parameters can introduce small differences compared to land suitability parameters, as their values have a smaller range of variation across studies. For example, the efficiency of solar panels varies between 11 and 14% among the surveyed literature. However, this is not the case for the density of capacity of wind turbines, which is applied uniformly to all regions in all studies, and ranges between 1 to 9 MW km −2 .
The methodology presented in this study can be applied in several potential analyses. First, the spatial matching of solar and wind power supply can be analyzed considering the location of energy demand rather than that of urban areas. This task requires the development of a map showing the location of populated areas together with the quantity of energy demanded in each location. Second, use of solar and wind resources for decentralized supply, in particular in rural areas, needs to be addressed. Rural electricity demand in the future may be significant at global scale due to improved access to electricity and growth in consumption levels. Therefore, it is worth analyzing the effective allocation of renewable resources, either for decentralized supply in rural areas, or for centralized supply in large scale plants. Third, sensitivity of parameters affecting the estimation of the energy potential and its distribution with respect to energy demand (urban areas) needs to be compared with other factors. These parameters involve large uncertainties which affect the scale of the impact of distance to energy demand on the technical potential. One major uncertainty in this study is the definition of which land areas are suitable for installing RE plants. Moreover, how much area of each land category is suitable for RE supply is subject to strong assumptions. The implications of using different assumptions for land suitability need further analysis. Also, the effect of changes in urbanization on a global scale should be investigated, as this factor can be important in the case of developing countries. In this respect, it is worth noting that the feasibility of the energy potential may be hindered in regions with poor economic conditions where electricity demands are marginal. Therefore, the actual deployment of the energy potential should aim to provide electricity supply in both poor and wealthy regions.
The characterization of the energy potential can be improved by analyzing the complementarity of solar and wind power farms, for example by considering hybrid energy systems. These systems, which combine the electricity supplied by solar panels and wind turbines in a given site, complemented by an energy storage system, can provide a smoother power output. For this analysis, data of high temporal resolution (e.g. hourly data) would be needed. Another important aspect affecting the estimation of the energy potential is the availability of improved and new solar and wind energy technologies. The future progress of these technologies may raise their conversion efficiency and their range of operational conditions. This will increase the estimates of the energy potential. Also, there are many alternative technologies and applications which are currently under development or at early stages of demonstration that can extend the availability of the energy potential in locations that currently are not feasible for the dominant technologies (such as floating solar panels, perovskite solar panels, airborne wind energy systems among others) or show advantages (proximity to existing power infrastructure, less land use conflicts).

Conclusions
This study estimated the global energy potential of solar PV and onshore wind power, and characterized its proximity/remoteness to urban areas with a method that measured distance avoiding unsuitable locations. It was found that the distribution of the potential of both solar and wind energies with respect to the closest urban areas is highly skewed towards the vicinity of urban areas, peaking at 0-30 km away from urban areas (depending on the resource grade considered) with a sharp decline afterwards. The global potential of solar PV (409 PWh yr −1 ) can cover current TPES by deploying high-grade resources within 200 km off urban areas. The potential of onshore wind (354 PWh yr −1 ) energy can cover the same amount of energy only if all resources (high to low grades) are deployed within 50 km off urban areas. Reaching levels equal to the current electricity demand is possible with high-grade resources of solar PV (CF >24%) within 30 km off urban areas, or with middle-grade resources of onshore wind (CF >20%) within 20 km off urban areas. The distribution of the energy potential was unequal among regions, although they showed a similar distribution with respect to the distance to urban areas. High-grade resource of solar PV in the proximity of urban areas are comparable to each region's TPES in Africa, Latin America, Middle East, Oceania and South Asia. For the onshore wind potential, high-grade resources in the proximity of urban areas are only comparable to regional electricity demands, in particular in Africa, Latin America, Oceania, the Former Soviet Union, part of Europe, Middle East and South Asia. This research can be extended to analyze the distribution of the energy potential for decentralized energy supply, and the effect of restrictions on the energy potential due to proximity to natural and human environments. Also, additional research is needed to clarify the role of some assumptions, such as those related to land suitability, and the updated distribution of urban areas or relevant infrastructure (such as electricity transmission lines). Finally, the assessment of the energy potential can be complemented with analysis of the economic aspects of the potential, and the interaction with other energy supply options by means of energy system models and integrated assessment models.