Solar potential of rooftops in Cáceres city, Spain

ABSTRACT Climate change is one of the challenges our society has to deal with nowadays. Photovoltaic is one of the main renewable energies and it is gaining ground all around the world. The prediction of solar radiation is crucial for this kind of energy. The aim of this work is to produce a solar potential map of rooftops the city of Cáceres (Spain) providing an overview of the employed methodology. The estimation of global radiation is based on LiDAR data of high density. Historical radiation records have been also employed to define the calculation parameters. The representation of the estimated global radiation on each building provides a wide range of new possibilities in the use of renewable energies and changes the conception of rooftops to a potential source of photovoltaic energy.


Introduction
The use of renewable energies is one of the most effective tools against climate change. As reported by the International Renewable Energy Agency, at the end of 2016, global renewable generation capacity amounted to 2006 GW. Photovoltaics (PV) and Wind Energy are key technology options for implementing the shift to a decarbonised energy supply and can be deployed in a modular way almost everywhere on this planet (Lacal Arantegui & Jäger-Waldau, 2017).
Policy-makers and investors tend to pay most of their attention to wind and solar electricity, while high-capital baseload technologies like nuclear, coal and natural gas are currently politically and economically less attractive (Paltsev, 2016). Solar energy is abundant, offers significant opportunities for climate change mitigation and it can be used to meet a variety of energy service needs (Hoggett, 2014). The International Energy Agency expects that in the next 5 years 30,000 solar panels will be installed every hour. Their previsions include China, USA and the EU at the top of the growing communities.
More efficient buildings are supporting the whole energy system transformation: Rapid deployment of high-efficiency lighting, cooling and other appliances. Solar building envelopes are increasingly attracting interest (Maurer, Cappel, & Kuhn, 2017). Additionally, urban energy landscapes are generating an urban energy transition towards experimentation in sustainability governance. It has a direct implication not just for planning, but also for the development of energy and urban policy that reflects the fundamental idea that spatial diversity relates to the production of innovation (Broto, 2017).
Recent studies like Bazán, Rieradevall, Gabarrell, and Vázquez-Rowe (2018) support that installing photovoltaic panels in rooftops would allow cities to be self-sufficient in electricity production for residential, commercial and public lighting purposes provided that these investments were to be performed in underutilised rooftops as well as supposes an attractive action in terms of climate change mitigation.
The prediction of solar radiation is crucial for this kind of PV energy, and has been a research topic in recent years and in several locations (Yaniktepe, Kara, & Ozalp, 2017). Usually, this kind of prediction is performed by Geographical Information Systems, but many other applications are being developed for these issues, like: Donatelli, Carlini, and Bellocchi (2006) who developed a software component called GSRad containing models to estimate extraterrestrial and ground-level solar radiation; Bezir, Akkurt, and Özek (2010) who proposed another programme that uses altitude, latitude, date and temperature of the place as an input for obtaining both daily and monthly solar radiation; Bayrakçı, Demircan, and Keçebaş (2017) (2015) an European web interface to produce calculations of solar radiation and PV system energy production; NASA (2016) a Web Mapping Application and Services contain geospatially enabled solar, meteorology and cloud related parameters formulated for assessing and designing renewable energy systems; Energy Sector Management Assistance Program (2016) an interactive atlas that provides longterm averages of solar resource to determine solar power generation; Solargis Company Ltd (2016) which includes a software for making a reliable calculation of PV electricity potential, and within the particular area, Ciemat (2012) have designed an interface called ADRASE that offers solar radiation data, with an approximate resolution of 5 × 5 km. The aim of this work is to perform a solar potential map of rooftops in Cáceres city (Spain) based on Light Detection and Ranging (LiDAR) data and historical radiation records.

Cáceres city
The studied area is located in western Spain. It is a small city of 36 km 2 with a population of around 100,000. It has a Mediterranean climate with smooth winters and warm summers, in which the average maximum temperature is 34°C. Rainfall is abundant in the months of October, November, March, April and May, but very intermittent.
As shown in Figure 1, Cáceres is positioned in the area with higher solar potential values in Europe, with approximately 5.1 kWh/m 2 of average global radiation in a year.
Constructions are medium sized in a large extent of the city centre, and the abrupt orography of the city is notable. As shown in Figure 2, there are approximately 15,200 buildings distributed along the urban extension, but nearly none of them have solar panels to leverage Figure 1. PV solar electricity potential in Europe (Source: http://re.jrc.ec.europa.eu/pvgis/). the received solar energy, with the exception of some residential houses, that use this energy to warm up their own water tanks.

LiDAR data and rooftop vectors
LiDAR has been used to record the Earth's surface since the beginning of the 1990s and developed to a reliable and economic recording method due to the development of efficient sensor and computer technology in recent years.
This method uses polar coordinates to relate Global Positioning System and Inertial Measurement Unit systems to survey the ground with high accuracy. Flight path position and orientation are recorded temporarily and spatially at the same time that the laser scanner measures angle and distance to the Earth's surface.
These LiDAR data have been proved to be a helpful tool for the analysis of the solar potential of partial urban sites in several works such as: Tereci, Schneider, Kesten, Strzalka, and Eicker (2009) or Brito, Gomes, Santos, and Tenedório (2012) with cases studies in suburbs of cities, the second one in 538 identified buildings of Lisbon; Szabó et al. (2016) that compared results of radiation with LiDAR and low-cost drone 3D model in 7 km 2 of urban area and Brito, Freitas, Guimarães, Catita, and Redweik (2017) modelling the radiation of two representative areas of 500 × 500 m 2 .
The airborne laser scanner survey of the area was performed by means of an ASL50-83 survey flight with an average point density of 1.5 pts/m 2 . A total of 102,987,890 measured points were obtained to generate the surface model of the city. The vertical accuracy of the obtained point cloud was of ±0.081 m (RMSE).
Additionally, a regular Digital Surface Model (DSM) of 1 m of spatial resolution, generated from the elevation of LiDAR points (Figure 3), was also provided. This raster DSM denotes the eminent differences in elevation between different spaces of the urban area.
The rooftop vectors representing buildings of the city (Figure 4) were obtained by means of photogrammetric techniques from digital images of 9 cm of pixel spatial resolution and a flight scale of 1:3500. Both LiDAR data and rooftop vectors were provided by the local council administration.

Historical radiation data
An historical series of radiation data was provided by the Spanish Meteorological Agency (AEMET). Daily Global radiation of years 2013, 2014, 2015 and 2016 of the city meteorological station was used for the determination of some crucial parameters of the calculation such as Transmittivity (T) and diffuse proportion (DP).

Methods
The solar radiation was calculated in a Geographic Information System (GIS) and based on methods from the hemispherical viewshed algorithm developed  by Hetrick, Rich, Barnes, and Weiss (1993) and further extended by Fu and Rich (1999, 2000a, 2000b. This algorithm presumes that solar radiation travels through the atmosphere; later on, it is modified by topography and other surface features, and finally, it is intercepted as direct, diffuse and reflected insolation components. Reflected radiation has to be considered only in places where ground surfaces have high albedo, for example, snow-covered surfaces (Fu & Rich, 1999). As it is not the case, in our study area global radiation has to be calculated only as the sum of direct and diffuse radiation.

Estimation of calculation parameters
The algorithm used does not model clouds inherently, because clouds are extremely hard to model or predict. However, the hypothetical effects of clouds could be reflected inherently with the transmittivity and DP parameters. Their values are crucial to obtain realistic radiation results of the studied area.
3.1.1. Transmittivity Solar radiation that reaches at the Earth's surface is only a percentage of the radiation received outside the atmosphere. As stated in Fu and Rich (2000b), Transmittivity is the ratio of the directly transmitted radiation incident on the surface after passing through the unit thickness of the atmosphere to the radiation that would be incident on the surface if the radiation had passed through a vacuum. In other words; Transmittivity is the ratio of the energy reaching the Earth's surface to that which is received at the upper limit of the atmosphere (extraterrestrial). Values range from 0 (no transmission) to 1 (complete transmission).
If the values of the direct, diffuse and global radiation are known, the transmittivity (T) can be obtained using the following equation defined by Galán, Arce, Koch, and Lara (2015): where GR is the global radiation and ER is the extraterrestrial radiation. The extraterrestrial radiation depends on the latitude (LAT) and the declination (DEC), and it was defined by (Duffie & Beckman, 2013  where GSC is the solar constant, n is number of the midmonth day, ω s is the sunset hour angle and it is defined by Equation (3): v s = − tan LAT tan DEC. (3) Following the above-described formulas, T was calculated for years 2013, 2014, 2015 and 2016.

Diffuse proportion
The DP is the fraction of radiation that has been scattered over all parts of the sky. This parameter encloses information about atmospheric turbidity: low values being associated with clean skies, and high values of diffuse fraction being linked to high turbidity (Batlles et al., 2008). Values range from 0 to 1. This value should be set according to atmospheric conditions by the following equation, described in Galán et al. (2015): where GR is the global radiation and DR is the direct radiation.
As in the previous parameter, the DP was calculated for years 2013, 2014, 2015 and 2016.

Other parameters
Two possible diffuse models have to be selected for the calculation. On the one hand, Uniform Overcast Sky (UOS), which is often applied in clear sky conditions (non-rainy months). This model assumes that the incoming diffuse radiation is the same from all sky directions. On the other hand, in a Standard Overcast (SO) diffuse model, diffuse radiation flux varies with zenith angle. It has a uniform component plus another component that increases towards the zenith (Kennelly & Stewart, 2014).

Calibration of calculation parameters
Once all the parameters were calculated for each year, the final calculation parameters were obtained as the average of the obtained parameters.
With those averaged parameters, an initial calibration was performed in order to calculate the adjustment between the modelled radiation and the registered global radiation at the AEMET meteorological station for each year.

Estimation of solar potential of rooftops
Once all the parameters were calibrated, the estimation of solar potential of the whole city was attained. The calculations were performed by means of ArcMap toolbox for solar radiation. Initially, the solar radiation was calculated throughout the whole DSM, in order to obtain real radiation with all the solar hidings produced by trees, other higher buildings, etc.
Finally, the obtained monthly rasters were trimmed off to consider only the rooftops of the city. Monthly quantification of all roofs was assessed and the final radiation map was designed, representing the average radiation in a year.

Estimation of calculation parameters
Firstly, the extraterrestrial radiation was obtained (Table 2) based on Equation (2).
From values in Tables 1 and 2 and Equations (1), (3) and (4), the values of monthly T and DP were calculated for each year and their average was considered as the final estimated parameters for the solar calculation (Table 3).
From an annual assessment of the accuracy, as shown in Table 4, the minimum differences represent dissimilarities in a day of only 10 Wh/m 2 (year 2014). Contrarily, the worst adjustment is of 0.85 h in a day (year 2015).

Estimation of solar potential of rooftops
Low carbon technologies are an important element of strategies to mitigate climate change, and diffusion of renewable energies for electricity generation potentially plays an important role within such strategies (Fischer, 2012).

Conclusions
In this work, a solar potential map of rooftops in Cáceres city (Spain) has been achieved. The representation of the estimated global radiation on each building provides a wide range of new possibilities in the use of renewable energies and, consequently, it could be of great help in the fight against climate change.
One of the crucial factors of this work is the precision and accuracy of the employed data. There are several studies that use Lidar data for solar modelling like Martínez-Rubio, Sanz-Adan, Santamaría-Peña, and Martínez (2016), Cheng et al. (2018), etc., but using low point density. Some studies like Jia et al. (2013) suggest that point density data set of at least 0.6 pts/m 2 is necessary to generate an accurate digital elevation model for the test of the measured urban environment. In this sense, the use of high precision LiDAR data of 1.5 pts/m 2 density confers to the present work a suitable accuracy.
Another major asset of this study is the use of rooftop vectors containing the real roof slopes, as it has been described, obtained by means of photogrammetric techniques with very high precision. These well-defined vectors, avoid the problem detected by Brito et al. (2012) when using only the LiDAR point cloud to define the edges of the roofs.
Additionally, the adoption of historical radiation records to configure the parameters of the calculations, confer the resulting map a pronounced accuracy and a proper fitting with real historical radiation data (an under estimation of less than an hour of energy in a day). In this way, the data and methodology calibration complies with one of the most important requirements stablished by Sergio, Deborah, and Daniel (2017) concerning the obtained result (final map) as a resource for PV decision-making policies.
Finally, focusing on the data obtained, the monthly radiation results in some of the buildings, in months like May, June and July, amounts to more than 7 kWh/m 2 . These results suggest that the city has a great renewable energy potential. Changing the perception of rooftops as a potential PV source would have huge consequences for renewable energy policy. Our map can inform decisions concerning PV panels. It can assist the siting of panels on roofs or even gables with optimum inclination, orientation and with no shadows from surrounding elements like buildings or urban vegetation.

Software
Microsoft Excel® 2013 and ESRI® ArcGIS® 10.4 were used to design maps, as well as to obtain radiation values with solar tools.