In today’s world, there is an ongoing search for renewable energy sources, specifically as potential alternatives to fossil fuels, which are known to have a large negative impact on the environment, human physical health and social peace (Meadows et al., 1972). Several energy sources have been developed in an effort to mitigate the use of fossil fuels, for example, nuclear energy, hydroelectric, solar, wind and natural gas (Gastli and Charabi, 2010). This paper will focus on photovoltaic (PV) solar energy, which harvests the sun’s radiation to produce electricity, and has the potential to greatly reduce fossil fuel use (Dresselhaus and Thomas, 2001). Specifically, this research will explore methods to increase the ease of planning city-wide rooftop solar energy systems by exposing rooftops which are well suited to host PV systems. There are three approaches to accomplish this: constant variable method, manual selection and GIS-based models (Gagnon et al., 2016), which are discussed below.
The constant value method is an extremely naive approach based on the assumption that the percentage of rooftop space suitable for hosting a PV system can be represented by a constant value (Gagnon et al., 2016). For example, if it is found that 20% of rooftops are suitable for PV systems, that number could then be applied to a large area. This method, though simple to apply, lacks in a few areas. Primarily, there is no specific indication of which rooftops are well-suited. Additionally, suitability is often poorly estimated and assumptions regarding slope and area must be made (Melius, Margolis, and Ong, 2013). Since PV suitability varies depending on local urban morphologies and geographic location, empirical methods are preferred.
Manual selection methods are an umbrella category for any approach which is not automated; human analysis of aerial imagery and in-person site visits are manual methods. These methods are extremely time intensive and are not well suited to large areas (Shahabi et al., 2014). Suzuki, Ito, and Kurokawa (2007) expanded on the analysis of aerial images by automating the process; analyzing the colours present in the imagery to determine rooftop suitability. The images are uploaded into a program which sorts data in each image into one of eight classes based on a maximum likelihood classifier. Although this method is relatively quick, it is known to be inaccurate due to various assumptions made by the model (Suzuki, Ito, and Kurokawa, 2007). For example, it is assumed that there are only three types of land cover: rooftop, vague (water, black) and non-roof (forest, grass). This assumption leads to an overestimation of rooftop availability (Suzuki, Ito, and Kurokawa, 2007).
Lastly, a geographic information system (GIS) may be used to identify suitable areas. GIS analysis approaches are well-suited to large-scale assessments of rooftops as they are designed to recognize and evaluate the relationships between spatial data (Shahabi et al., 2014). Using a GIS often amalgamates the advantages of the previous techniques; processing large areas (like constant value) while maintaining much of the precision manual analysis affords (Gagnon et al., 2016).
Within GIS-based analyses, many studies have been conducted on the use of light detection and ranging (LiDAR) data to inform rooftop suitability for PV systems. LiDAR is a method of remote sensing which uses pulses of light to measure the distance between the sensor (typically airborne) and the surface below it. This pulse data, when combined with other measurements from the sensing device, creates a precise three-dimensional model of the surface (What is LIDAR? 2015). In assessments of rooftop PV suitability, LiDAR data are typically converted to a digital surface model (DSM) and used in conjunction with solar radiation and climate models as well as algorithms to determine shading (Redweik, Catita, and Brito, 2013; Boz, Calvert, and Brownson, 2015). Boz, Calvert, and Brownson (2015) specifically propose a method which uses both LiDAR and building footprint data to extract rooftop segments which could host a PV system; taking into account the segment’s slope, aspect, shading, and area. The use of these factors coupled with the precision of LiDAR data led Boz, Calvert, and Brownson (2015) to successfully identify suitable rooftops within their study site. This research will expand on the methods used by Boz, Calvert, and Brownson (2015) in an effort to increase the accuracy of their model.
The analysis that Boz, Calvert, and Brownson (2015) used to perform their assessment of rooftop suitability is very complex, hinging on the use of spatial analysis methods; specifically interpolation of elevation, slope, aspect and solar radiation. A geographic information system (GIS) must be used to complete this analysis due to its ability to perform the required interpolations, but also to account for the spatial relationships between data. Additionally, a GIS allows its users to perform analyses that would be extremely complex to carry out by hand, especially for large datasets. This research will use ESRI’s ArcMap for the above reasons as well as its ability to create automated workflows which can be easily shared with interested parties.
Purpose of the Research
This study verifies and improves the model proposed by Boz, Calvert, and Brownson (2015) to process LiDAR data in order to characterize rooftop solar opportunities. The analysis is taking place in Louisville, KY and is expected to further the application of current remote sensing technologies to renewable energy development.