I selected a study area based on two criteria
- high media household income – people are more able to afford solar panels
- high residential building density – to study the effects on different houses
After investigating the Median After-tax Household Income provided by the City of Surrey, I chose the area of Cloverdale.
Building DEM
First, I converted the LiDAR data in LAS format to LAS datasets that were readable by ArcGIS. Two LAS datasets were created for the purpose of this analysis, they were building data (return value=6) and ground data (return value =2). They were then converted again into raster DEM for further analysis. The z-value in the building DEM layer contained the elevation of the ground, i.e. the ground layer height. In order to obtain the height of the buildings themselves, the ground layer was subtracted from the building layer. A DEM layer with building height was created (Figure 1).
Figure 1. Obtaining the height of the buildings
Data examination and correction
Land use and building shapefile were equally important pieces of information to perform this study. The accuracy of these two layers could directly influence the result of the analysis. Hence, before I started my analysis, I thoroughly examined the data and corrected all the errors. To make sure the building shapefiles polygons were in the right place, I compared the building layer against the orthophoto taken by the City of Surrey in the same year. I found that some of the building footprints polygons were incomplete (figure 2a). I manually corrected the shapes according to the actual building outlines in the orthophoto (figure 2b), and deleted those that were out the boundary of the orthophoto to enhance the value of the data layer.
Figure 2. Data correction: Edit incomplete polygons in the building shapefile
Furthermore, there were rooftops too small to even support 2 solar panels. They were mostly small garage roofs (figure 3). There roofs were deleted from the data set.
Figure 3: Data correction: Remove unsuitable polygons
Analysis started after data accuracy was ensured. I began by singling out the residential buildings from the land use layer, then I created a new layer for residential buildings. I clipped the building shapefiles using residential land use layer to identify buildings that were suitable for installing household rooftop solar panels. The next step was to combine the height from the DEM to the residential buildings shapefiles. Since residential rooftops were the interest of study, only cells in a raster that corresponded to the building footprint was extracted (figure 4). The Extract by Mask tool was used.
Figure 4. Extract the cells of a raster that correspond to the building mask
Solar radiation
To model and understand the effects of the sun on residential building, the Area Solar Radiation tool was used. It allowed me to calculate the incoming solar radiation (insolation) across my study area. The following parameters were adopted when modelling insolation in summer, winter, and all year round on residential roofs.
Below are the final solar maps in different periods of time:
Map 1. Solar Map for Summer and Winter. Please refer to this winter and summer PDF file for a higher resolution map.
Map 2. Annual Solar Map . Please refer to this annual PDF file for a higher resolution map.
To get a general view of the solar energy generation potential, the annual solar radiation model was used in the following analysis. The values of solar radiation were classified into 5 classes by natural distribution. Each class represented a level of suitability. These values were reclassified. The classification was as below:
Normally, raster layers do not come with an attribute table. However, an attribute table was needed to be extracted for the calculation of the total solar radiation rooftops could receive. To do so, I first made all the values of the raster cells into integers using the Raster Calculator. This allowed me to convert the floating type raster to an integer type raster. An attribute table could be built with the Build Raster Attribute Table tool from the integer raster layer.
We can see from the figure below that aspect plays an important role in the amount of insolation received by roofs. In the northern hemisphere, where Canada is located, a south-facing slop is warmer and more exposed to solar radiation as the sun stays mostly in the southern half of the sky (Pelland et al., 2003). As a result, solar panels should be placed on the south side of the residential buildings to maximize the performance of solar energy generation. The influence of aspect is exemplified in map 2. The highest solar cells aligned very well with the south-facing roofs.
Map 3. Rooftop Aspect and Solar Panel Suitability Map. Please refer to this Aspect and suitability PDF file for a higher resolution map.
These solar radiation data needed to be converted into polygons, After reclassification, data was generalized using the tool Majority Filter to remove anomalies (figure 5). This tool smooths cell surface by looking at the majority of a cell’s contiguous neighboring cells and determining the most appropriate value for that cell. Anomalies included gaps in a high insolation cell due to the variation in insolation.
Figure 5. Data generalization
Generalized raster cells were converted to polygons. After conversion, problems with solar radiation raster cells could be seen. Some raster cells extended outside of the building footprint due to rigidity of a square cell. To solve this problem, the solar radiation polygons were clipped to fit the shape of the buildings.
Figure 6. Clipping of solar radiation polygon to building footprint
A suitability map is now done! Final maps are presented under the result tab.
Below is the flowchart for this project:
