ANALYZING OPTIMAL LOCATION OF SOLAR FARMS IN BUTUAN CITY USING
GEOGRAPHIC INFORMATION SYSTEM
Sheila B. Refamonte1, MeriamMakinano-Santillan1, 2
1
Divison of Geodetic Engineering, College of Engineering and Information Technology
Caraga State University,Ampayon, Butuan City, 8600, Philippines
Email: sheilarefamonte94@gmail.com
2
Caraga Center for Geoinformatics, Caraga State University, Ampayon, Butuan City, 8600, Philippines
Email: meriam.makinano@gmail.com
KEYWORDS: Solar Radiation, Land Cover, Weighted Overlay, GIS
ABSTRACT: This study applies Geographic Information System techniques to locate the optimal area in sitting
solar farms that would be able to paramount the collection of solar energy. To achieve the main objective of this
study, the factors that affect the amount of solar energy collected should be considered: solar radiation, slope,
hillshade, aspect, land-cover, distance to city proper, and distance to electric substations. The processes are done
with the aid of ArcGIS software. Digital Terrain Model (DTM) was used to generate the aspect, hillshade, slope,
and solar radiation parameters. Landsat 8 satellite image was used to generate the land cover of Butuan City.
Substations location data in Butuan City from Agusan del Norte Electric Cooperative (ANECO) was converted into
point shapefiles and was then used to generate the distance to substation raster. A polygon shapefile of the city
proper of the said study area was used to generate a distance to city raster. The classified pixel values of the GIS
based sitting models of the different factors are reclassified into 100, 75, 50, 25, and 0 values according to the
criteria and constraints of sitting the solar farms. The weights of solar radiation, slope, hillshade, aspect, land-cover,
distance to city, and distance to substations are 0.30, 0.15, 0.15, 0.15, 0.10, 0.10, and 0.05 respectively. The initial
result from weighing is analysed which among the areas is optimal considering the minimum area of 1 acre or
4046.86 square meter. The output of this study is a map showing the optimal location of solar farms in Butuan City.
1. INTRODUCTION
Greenhouse gases concentrations have risen over the last 250 years from greater fossil fuel use, modern wide-scale
agriculture, and land-use alteration. Although fossil fuels are still plentiful and inexpensive, the threat of global
warming has caused many to explore a switch to alternative, renewable energy source (Janke, 2009). The sun, the
earth’s primary source of free and limitless energy, provides a tremendous resource for generating clean and
sustainable electricity and new solar energy technologies are being developed in attempts to find environmentally
friendly alternative energies. Solar farm, also known as photovoltaic power station, is a large collection of
connected solar panels that work together to produce electrical energy. It has been constructed all over the world to
convert some of the solar energy into electricity, which is much more environmentally friendly than traditional oil
and coal energy sources (Mckinney, 2014). A primary objective for any power plant is to ensure the plant reliably
generates the maximum economic and energy performance returns (Belfiore, Taylor, Moisan, Zappia and Cinarelli,
2013).
This paper presents the area in Butuan City with the highest solar potential considering the criteria and constraints
of sitting the solar farms with the help of ArcGIS software. The study area is limited to the area in Butuan City with
available Digital Terrain Model as shown in Figure 1.
Figure 1 The DTM used in the study with a cell size of 1m.
2. MATERIALS AND METHODS
2.1 Software and Tools
Data processing and data analysis are done with the aid of ArcGIS 10.1. Some tools from the Spatial Analyst
tools,Data Management tools and Conversion tools from ArcToolBox are the major tools used to process the data to
achieve the objective of the study.
2.2 Data Used
The Digital Terrain Model of Butuan City was obtained from the CSU Phil-Lidar 1 project. The DTM was used to
generate the solar radiation, slope, hillshade, and aspect rasters. The DTM has a dimension of 31286 by 30686
pixels and a cell size of 1m by 1m with Universal Transverse Mercator (UTM) Zone 51 projection and the World
Geodetic System (WGS) 1984 as the horizontal datum. The City boundaryshapefilewas used to generate distance to
city proper raster. The Landsat 8 images in geotiff format were downloaded from http://earthexplorer.usgs.gov/.
Three Landast images captured at three different periods: March 18, 2015, April 3, 2015, and August 2014; was
used to generate the cloud-free and shadow-free land-cover raster. The substation location data was acquired from
Agusan del Norte Electric Cooperative (ANECO) and was used to generate distance to substation raster.
2.3 Data Processing
To produce the different raster files, the data gathered underwent several processes as explained below.
2.3.1 Solar Radiation Raster
Using the Area Solar Radiation tool, the DTM is loaded and the parameters are inputted. The central latitude of the
raster is automatically calculated and shown in the latitude bar in the area solar radiation box upon loading the
DTM. The time configuration was set into whole year with monthly interval and the year was set into 2015. The Z
factor and other parameters were set into default.
2.3.2 Slope Raster
The slope tool helps to determine the rate of maximum change in z values. The surface raster is loaded on the slope
box and set the output measurement into degrees. The Z factor is set into 1 (default).
2.3.3 Hillshade Raster
By using the hillshade tool, the shaded relief of the study area is shown considering the illumination source angle
and shadows. The DTM of the study area is loaded and the model shadows is checked while leaving other
parameters in default: Z factor is 1, azimuth is 315 degrees, and altitude is 45 degrees.
2.3.4 Aspect Raster
The aspect is the slope direction of each cell in the DTM. The surface raster is loaded and the output folder is set
into the user’s directory.
2.3.5 Distance to City Proper Raster
Euclidean distance tool is used to determine the cells that are closest to the City Proper of Butuan City. The city
proper shapefileis used to generate the distance to city raster. The extent is set into the range of the available DTM
by loading the DTM of Butuan City in the Environment Setting in the Euclidean Distance Box.
2.3.6 Land-cover Raster
The landsat 8 image acquired on March 18, 2015 is used to generate land-cover raster. To fill the clouded and
shadowed areas with the true data of the land or to make the image a cloud-free and shadow-free image, the landsat
8 images acquired on April 3, 2015 and August 2014 are used. First is the application of the atmospheric
correction- the conversion of the Digital Numbers (DN) of the three images to Top-of-Atmospheric (TOA)
reflectance followed by the Sun Angle correction using the raster calculator tool. The formulas areshown in
equations 1 and 2.
TOA reflectance= Reflectance_Mult_Band x DN values + Reflectance_Add_Band (equation no. 1)
Correction for Sun Angle= TOA reflectance/sin(SUN Elevation) (equation no. 2)
The needed inputs for the atmospheric correction are seen in the metadata in text document type file. After the
application of the correction, the images are clipped based on the available extent of the DTM.Then, the clipping of
cloud-free and shadow-free areas on April 2015 and August 2014 images followed using the Clip tool considering
the areas are clouded and shadowed on the March 2015 image. The clipped rasters and March 2015 image is then
mosaicked using the Mosaic tool. The initial output is processed under band rationing specifically the Normalized
Difference Vegetation Index (NDVI) to distinguish the greenness of the vegetation. The point shapefile is then
created in the ArcCatalogwith a spatial reference of WGS84_UTM_Zone_51N. By using the point shapefile, points
are collected in the landsat image with the assistance of NDVI and attributed to the class where the points belong.
The land-cover classes are River/Sea, Ponds, Built-ups, Barren, Shrubs/Grassland, Ricefield and Forest. The point
shapefile is converted into signaturefile using the Create Signatures Tool. Using the Maximum Likelihood
Classification as the method of image classifying, the signature created is loaded in Maximum Likelihood
Classification Box as well as the cloud-free and shadow-free Landsat image.
In the accuracy assessment, ground truth points are collected on the high resolution orthophoto images acquired on
the same date of the DTM used. According to the law of thumb,the total number of ground truth points to be
collected per class should be 10 times the total number of land-cover classes of the study area. The ground truth
shapefile is converted to a raster file using the Point to Raster tool. The converted points and the image classified
are combined using the Combined tool to evaluate the number of pixels in each class that were correctly and
incorrectly classified. The values showed in the combined rasters are analysed to come up with the accuracy of the
land-cover classification. The formula of computing the different accuracies are shown below:
Class Accuracy= (Training Points Classified Correctly/Total Ground Truth) *100 (equation no. 3)
Comisionclass= (Misclassified Points/Total Points)training points*100 (equation no. 4)
Omisionclass= (Misclassified Pixels/Total Pixels)ground truth*100 (equation no. 5)
Producer’s Accuracy= (Training Points Classified Correctly/Total Ground Truth)class *100 (equation no. 6)
User’s Accuracy= (Training Points Classified Correctly/Total Training Points) class*100 (equation no. 7)
Over-all Accuracy = (Sum of all correctly classified pixels)/Total Ground Truth*100 (equation no. 8)
Kappa Coefficient= [(Total Pixels*Sum of Correct)-(Sum of all the (Row Total*Column Total)]/Total pixels
Squared-Sum of all the (Row total*Column Total) (equation no. 9)
2.3.7 Distance to Substations Raster
The data from the ANECO showing the coordinates of the substations in Butuan City are processed using Euclidian
Distance tool to determine the closest cells from the substations.The extent is again set into the range of the
available DTM by loading the DTM of Butuan City in the Environment Setting in the Euclidean Distance Box. This
procedure is the same with the process in generating distance to city raster.
2.3.8 Reclassification
The rasters are reclassified using the Reclassify tool to the values of 0, 25, 50, 75, and 100 based on the criteria and
constraints of sitting the solar farm.
Table 1. Reclassification of cell values of each raster
New Solar Radiation Slope Distance to City Distance to
Hillshade Aspect Land-cover
Values (Wh/m2) (degrees) Proper (m) Substation (m)
100 >1.75x106 <3 202 - 254 Flat, S 0 - 4000 Shrubs/Grassland/Barren 0-2500
75 1.6x106 to 1.75x106 3 - 7 176 - 202 SW, W 4000 - 8000 --- 2500-7500
50 1.5x106 to 1.6 x106 7 - 13 151 - 176 SE 8000 - 12000 --- 7500-12500
25 <1.5 x106 13 - 21 122 - 151 E, NE 12000 - 16000 --- 12500-17500
0 ------ > 21 0 - 122 N, SW >16000 Other Classes >17500
2.4 Weighing and Analysis
The reclassified raster files are calculated according to the weights of each factor using the Weighted Overlay tool.
The weights or the influence of solar radiation, slope, hillshade, aspect, land-cover, distance to city proper, and
distance to substations are 0.30, 0.10, 0.10, and 0.05, respectively.The scale values are set into 1,2,3, 4 and 5 if the
reclassified values are 100, 75, 50, 25, and 0 respectively. In the analysis of the initial output, the researcher also
considered the minimum area of 1 acre or 4046.86 m2.
3. RESULTS AND DISCUSSION
The initial results are shown in the figures below. In solar radiation factor, the output shown in Figure 2 is a
modelof Butuan City showing the incoming solar radiation from the loaded DTM and the cell value ranges from
134,673 wh/m2 to 1.8x106wh/m2. In slope factor, the output cell value ranges from 0 o to 85.34o and the result is
shown in Figure 3.The output cell value of hillshade model ranges from 0 to 254 from darkest to brightest and the
result is shown in Figure 4.In aspect factor, shown in Figure 5, the cell value of the model rangesfrom -1o to 360o
reckoned from north, also classified as flat, N, NE, E, SE, S, SW, W, and NW.
Figure 2. Solar Radiation Model of Butuan City Figure 3. Slope Model of Butuan City
Figure 4. Hillshade Model of Butuan City Figure 5. Aspect Model of Butuan City
Figure 6. Distance to Butuan City Proper Model Figure 7. Distance to Butuan City’s Substations
The maximum distance from the city proper, where the
extent is in the maximum x-coordinate and maximum y-
coordinate of the study area is 26,691.365m as shown in
Figure 6. In Figure 7, the locations of the substations in
Butuan City are shown as well as the generated distance to
substations model from the said shapefile. The land-cover
raster, shown in Figure 8, has seven (7) classes: River/Sea,
Ponds, Built-ups, Barren land, Shrubs/Grassland, Ricefield,
and Forest. The values brought-out from combining the
land-cover and the raster converted from the ground truth
points are shown in Table 2. The accuracies of the land-
cover classification based on the equations no. 3 to no.9 are
shown in Table 3.
Figure 8. Land-cover Model of Butuan City (March
18, 2015)
Table 2. The Matrix of the Ground Truth and the Land-cover of Butuan City
GROUND TRUTH to No.of Training
Land-cover
Raster1 Raster2 Raster 3 Raster 4 Raster 5 Raster 6 Raster 7 Points
River/Sea 65 2 0 0 0 0 0 67
Ponds 3 63 1 2 1 2 0 72
Built-ups 0 3 68 1 1 0 0 73
Barren land 2 2 1 64 6 2 0 77
Shrubs/Grassland 0 0 0 3 60 1 1 64
Ricefield 0 0 0 1 2 63 2 68
Forest 0 0 0 0 0 2 67 69
No. of GT points 70 70 70 70 70 70 70 490
Table 3. The accuracy of the land-cover
Class Producer’s User’s Over-all Kappa
Land-cover Commission Omission
Accuracy Accuracy Accuracy Accuracy Coefficient
River/Sea 92.86 92.86 97.01 2.98 7.14
Ponds 90 90 87.5 12.5 10
Built-ups 97.14 97.14 93.15 6.85 2.86
Barren land 91.43 91.43 83.12 16.88 8.57 91.84 90.48
Shrubs/Grassland 85.71 85.71 93.75 6.25 14.29
Ricefield 90 90 92.65 7.35 10
Forest 95.71 95.71 97.10 2.90 4.29
The reclassified rasters are shown in Figure 9, Figure 10, Figure 11, Figure 12, Figure 13, Figure 14, and Figure 15
and the new values are 0, 25, 50, 75, and 100 based on the criteria and constraints in Table 1.
Figure 9.The Reclassified Solar Radiation Model Figure 10. The Reclassified Slope Model
Figure 11.The Reclassified Hillshade Model Figure 12.The Reclassified Aspect Model
Figure 13. Reclassified Distance to City Proper Model Figure 14. Reclassified Distance to Substations Model
Figure 15.The Reclassified Land-cover Model
The result from the multi-criteria weighing of the seven reclassified rasters considering the influences of each factor
of sitting the solar farms is shown in Figure 16. The most favourable areas for solar farmsin Butuan City with
18015 pixels are in black pixels. Since the pixel size is 30m by 30m, then the total suitable areas for solar farms in
Butuan City is 16,213,500m2 or 1621.35 has.
Figure 16. Showing the Suitable Areas for Solar Farms in Butuan City(1 to 4, most favourable to less favourable)
From the 273 polygons (connected most suitable pixels), the researcher founds out that only 171 polygons have a
minimum area of 1 acre. Through another researcher’s analysis base on the orthophoto image, the table below
shows the initial result.
Table 4. Location of suitable areas for solar farms
Number of Polygons
Location Need Field Validation
Good
Questionable No data on Ortho
Ampayon 4 2 6
Taguibo 1 0 2
Sumilihon 0 0 1
Cabcabon 0 1 1
Tiniwisan 1 0 0
Baan Km3 5 1 0
Lemon 2 0 0
Basag 1 0 1
Camayahan 0 0 2
Pigdaulan 0 2 0
Mahay 1 0 0
Pangabugan 2 0 0
Villa Kananga 1 2 0
Bonbon 0 1 0
City Proper 1 1 0
Doongan 2 1 0
Libertad 5 0 0
Ambago 2 1 0
Bancasi 0 0 7
Pinamanculan 0 0 4
Total number of Polygons 28 12 24
The land-cover verified sites which is good for solar farm is in the following barangays: Libertad, Baan Km3,
Ampayon, Ambago, Doongan, Pangabugan, Lemon, Taguibo, Tiniwisan, Basag, Mahay, Villa Kananga, and City
Proper.The total number of polygons left after the analysis is 54 including the sites that needed field validation base
on its land-cover.
4. CONCLUSION
Analyzing the optimal location for solar farms required to locate the suitable areas for solar farms. In finding the
suitable areas, the researcher should consider the factors that would affect the amount of solar energy collected by
the farm and the factor that affect the capital of sitting the solar farm. These factors are solar radiation, slope,
hillshade, aspect, distance to city proper, land-cover and distance to substations. In the city of Butuan, out
of1621.35 hectares- total area suitable for solar farms, only 26.91 hectares are found to be good for sitting of solar
farm.
ACKNOWLEDGEMENT
The researcher acknowledged the CSU Phil-Lidar 1 project and ANECO for the data provided to the researcher.
REFERENCES
Belfiore, F., Taylor, T., Moisan, B., Zappia, M., Cinarelli, E., 2013. Risks and Opportunities in the Operation of
Large Solar Plants, pp. 1-8.
Hala Adel Effat, 2013. Selection of the Potential Sites for Solar Energy Farms in Ismailia Governotate, Egypt using
SRTM and MulticriteriaAnalsis, v2. 1, pp 205-220
Jason R. Janke, 2009.Multicriteria GIS modeling of wind and solar farms in Colorado, pp. 2228-2234.
Marcus Mckinney, 2014. Site Suitability Analysis for a Solar Farm in Watauga County, NC, v4. 1, pp. 1-14.
MasoudMomeni, Ryan Pemberton, Xi Yang, 2012. Statewide Analysis for Optimized Solar Farm Location
