The purpose of this Master's thesis was to create a new model for screening possible optimal locations for utility-scale solar power plants (i.e. solar parks, solar power stations and solar farms) in larger city areas. The model can be used as a part of a decision making when examining site potentiality in a particular city of interest. The model includes forecasts for the year 2040. The main questions of the thesis are as follows: 1) What are the main criteria for a good location for a utility-scale solar power plant and 2) how to build a geographic information system (GIS) model for solar power plant location optimization? Solar power plants provide an alternative to producing renewable energy due to the enormous distribution potential of solar energy. A disadvantage of utility-scale solar energy production is the fact that it requires larger areas of land than the more traditional power plants. Converting land to solar farms might threaten both rich biodiversity and food production, which is why these factors are included in the model.
In this study, methods from the field of geographic information science were applied to quantitative location optimization. Spatial analytics and geostatistics, which are effective tools to narrow down optimal geographical areas, were applied for finding optimal locations for solar power plants, especially in larger city regions. The model was developed by an iterative approach. The resulting model was tested in Harare (Zimbabwe), Denver (United States) and Helsinki (Finland).
The optimization model is based on three raster datasets that are integrated through overlay analysis. The first one contains spatial solar radiation estimates for each month separately and is derived from a digital elevation model and monthly cloud cover estimates. The resulting radiation estimates are the core factor in estimating energy production. The second and the third dataset are two separate global datasets, which were used to deal with land use pressure issues. The first of these is a hierarchically classified land systems model based on land cover and intensiveness of agriculture and livestock, while the second is a nature conservation prioritization dataset, which shows the most important areas for conserving threatened vertebrate species. The integration of these datasets aims to facilitate smart and responsible land use planning and sustainability while providing information to support profitable investments.
The model is based on tools implemented in the ArcGIS 10 software. The Area solar radiation tool was used for calculating the global and direct radiation for each month separately on clear sky conditions. An estimate of the monthly cloud coverage was calculated from 30 years' empirical cloud data using a probability mapping technique. To produce the actual radiation estimates, the clear sky radiation estimates were improved using the cloud coverage estimates. Reclassifying the values from land use datasets enabled the exclusion of unsuitable areas from the output maps. Eventually, the integration and visualization of the datasets result in output maps for each month within a year. The maps are the end product of the model and they can be used to focus decision making on the most suitable areas for utility-scale solar power plants.
The model showed that the proportion of possible suitable areas was 40 % in Harare (original study area 40 000 km2), 55 % in Denver (90 000 km2) and 30 % in Helsinki (10 000 km2). This model did not exclude areas with low solar radiation potential. In Harare, the yearly variation in maximum radiation was low (100 kWh/m2/month), whereas in Denver it was 2.5-fold and in Helsinki 1.5-fold. The solar radiation variations within a single city were notable in Denver and Harare, but not in Helsinki. It is important to calculate radiation estimates using a digital elevation model and cloud coverage estimates rather than estimating the level of radiation in the atmosphere. This spatial information can be used for directing further investigations on potential sites for solar power plants. These further investigations could include land ownership, public policies and investment attractiveness.
