Abstract

As a response to the increasing demand for renewable energy, the global solar photovoltaic (PV) market is growing fast. In addition to PV systems, the energy from solar radiation can be used in solar thermal (ST) systems. Both use and installation rate of PV and ST systems have increased due to their scalability. The market share of distributed systems are about 40% and 60% for PV and ST systems respectively and the increasing number of systems and use of different records have made it difficult to keep track of both PV and ST deployment. As a result, new methods to assess installation volumes have been developed. One method is to identify and measure solar energy systems using aerial imagery and deep machine learning, creating a database with detailed information about the location.
This study aims to use this type of method, still in its’ development phase, to create and evaluate an inventory of solar energy systems in three Swedish municipalities and cross-check the identified systems with available databases and on-site inspections. In addition, socioeconomic and demographic data connected to the locations of the systems are analyzed to understand the relation between different variables, PV and ST adoption. Where the variables are age, sex, birth region, education, unemployment, average income and economic standard. Information about the locations also include owner, time at residence, tax value, purpose of property and purpose of building. The relation is evaluated and analyzed at three different granularity levels: households, demographic statistical areas and municipalities, with different variables available at different levels. A correlation analysis, including correlation matrices and regression analysis are performed. In addition, a sensitivity analysis evaluates the result from the analysis by removing outliers.
Out of 692 inventoried PV systems and 399 ST systems, the majority was installed in rural or regional center areas. The highest density of installed systems was found in rural areas and the most common buildings were residential and complementary. Most of the properties were owned by individuals, and in a larger extent for ST systems compared to PV systems. Also, the tax value of properties was in general lower for ST systems, indicating that it is more common for companies to install PV systems and at larger properties. The average income, age and percentage of males are higher for households with PV and/or ST systems compared to the municipalities in whole. However, the difference is clearer for PV systems compared to ST systems.
The correlation, regression and sensitivity analysis conclude that share of the population in age group 45-64 years, share of males in the population, share of the population born in Sweden and a high average income and economic standard have a positive impact on PV adoption. Share of the population within age group 25-44 years, unemployment and low economic standard have a negative influence on PV deployment. Positive impact on ST adoption is found for population within age group 45-64 years and share of the population born in Sweden. population in age group 24-44 years and unemployment have a negative impact on ST deployment.