Large-scale rooftop solar photovoltaic technical potential estimation using Random Forests

Dan Assouline, Nahid Mohajeri and Jean-Louis Scartezzini

Applied Energy, 2018, vol. 217, issue C, 189-211

Abstract: Photovoltaic (PV) panels are a very promising technology that answers part of the increasing need for global renewable energy production, particularly in urban areas. We present a novel methodology combining Geographic Information Systems (GIS), solar models and a Machine Learning (ML) algorithm, Random Forests, to estimate the potential for rooftop PV solar energy at the scale of a country. We use a hierarchical approach which divides the computation of the final potential into several steps. Each step is reached by estimating multiple variables of interest using widely available data, and combining these variables into potential values. The method for estimating each variable of interest is as follows: (1) collect all the data related to the variable, (2) train a Random Forest model based on the collected data and (3) use the model to predict the variables in unknown locations. The variables of interest include available area for PV installation on rooftops, shape, slope and direction of rooftops, global solar horizontal and tilted radiations, as well as shading factors over rooftops. The study focuses on Switzerland and provides the rooftop PV technical potential for each (200 × 200) [m2] pixel of a grid covering the entire country. The methodology, however, is generalizable to any region for which similar data is available and could therefore be useful for researchers, energy service companies, stockholders and municipalities to assess the rooftop PV capacity of the region. Prediction Intervals are also provided for the different estimated variables, in order to measure the uncertainty of the estimations. The results show that Switzerland has a large potential for rooftop PV installations. More specifically, for roofs orientated at ±90° from due south, the total estimated potential PV electricity production is about 16.29 TWh/year, which corresponds to 25.3% of the total electricity demand in 2017.

Keywords: Rooftop photovoltaics; Solar energy potential; Geographic Information Systems; Machine Learning; Random Forests (search for similar items in EconPapers)
Date: 2018
References: View references in EconPapers View complete reference list from CitEc
Citations: View citations in EconPapers (47)

Downloads: (external link)
http://www.sciencedirect.com/science/article/pii/S0306261918302496
Full text for ScienceDirect subscribers only

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:eee:appene:v:217:y:2018:i:c:p:189-211

Ordering information: This journal article can be ordered from
http://www.elsevier.com/wps/find/journaldescription.cws_home/405891/bibliographic
http://www.elsevier. ... 405891/bibliographic

DOI: 10.1016/j.apenergy.2018.02.118

Access Statistics for this article

Applied Energy is currently edited by J. Yan

More articles in Applied Energy from Elsevier
Bibliographic data for series maintained by Catherine Liu ().