A new approach to determine solar potential using terrestrial images

Before installing Photovoltaic (PV) panels at a place it is important to estimate the solar potential of the place. Most approaches available in literature do not consider the presence of surrounding obstructions, thus leading to wrong estimates. Light Detection and Ranging (LiDAR) data or 3D (Geographic Information System) GIS based approaches consider obstructions but prove cost-effective only at city-wide scales. This paper presents a cost-effective, accurate, and scalable approach for this purpose. The proposed approach utilizes terrestrial images of surroundings and identifies obstructions present therein. Using the azimuth and elevation angles of the principal axes of the terrestrial images the azimuth and elevation angles of each pixel in these images are determined. Using thresholding and morphological closing the terrestrial images are segmented into sky and non-sky zones. Sun is considered visible at the point of interest if it lies in the sky zone. The position of the Sun is determined using the solar position algorithm and the irradiance reaching the Earth surface is computed using the modified radiative transfer model. Finally, the total irradiance over a chosen time period at a given location is estimated by integrating the irradiances for the duration when the Sun is visible.