American scientists have developed a method to predict solar radiation at any location using a single high-resolution hemisphere image captured on site. By extracting information about the sky, sun and surrounding scenes from visual cues, this approach enables accurate long-term energy forecasts without relying on detailed 3D city models.
Researchers at Columbia University in the United States have developed a new technique to measure solar radiation at any location using a single rectangular image.
The method uses a single hemispherical image taken at the panel location to infer scene geometry, sky visibility and lighting conditions, which are then used to predict future solar radiation over time.
“Our method can be used in two situations,” said the corresponding author Shree K. Nayar told pv magazine. “First, before a panel is installed in a specific location on a roof or vertical wall, the method can be used to determine the annual energy that the panel would produce. Second, in the case of a panel mounted on a pole in an urban canyon, we can determine what the best orientation of the panel would be.”
Nayar explained that the only other way to get energy forecasts for solar panels is to use 3D city models. In this case, graphical simulations are used to estimate the energy received by a panel.
“Unfortunately, these 3D models are simply not accurate enough to provide accurate energy estimates,” he continued. “This is because the reflections and shadows on a panel are affected not only by large buildings, but also by small objects such as signs, HVAC vents, parapets and window sills, which are often missing in 3D scans of cities. A small vent close to a panel can have the same effect as a large building across the street.”
The new approach uses a high-resolution, 360°, high-dynamic-range, in-situ image that is said to capture details of the surrounding structures, small and large, while also providing useful clues regarding the reflections of the structures around the panel.
Because inertial sensors (IMUs) are unreliable in urban environments due to magnetic disturbances, this method relies on visual signals instead. These are observable features in an image, such as shadows, edges, textures, lighting patterns, and structural lines, that provide information about the geometry, orientation, and lighting of a scene.
Image: Nayar Lab/Columbia Engineering
The proposed technique is based on a neural network trained to predict both the sun’s direction and gravity based on the image. The training is performed on large-scale synthetic hemispheric images and refined on real urban datasets. Once the sun and gravity vectors in the camera space are estimated, they are compared to their counterparts in the Earth frame, calculated from time, date and GPS.
The method predicts solar radiation by modeling both sky and scene lighting over time. This makes it possible to predict how the contribution to the sky will change with future solar positions and weather conditions. The sun’s contribution is calculated geometrically based on whether it falls within the visible sky opening.
At the same time, the irradiance of surrounding buildings is modeled separately, using the insight that the illumination of the scene varies smoothly and is scale invariant. Finally, the total irradiance is obtained by adding the components of the sun, sky and landscape over time.
“One of the interesting aspects of our algorithm for estimating irradiance on a panel is that it not only calculates the contribution of the sky, but also the contribution of the structures surrounding the panel,” Nayar points out. “We call the latter the scene irradiance component and have found that on average it represents about 12% of the energy received by the panel.”
The method was validated in four urban environments using a single hemispherical image per location and ground truth measurements from a pyranometer. On rooftops and deep urban canyons, the model was found to track daily radiation “accurately” under clear, partly cloudy and overcast conditions. The sharp changes caused by the sun entering or leaving the sky opening are captured “correctly,” especially when the sky is unobstructed. Errors generally remain low, although they increase slightly in more complex lighting conditions.
“This technology is cheap and portable, meaning it can be used by both homeowners and large companies during the planning phase of a solar project to maximize the return on their investment,” Nayar said. “Right now, that process is slow, expensive and approximate at best.”
“Our approach can also be used to assess panels mounted on vertical walls. Facades of tall buildings offer a great opportunity because they have a much larger surface area than roofs. The sides of tall buildings often receive more direct sunlight than the roof,” he concluded.
The new methodology was presented in “Solar energy prediction using a single image”, published in Solar energy.
This content is copyrighted and may not be reused. If you would like to collaborate with us and reuse some of our content, please contact: editors@pv-magazine.com.
Popular content
