A review on physical and data-driven based nowcasting methods using sky images

TL;DR

This paper reviews short-term solar nowcasting methods using sky images, highlighting the significance of various sky image features and comparing physical and data-driven approaches for solar irradiance prediction.

Contribution

It provides a comprehensive review of sky image-based nowcasting techniques and discusses key features influencing solar irradiance forecasting.

Findings

Sky image features significantly impact nowcasting accuracy

Comparison of physical and data-driven methods reveals strengths and limitations

Highlights the importance of real-time sky image analysis for reliable solar forecasts

Abstract

Amongst all the renewable energy resources (RES), solar is the most popular form of energy source and is of particular interest for its widely integration into the power grid. However, due to the intermittent nature of solar source, it is of the greatest significance to forecast solar irradiance to ensure uninterrupted and reliable power supply to serve the energy demand. There are several approaches to perform solar irradiance forecasting, for instance satellite-based methods, sky image-based methods, machine learning-based methods, and numerical weather prediction-based methods. In this paper, we present a review on short-term intra-hour solar prediction techniques known as nowcasting methods using sky images. Along with this, we also report and discuss which sky image features are significant for the nowcasting methods.