Solar Irradiance Forecasting Meta-review: An In-Depth Systematic Meta-review on Solar Irradiance Forecasting Datasets, Predictive Pipelines and Performance

Overview

Solar irradiance forecasting is central to bringing more solar power into existing grids, but the research on it is spread across a large, fragmented body of reviews. That makes it hard to see which methods work, on which data and over which forecasting horizon.

This meta-review analyses 72 review articles through a structured, reproducible screening framework. It organises the forecasting horizons, compiles benchmark datasets and open-source tools from around the world, and sorts forecasting models into eight families, summarising the methods, metrics and best reported results for each.

It also maps where the field publishes and closes with the open research gaps, pointing to explainable AI and physics-informed neural networks as the most promising directions.

Key Findings

• —72 review articles screened and analysed through a structured, reproducible framework.
• —Forecasting horizons organised into very short-term, short-term, medium-term and long-term.
• —Models classified into eight families: physical, statistical, machine learning, deep learning, ensemble, hybrid, fuzzy logic and persistence.
• —A repository of benchmark datasets and open-source tools, to make results reproducible and comparable.
• —Explainable AI and physics-informed neural networks identified as the key directions for future work.

This is our summary of the paper. The full text, with its abstract, figures and data, is on the publisher’s site.