In the lush, sun-drenched landscapes of West Java, Indonesia, a groundbreaking study led by Ahmad Gufron from the Department of Geography, Universitas Indonesia, and the Department of Mechanical Engineering, Kookmin University, is shedding new light on solar energy potential. The research, published in Renewable and Sustainable Energy Transition, focuses on estimating hourly solar radiation with unprecedented accuracy, a critical factor for harnessing solar power in tropical regions.
The study addresses a significant challenge in solar energy development: the extreme fluctuations in solar radiation intensity. “Solar radiation can vary dramatically within short periods, making it difficult to predict and optimize solar power generation,” Gufron explains. To tackle this issue, Gufron and his team employed a sophisticated Long Short-Term Memory (LSTM) machine learning model, integrating data from eight meteorological stations and satellite imagery from the Geo-KOMPSAT-2A satellite.
The LSTM model, combined with the Inverse Distance Weighting (IDW) method for spatial interpolation, transforms point-based solar radiation estimates into a continuous spatial dataset. This approach not only enhances the accuracy of solar radiation predictions but also provides a more comprehensive understanding of solar energy distribution across West Java. “By incorporating spatial aspects, we can better identify optimal locations for solar farms and improve the overall efficiency of solar energy systems,” Gufron notes.
The model’s performance was impressive, achieving a Root Mean Square Error (RMSE) of 149.46 W/m² and a relative RMSE (rRMSE) of 39.99%. However, the study also highlighted some limitations, particularly in high-altitude areas where accuracy variations were more pronounced. These findings underscore the need for further research into hybrid machine learning models and advanced spatialized techniques to refine solar radiation estimation.
The implications of this research for the energy sector are profound. Accurate solar radiation estimation is crucial for the development of solar-based electricity, a renewable energy source that can significantly mitigate the environmental impacts of fossil fuel consumption. By providing a more precise and spatially detailed understanding of solar energy potential, this study paves the way for more efficient and effective solar power projects in West Java and beyond.
As the world continues to transition towards renewable energy, innovations like Gufron’s LSTM-IDW model will play a pivotal role in shaping the future of solar energy. By bridging the gap between data-driven predictions and practical applications, this research offers a promising pathway for enhancing solar energy development in tropical regions. The study, published in the journal Renewable and Sustainable Energy Transition, marks a significant step forward in the quest for sustainable energy solutions.
