Deep FS: A Deep Learning Approach for Surface Solar Radiation.

Area of Science:

• Environmental Science
• Climate Science
• Renewable Energy

Background:

• Climate change presents significant environmental challenges, with drastic climate shifts being a primary driver.
• Accurate prediction of solar radiation, specifically Global Horizontal Irradiance (GHI), is essential for solar energy applications and meteorological forecasting.
• GHI is influenced by atmospheric conditions, geography, and temporal dynamics.

Purpose of the Study:

• To introduce a novel methodology for estimating surface sun exposure using advanced deep learning.
• To validate the proposed method using the NASA SORCE dataset.
• To enhance the accuracy of solar radiation prediction for improved solar energy and forecasting applications.

Main Methods:

• Feature extraction using a deep learning method, Deep-FS, to identify the most relevant predictors for surface exposure.
• Time series analysis employing Convolutional Neural Networks (CNNs) for prediction.
• Validation and comparison against traditional methodologies using standard performance metrics.

Main Results:

• The Deep-FS method successfully extracted key features crucial for accurate prediction.
• CNNs demonstrated superior performance in time series analysis for solar radiation prediction.
• The proposed deep learning approach significantly outperformed traditional methods across standard performance metrics.

Conclusions:

• The developed deep learning methodology, integrating Deep-FS and CNNs, provides a highly effective approach for estimating surface sun exposure.
• The novel method offers improved accuracy in Global Horizontal Irradiance (GHI) prediction compared to existing techniques.
• This advancement holds significant potential for optimizing solar energy utilization and meteorological forecasting.