Multicut-HDMR with an application to an ionospheric model

Area of Science:

• Computational Physics
• Atmospheric Science
• Data Analysis

Background:

• High Dimensional Model Representation (HDMR) is a set of tools for analyzing high-dimensional input-output systems.
• Existing Cut-HDMR methods struggle with large input spaces where function expansion may not converge.
• Accurate modeling of complex systems requires efficient approximation techniques.

Purpose of the Study:

• Introduce and apply a novel Multicut-HDMR tool for enhanced model assessment.
• Address limitations of Cut-HDMR in approximating functions over large input domains.
• Improve the efficiency and accuracy of ionospheric electron density modeling.

Main Methods:

• Developed Multicut-HDMR, a new technique utilizing multiple reference points in the input space.
• Applied Multicut-HDMR to an ionospheric electron density model.
• Compared Multicut-HDMR performance against traditional HDMR methods.

Main Results:

• Multicut-HDMR effectively approximates the ionospheric electron density model.
• The new technique overcomes convergence issues associated with Cut-HDMR in large input spaces.
• Demonstrated improved accuracy and efficiency in deducing system behavior.

Conclusions:

• Multicut-HDMR offers a significant advancement for modeling complex systems, particularly in atmospheric science.
• The method provides a robust solution for approximating functions where traditional HDMR techniques fail.
• This tool enhances the capability for accurate ionospheric electron density prediction and analysis.