m-NLP Inference Models Using Simulation and Regression Techniques.

Area of Science:

• Space Physics
• Plasma Diagnostics
• Computational Physics

Background:

• Current multi-needle Langmuir probe (m-NLP) data analysis relies on Orbital Motion-Limited (OML) theory, which uses simplifying assumptions often unmet in real conditions.
• This leads to significant uncertainties in derived plasma parameters like electron density and satellite potential.
• Existing methods struggle with the complexities of actual experimental data, necessitating improved inference techniques.

Purpose of the Study:

• To develop and validate novel regression-based models for inferring plasma parameters from m-NLP data.
• To assess the performance of these new techniques against traditional methods using synthetic and real satellite data.
• To improve the accuracy and reliability of space plasma measurements from m-NLP instruments.

Main Methods:

• Utilized three-dimensional kinetic particle-in-cell simulations to generate a synthetic dataset for training and validation.
• Developed and trained regression models to infer electron density and satellite potentials from 4-tuples of currents.
• Applied the trained models to real data from the NorSat-1 satellite and compared results with existing techniques.

Main Results:

• Regression techniques achieved Root Mean Square (RMS) relative errors below 20% for plasma density inferences.
• Satellite potential inferences showed RMS errors less than 0.2 V for potentials ranging from -6 V to -1 V.
• The new methods demonstrated improved performance compared to state-of-the-art techniques when applied to NorSat-1 data.

Conclusions:

• The developed regression-based models offer a more robust and accurate approach to m-NLP data analysis.
• These advanced inference techniques effectively address the limitations of OML theory in practical space plasma environments.
• The findings pave the way for more reliable characterization of plasma properties in space missions.