Detection and tracking of barchan dunes using artificial intelligence.

Area of Science:

• Geology and Planetary Science
• Artificial Intelligence
• Remote Sensing

Background:

• Barchan dunes are crescent-shaped sand formations found on Earth and Mars.
• Satellite imagery and Artificial Intelligence (AI) are used for dune monitoring.
• Previous AI methods struggle to detect interacting barchans in fields.

Purpose of the Study:

• To develop and validate an AI model for automatic detection and tracking of interacting barchan dunes.
• To demonstrate the model's effectiveness on both experimental and satellite imagery from Earth and Mars.

Main Methods:

• Training a neural network using images from controlled experiments with interacting dunes.
• Applying the trained neural network to satellite images of barchan fields on Earth and Mars.
• Evaluating the model's performance using confidence scores (accuracy).

Main Results:

• The AI model successfully identifies and tracks interacting barchans in diverse environments and image types.
• Accuracy scores consistently exceeded 70% for detection and tracking of complex dune interactions.
• This represents a significant advancement over previous methods limited to isolated dunes.

Conclusions:

• A properly trained neural network can effectively monitor interacting barchans across different celestial bodies.
• This technology has significant implications for planetary exploration and terrestrial dune management.
• The AI model provides a robust tool for advancing our understanding of dune dynamics.