Aerosol Iron Solubility Specification in the Global Marine Atmosphere with Machine Learning

Area of Science:

• Atmospheric Chemistry
• Oceanography
• Artificial Intelligence

Background:

• Aerosol iron (Fe) solubility is vital for marine nutrient input assessment but poorly understood in models.
• The precise mechanisms governing aerosol iron solubility remain unclear, limiting model accuracy.

Purpose of the Study:

• To develop a deep learning model for projecting aerosol iron solubility.
• To identify key factors influencing aerosol iron solubility using observational data.

Main Methods:

• A deep learning model was developed using observational data from a coastal Chinese city.
• Input variables included particle size, relative humidity, and ratios of sulfate, nitrate, and oxalate to total iron (TFe).

Main Results:

• The model demonstrated excellent agreement (Pearson correlation coefficients r=0.73-0.97) with literature values across most global seas.
• The ratio of oxalate to total iron (oxalate/TFe) was identified as the most influential variable.
• The model showed good agreement in the Atlantic Ocean when trained with local data (r=0.34-0.66).

Conclusions:

• Deep learning models can effectively predict aerosol iron solubility using key atmospheric and chemical factors.
• The findings offer improved methods for quantifying aerosol soluble iron deposition in marine environments.
• The study highlights the potential of AI in atmospheric and oceanographic research.