Mathematical image analysis techniques are implemented in a novel manner to analyze aerosol particles in scanning electron microscope (SEM) data. By evaluating and removing the variable background in a digital SEM image using dynamic thresholding and then calculating the area and perimeter of each particle, one can study the relationship between particle area and particle shape. It is possible to distinguish in this manner between at least two different particle populations, separated by their fractal dimension, in a set of samples collected during 1 day in a heavy dust storm in Israel. ¿ 2001 American Geophysical Union