Characterizing the Particle Sizes and Shapes of Mount St. Helens Ash

Introduction

On May 18, 1980, the most destructive volcanic event in US history occurred when Mount St. Helens experienced a lateral eruption in Skamania County, Washington in the Pacific Northwestern region of the United States.

The eruption was triggered by a magnitude 5.1 earthquake and subsequent magma rise. The eruption began with a debris avalanche, which reduced the mountain’s summit from 9,677 feet to 8,363 feet, and was followed by a lateral blast that left a one-mile-wide horseshoe-shaped crater in the center of the mountain and scorched land up to 17 miles away. The column of ash was 15 miles high and covered over 20,000 square miles. Trace amounts of ash were found as far as 500 miles to the east in western Montana.

In this article, the particle size distributions were measured for three ash samples that were obtained from the 1980 Mount St. Helens eruption. The ash was purchased from a souvenir shop in Washington State during a family trip taken by the author when he was a boy. The desolation visited upon the area by the volcano was unforgettable. The samples were collected from distances of 5, 22, and 250 miles from the volcano vent, and were numbered 1 through 3, respectively. The author was curious to know how the particle sizes and shapes changed with increasing distance from the volcano.

Experimental Procedure

Each powder was dispersed onto a glass slide and imaged under a low-power stereomicroscope using coaxial illumination. Approximately 50 images were captured of each sample, and a semi-autonomous image processing method was used to threshold and measure the projected area of the particles in the images. Figure 1 shows a representative image of each sample captured at 60X alongside the thresholded image (right column) that was saved during processing. An extremely thin blue outline is drawn around the measured particles that were included in the sizing. The scale bar and particles that were touching the image boundary were excluded from the measurement area.

Results and Discussion

Figures 2-4 each display a particle size histogram and shape plot for each of the three samples. Some of the parameters were held constant across all three plots for sample comparison. For example, the bin size was set to 2 µm and the size range was also held constant. The y-axis of the particle size histogram is presented as the probability of finding a particle of that size in the size distribution.

The shape plot is a two-dimensional histogram where the color bar lists the number of particles that were observed in each colorized bin. One of the advantages of viewing the shape plots is that it can tell you if the shapes are more equant or acicular, which can inform the way particle size is measured and presented. For example, the shape plots observed here suggest that both the roundness and circularity of the particles are mostly greater than 0.5, which means the particles are fairly equant in shape. Therefore, equivalent spherical diameter is a suitable representation for the particle sizes. If the particle shapes were more acicular, then perhaps major and minor axes measurements would be a more accurate representation of particle size. Our image processing software calculates all of these parameters automatically from the images, so it is just a matter of choosing the right parameters that best describe a given sample.

Violin plots were then used to compare different features of the particle sizes and shapes across all three samples. These plots combine a few helpful statistical parameters into a single visual representation. The black line down the middle is actually a small box plot outlining the quartiles for each sample set with the white dot representing the median. A kernel distribution function is then displayed on either side, giving each plot its violin-like appearance. This makes it easy to visually compare size or aspect ratio distributions between samples. The mean, standard deviation, and quartiles including the min and max values for each distribution are also available in Table 1.

We are now able to compare these samples using the histograms, violin plots, and statistics table to gain insight into what might cause these three samples to be different. Both the mean and median particle sizes for each sample appear to be increasing as ash is collected further away from the volcano. This is a bit counterintuitive, since one would expect the further reaching particles to be smaller and lighter. However, there appear to be a large fraction of very small particles (15 µm and smaller) within the blast region. From the quartile percentages listed in Table 1, we see that over 75% of Sample 1, almost 50% of Sample 2, and about 25% of Sample 3 are made up of particles that were observed to be approximately 15 µm or smaller. We also notice from the shape plots in Figures 2 through 4 that Sample 1 has a wider variety of shapes than the other two samples.

Concluding Remarks

As was stated in the introduction, these samples were purchased from a gift shop, so the methods for sample collection may have been less than ideal or scientifically rigorous. Nevertheless, performing particle size characterization on these samples was possible, and the results were somewhat unexpected. The large number of very fine particles so close to the volcano vent may be related to the blast, or it may be the result of sample collection error.

In a future article we will use a combination of X-ray diffraction and elemental analysis to explore the percentages of crystalline compounds present in these powder samples.