Experimental simulations of volcanic ash resuspension by wind under the effects of atmospheric humidity

Abstract

Ash deposited during volcanic eruptions can be resuspended by wind and become hazardous for health and infrastructure hours to decades after an eruption. Accurate resuspension forecasting requires accurate modelling of the threshold friction velocity of the volcanic particles (U_th*), which is the key parameter controlling volcanic ash detachment by wind. Using an environmental wind tunnel facility this study provides much needed experimental data on volcanic particle resuspension, with the first systematic parameterization of U_th* for ash from the regions Campi Flegrei in Italy and also Eyjafjallajökull in Iceland. In this study atmospheric relative humidity (and related ash moisture content) was systematically varied, from <10% to >90%, which in the case of the Eyjafjallajökull fine ash (<63 μm) produced a twofold increase in U_th*. Using the Campi Flegrei fine ash (<63 μm) an increase in U_th* of only around a factor of 1.5 was observed. Reasonable agreement with force balance resuspension models was seen, which implied an increase in interparticle adhesion force of up to a factor of six due to high humidity. Our results imply that, contrary to dry conditions, one single modelling scheme may not satisfy the resuspension of volcanic ash from different eruptions under wet conditions.

Figure 1

Examples of volcanic ash resuspension by wind around the Eyjafjallajokull volcano (Iceland) in May 2010 (a), and by road cleaning at Sakurajima volcano (Japan) in July 2013 (b). External view of the environmental chamber at Aarhus University showing Laser Doppler Anemometry (LDA) measurements (c). Internal view showing the test section of the wind tunnel (40 cm in width). The LDA and sample exchange windows are indicated (d). View of the test section from the LDA window, including the position (contoured) of the 20 × 20 cm plate (e). The experimental sample plate loaded with ash before an experimental run (f), and during an intermediate removal step (g). Schematic of the wind tunnel orientation used in this study as well as the LDA position for shear stress measurement (located 2.1 cm above the floor of the test section) (h). All Photographs and drawings were created by the contributing authors.

Figure 2

Grain size distribution of the Campi Flegrei-Pomici Principali (CF, a–c) and Eyjafjallajokull (EY, d–i) ash samples in the three size classes (0–63 μm, 63–125 μm, and 125–250 μm) used in wind tunnel detachment experiments. In red, the cumulative distribution curves, and in the text boxes the distribution parameters. Note the fines-enriched and broader distribution of the 63–125 μm EY1 sample (e) with respect to the corresponding EY2 one (h).

Figure 3

Properties of the CF and EY ash samples starting material. Selected Scanning Electron Microscope images of ash thin sections after gray scale thresholding, showing particles area (black) and internal vesicles (a–b). Note the more vesicular appearance of the CF particles (a) with respect to the EY ones (b). Box diagram showing the median value (red line), the 25^th and 75^th percentiles (box size), 99.3 percent of data (whiskers), and outliers (red crosses) for the distribution of density and 2-D vesicularity of the particles as a function of grain size class. Open circles represent the density of the solid fraction (c–d). Variation of the moisture content of samples at RH 100% as a function of the sample size class (e). Error bars in the moisture reflects the weighting error. Note the divergent trend for the coarsest size class.

Figure 4

Removal curves for CF (a–c) and EY (d–f) ash samples of variable moisture content, sieved in the 0–63, 63–125 and 125–250 μm size classes. Each data point represents the cumulative mass of particles (error ± 0.3 wt. %) removed at the corresponding friction speed step (error ± 0.08 m/s), colour-coded by wind tunnel humidity range (RH). CF* indicate an experiment carried out with a dry sample (nominal moisture content 0 wt. %), in high humidity environment (RH 90–100%).

Figure 5

Measured threshold friction velocity U*_th (error of ±0.08 m/s, see text) for the different diameters d (median of each size distribution, with error bars marking the 25^th to the 75^th percentile range) for the CF and the two EY samples. The data are colour-coded by humidity range (RH) (a). Fit to the CF (b) and EY (c) data is provided by the Eq. 1 using fixed C_T and C_L coefficients for the minimum (red) and the maximum (blue) humidity ranges. The dash-dotted lines are for C_T = 0, where Eq. 1 equals the model of Shao and Lu. The dashed lines are for C_T = 1.6 × 10⁶ m⁻¹. Coloured arrows indicate overlapping data points.

Figure 6

Comparison between: i) our ‘wet’ and ‘dry’ experimental data points (symbols as per Fig. 5), ii) our model fits (Eq. 1) with coefficients from Table 1 (solid lines), and iii) literature models^,(dashed and dotted lines) with parameters previously used for the resuspension of volcanic ash particles by Folch et al.. For sake of visibility, literature models are only shown for the CF sample, differences between the two samples being negligible. The grey dashed line marks the U_th* = 0.4 m/s value previously used to model Eyjafjallajökull resuspension events^,,.