Table 2. Best-fit results obtained from reconstructing the CI super-eruption as a two-phase and single-phase event.
| Modelled dispersion parameters | Explored Range | TWO-PHASE |
SINGLE-PHASE | ||
|---|---|---|---|---|---|
| Plinian phase | Co-ignimbrite phase | Combined phases | Single phase event | ||
| Tephra mass (kg) | Calculated | 5.40 × 1013 | 1.54 × 1014 | 2.08 × 1014 | 2.11 × 1014 |
| Average deposit density (kg/m3)(a) | Assumed | 1,000 | 1,000 | 1,000 | 1,000 |
| Tephra volume (km3) | Calculated | 54 | 153.9 | 207.9 | 211.1 |
| Tephra volume DRE (km3) | Calculated | 22.6 | 61.6 | 84.2 | 84.4 |
| Duration (h) | 12–48 | 4 | 19 | 23 | 23 |
| Mass eruption rate (kg/s) | 108–1010 | 3.75 × 109 | 2.25 × 109 | 2.51 × 109(b) | 2.55 × 109 |
| Column height (km) | 20–50 | 44 | 37 | 38(b) | 38 |
| TGSD modes (Φ)(c) | 0–3/6–9 | 2.5(d) | 5(d) | — | 2.0/6.5(c) |
| TGSD variances (Φ)(c) | 1–3/1–3 | 1.16 | 1.22 | — | 2/2(c) |
| Suzuki coefficient A (−)(e) | 2–9 | 4 | 9 | 8(b) | 9 |
| Density of aggregates (kg/m3)(f) | 100–500 | 350 | 350 | 350 | 350 |
| Diameter of aggregates (in Φ–unit)(f) | 2–3 | 2.3 | 2.3 | 2.3 | 2.3 |
| Pearson correlation coefficient (R)(h) | Calculated | 0.76 | 0.83 | 0.81 | 0.79 |
| Root-mean-square error (RMSE)(h) | Calculated | 0.10 | 0.30 | 0.18 | 0.27 |
| Aida indexes K/k (−)(i) | Calculated | 1.01/1.05 | 1.03/1.07 | 1.02/1.14 | 1.02/1.15 |
The combined phase column is obtained by using the optimal ESPs resulting from the Plinian and co-ignimbrite phase inversions.
(a)This value is used to convert mass loading to deposit thickness and thereby to calculate tephra volume from mass, whereas a bulk density of 2500 kg/m3 was considered to convert into DRE volume.
(b)Weighted sum of input parameters for each phase.
(c)Total grain size distribution (TGSD) for the single phase reconstruction is assumed bi-Gaussian in Φ with maxima at μ1 and μ2 and corresponding variances σ1 and σ2.
(d)TGSD for the two-phase reconstruction was determined by Voronoi tessellation55.
(e)The eruption source is described in a purely empirical way using the Suzuki distribution30,36 for mass release along the column.
(f)Aggregation is accounted for using the empirical model of Cornell et al. (1983), assuming that 50% of the 63–44 μm ash, 75% of the 44–31 μm ash, and 95% of the less than 31 μm ash fell as aggregated particles, with diameter and density of aggregates found through the best-fit.
(h)Pearson correlation (R) and root-mean-square error (RMSE) based on the differences between log (measured thickness) and log (simulated thickness).
(i)Aida index for geometric average (K) and geometric standard deviation (k) of the distribution.