11:45 AM - 12:00 PM
[MIS23-05] Characteristics of spatio-temp segragation of tephra particles from plumes of vulcanian eruptions at Sakurajima volcano
Keywords:Vulcanian eruption, Sakurajima, vertical profiles of tephra segregation, plume dynamics, disdrometer
1. Introduction
Observation-based vertical profiles of tephra segregation (hereafter “profiles”) are not investigated sufficiently. Conventionally, simulations have used functional models (e.g. Suzuki, 1983) and analytical models (e.g. Woods, 1988). Based on in-situ tephra fall measurements using disdrometers, Takishita et al. (2020) estimated that 6 Vulcanian eruptions at Sakurajima volcano show bimodal profiles. The higher peak corresponds to the top of the mushroom-shaped plume, and the lower one indicates the segregation from the lower part of the plume. This implies eruption phase transition from instantaneous explosion to quasi-steady emission of tephra. In this study, we analyzed larger number of eruptions to verify if such profiles are common among Vulcanian eruptions.
2. Observations and Eruption characteristics
We counted the number of tephra particles in every combination of diameter and settling velocity class minutely using a total of 17 optical disdrometers (OTT Parsivel2) installed around Sakurajima volcano to obtain load by the empirical formula (Takishita et al., 2022).
We analyzed 39 eruptions in 2018 and 2019 whose plume heights are recorded by JMA or DPRI, Kyoto University. They did not accompany the rain, and tephra fall was detected in at least 3 sites. Based on the minutely variation of mass eruption rate estimated from the seismic and ground deformation signals (Iguchi, 2016), the eruptions are classified into 4 types: isolated (Type I: 7 cases), in which particle ejections finish just after the explosion; quasi-stationary (Type QS: 7 cases), continuous 10 t/min particle ejections for tens of minutes; repeated (Type R: 18 cases), several 100 to 1000 t/min particle ejections repeated every couple of minutes; and combination (Type C: 7 cases).
3. Decision of profiles
The model suggested by Suzuki (1983) is adopted as the profile function. The set of profiles consists of various combinations of parameters of maximum height and the shape factor, and combinations of these profiles with various maximum heights. Each profile is adopted into the simulation using an advection-diffusion model Tephra4D (Takishita et al., 2021). Tephra4D assumes the settling velocity as the sum of terminal velocity and downwards wind. We consider the topography, temporal evolutions, and horizontal heterogeneity in the meteorological fields. The optimal profile is determined by residual evaluations in 4 groups of settling velocity classes (hereafter “slow particles”, “semi-slow particles”, “semi-fast particles” and “fast particles” respectively in order of lateness).
4. Results
The tephra fall predictions were estimated in the order of Type QS, I, and R with good accuracy. Many profiles have multiple modes, and most are the combinations of multiple unimodal profiles with different settling velocity, implying multiple modes result from settling velocity variations. Many profiles of Type QS eruptions have a lower peak. In Type R eruptions, many cases of the semi-fast particles have profiles with higher peaks and the number of profiles with higher and lower peaks is competitive in the semi-slow particles, reflecting the characteristics of the particle ejections in each type. In all types, dominant cases have a higher peak in the profiles of the semi-fast particles than of the semi-slow particles.
5. Discussions
The variations of peak height between different velocity groups are inconsistent with the analytical models. As the discussed settling velocity is limited to the several weight percent of erupted particles, further investigation is required.
We suggest that the variation of the estimation accuracy between eruption types is due to the assumption that all particles are ejected at the start of the eruption with at most two peaks of segregation heights. Such assumptions are most far from the reality in Type R eruptions, the type with the worst accurate estimations, suggesting that the temporal variation of profiles needs to be considered.
Observation-based vertical profiles of tephra segregation (hereafter “profiles”) are not investigated sufficiently. Conventionally, simulations have used functional models (e.g. Suzuki, 1983) and analytical models (e.g. Woods, 1988). Based on in-situ tephra fall measurements using disdrometers, Takishita et al. (2020) estimated that 6 Vulcanian eruptions at Sakurajima volcano show bimodal profiles. The higher peak corresponds to the top of the mushroom-shaped plume, and the lower one indicates the segregation from the lower part of the plume. This implies eruption phase transition from instantaneous explosion to quasi-steady emission of tephra. In this study, we analyzed larger number of eruptions to verify if such profiles are common among Vulcanian eruptions.
2. Observations and Eruption characteristics
We counted the number of tephra particles in every combination of diameter and settling velocity class minutely using a total of 17 optical disdrometers (OTT Parsivel2) installed around Sakurajima volcano to obtain load by the empirical formula (Takishita et al., 2022).
We analyzed 39 eruptions in 2018 and 2019 whose plume heights are recorded by JMA or DPRI, Kyoto University. They did not accompany the rain, and tephra fall was detected in at least 3 sites. Based on the minutely variation of mass eruption rate estimated from the seismic and ground deformation signals (Iguchi, 2016), the eruptions are classified into 4 types: isolated (Type I: 7 cases), in which particle ejections finish just after the explosion; quasi-stationary (Type QS: 7 cases), continuous 10 t/min particle ejections for tens of minutes; repeated (Type R: 18 cases), several 100 to 1000 t/min particle ejections repeated every couple of minutes; and combination (Type C: 7 cases).
3. Decision of profiles
The model suggested by Suzuki (1983) is adopted as the profile function. The set of profiles consists of various combinations of parameters of maximum height and the shape factor, and combinations of these profiles with various maximum heights. Each profile is adopted into the simulation using an advection-diffusion model Tephra4D (Takishita et al., 2021). Tephra4D assumes the settling velocity as the sum of terminal velocity and downwards wind. We consider the topography, temporal evolutions, and horizontal heterogeneity in the meteorological fields. The optimal profile is determined by residual evaluations in 4 groups of settling velocity classes (hereafter “slow particles”, “semi-slow particles”, “semi-fast particles” and “fast particles” respectively in order of lateness).
4. Results
The tephra fall predictions were estimated in the order of Type QS, I, and R with good accuracy. Many profiles have multiple modes, and most are the combinations of multiple unimodal profiles with different settling velocity, implying multiple modes result from settling velocity variations. Many profiles of Type QS eruptions have a lower peak. In Type R eruptions, many cases of the semi-fast particles have profiles with higher peaks and the number of profiles with higher and lower peaks is competitive in the semi-slow particles, reflecting the characteristics of the particle ejections in each type. In all types, dominant cases have a higher peak in the profiles of the semi-fast particles than of the semi-slow particles.
5. Discussions
The variations of peak height between different velocity groups are inconsistent with the analytical models. As the discussed settling velocity is limited to the several weight percent of erupted particles, further investigation is required.
We suggest that the variation of the estimation accuracy between eruption types is due to the assumption that all particles are ejected at the start of the eruption with at most two peaks of segregation heights. Such assumptions are most far from the reality in Type R eruptions, the type with the worst accurate estimations, suggesting that the temporal variation of profiles needs to be considered.