11:00 AM - 1:00 PM
[SVC31-P17] Relationship between eruption size and duration based on the characteristic shape of high-frequency eruption tremor waveforms
Keywords:explosive eruption, volcano seismology, volcano monitoring
Eruption size of sustained explosive eruptions such as Plinian eruptions is generally considered to increase with eruption duration (T). In these eruptions, it is fundamentally important to immediately estimate eruption size and duration during an eruption for residents and aviation safety. However, no methods to estimate them in real-time before the peak of an eruption have been established. Mori and Kumagai (GJI, 2019) showed that plume height (H) during explosive eruptions is scaled to seismic source amplitude (As) determined by using the amplitudes of high-frequency (5−10 Hz), in which S waves show isotropic radiation patterns due to highly heterogenous volcano structures. Mori and Kumagai (2019) also showed that As is proportional to eruption rate (q), suggesting that H and eruption volume (V) can be predicted by estimating As in a shortly time and cumulating As during an eruption, respectively. However, tremor waveform features and their relations to As and T are yet to be understood.
In this study, we performed detailed analysis of the waveform features of eruption tremor for sub-Plinian eruptions at Tungurahua (Ecuador), Pavlof (Alaska), and Kirishma (Japan). We applied a band-pass filter in 5-10 Hz to tremor waveforms and calculated their envelopes. We estimated the seismic source amplitude in every 10 s (Ask), which is the average amplitude in each window corrected for the effects of geometric spreading and intrinsic attenuation, to construct the source amplitude function (SAF) for each eruption. We also estimated As and cumulative source amplitudes (Is), which correspond to the maximum and cumulation of Ask during eruption tremor, respectively, and calculated the envelope width (p = Is/As) to evaluate the eruption tremor duration.
Our estimated SAFs are divided into three phases: the amplitudes gradually increase to the peak level (τ1), remain around the peak level (τ2), and decrease to the noise level (τ3). Based on the above feature, we assumed that the SAF shape can be approximated by trapezoid, where height (α), bottom, and area correspond to As, T and Is, respectively. We determined trapezoid shape by a parameter search by using cosine taper windows, and estimated the slope during τ1 (γ = α/τ1). We found power law relations among p, As and Is. We also found that τ2 and T tend to increase with decreasing γ, but T has little correlation with As and Is.
Mori and Kumagai (2019) theoretically derived the proportionality between As and q, which is expressed by a product of magma rise velocity (v) and cross-sectional area of a conduit (σ). Based on this source model, the trapezoidal shape of the SAFs can be interpreted as follows: (1) q and As gradually increase by opening the conduit and increasing v during τ1, (2) the conduit is stably open and q and As are kept around the peak level during τ2, and (3) the conduit closes and q and As decrease during τ3. Our results indicate that when γ is smaller, the duration to reach maximum values of σ and v is longer and T also tends to be longer. Thus, V and T of sustained explosive eruptions may depend on the peak values of σ and v and their increasing rates during τ1.
Our results suggest that the SAFs are useful to predict eruption size and duration and to understand eruption processes. We roughly reproduced observed durations of eruption tremors using the relation between γ and T. Therefore, when the SAF can be approximated by trapezoid shape, we may predict eruption duration by using the slope to the peak level during an amplitude increasing phase of eruption tremor. We need to investigate the SAFs for eruption tremor at various volcanoes to discuss the universality of our estimated relations.
In this study, we performed detailed analysis of the waveform features of eruption tremor for sub-Plinian eruptions at Tungurahua (Ecuador), Pavlof (Alaska), and Kirishma (Japan). We applied a band-pass filter in 5-10 Hz to tremor waveforms and calculated their envelopes. We estimated the seismic source amplitude in every 10 s (Ask), which is the average amplitude in each window corrected for the effects of geometric spreading and intrinsic attenuation, to construct the source amplitude function (SAF) for each eruption. We also estimated As and cumulative source amplitudes (Is), which correspond to the maximum and cumulation of Ask during eruption tremor, respectively, and calculated the envelope width (p = Is/As) to evaluate the eruption tremor duration.
Our estimated SAFs are divided into three phases: the amplitudes gradually increase to the peak level (τ1), remain around the peak level (τ2), and decrease to the noise level (τ3). Based on the above feature, we assumed that the SAF shape can be approximated by trapezoid, where height (α), bottom, and area correspond to As, T and Is, respectively. We determined trapezoid shape by a parameter search by using cosine taper windows, and estimated the slope during τ1 (γ = α/τ1). We found power law relations among p, As and Is. We also found that τ2 and T tend to increase with decreasing γ, but T has little correlation with As and Is.
Mori and Kumagai (2019) theoretically derived the proportionality between As and q, which is expressed by a product of magma rise velocity (v) and cross-sectional area of a conduit (σ). Based on this source model, the trapezoidal shape of the SAFs can be interpreted as follows: (1) q and As gradually increase by opening the conduit and increasing v during τ1, (2) the conduit is stably open and q and As are kept around the peak level during τ2, and (3) the conduit closes and q and As decrease during τ3. Our results indicate that when γ is smaller, the duration to reach maximum values of σ and v is longer and T also tends to be longer. Thus, V and T of sustained explosive eruptions may depend on the peak values of σ and v and their increasing rates during τ1.
Our results suggest that the SAFs are useful to predict eruption size and duration and to understand eruption processes. We roughly reproduced observed durations of eruption tremors using the relation between γ and T. Therefore, when the SAF can be approximated by trapezoid shape, we may predict eruption duration by using the slope to the peak level during an amplitude increasing phase of eruption tremor. We need to investigate the SAFs for eruption tremor at various volcanoes to discuss the universality of our estimated relations.