Você está na página 1de 14

Ideas to improve pyroclast density

analysis

Benjamin Bernard1, Ulrich Kueppers2, Hugo Ortiz1


1
Instituto Geofísico, Escuela Politécnica Nacional, Quito - Ecuador
2
Ludwig-Maximilians-Universität München (LMU), – Germany

bbernard@igepn.edu.ec
Session V13E: What pyroclasts can tell us?
Pyroclasts speak a different
language
● Shape
● Bulk chemical
Composition
● Glass
composition
● Paragenesis
● Crystallinity
Cauliflower bomb from a PDC deposit
● Porosity at Tungurahua volcano
Porosity
● Fracture porosity produced by thermo-mechanic
processes (generally «5%)
● Vesicularity due to exsolution of volatiles (up to
90%)
ρP
● Porosity is calculated by: Φ=1−
ρDR
– Where ρP = pyroclast density and ρDR = dense rock
density
➢ Problem: ρP is not unique (ρDR neither but to a
lesser extent ≈2%)
➢ Solution: make lots of measurements
Classical density analysis
Houghton and Wilson (1989)
● About 30 pyroclasts
(but typically n > 100)
of the 16-32 mm
fraction
● Density histogram
● Density averages
● Standard deviation
● Density range
Density distributions of pyroclasts
➢ Problem: ρ is an from pumice fall and flow deposits,
Mount St. Helens, June 12th 1980
intensive property (Cashman and McConnell 2005)
➢ Solution: weighting
Weighting equation
mP
● Density: ρ P=
VP

VP
● Pyroclast Representativeness: RV = P n

∑Vi
i=1

n ∑ mi
● Weighted average: ρ˙V =∑ (RV ×ρi )= i=1
i n
i=1
∑Vi
i=1
Is there a big difference?
chachimbiro unzen ➢ We used two large datasets
2500
from Unzen (dome collapse)
Density frequency average (kg/m³)

2400 and Chachimbiro (dome


2300
lateral blast)

2200
➢ Higher difference with the
classical method (up to ± 4%;
2100
Bernard et al., 2014) than
2000 with large pyroclasts (< ±
1900
2%; Kueppers et al., 2005)
1800
➢ Absolute error is not as
1700
important as relative
1700 1800 1900 2000 2100 2200 2300 2400 2500 difference, because the latter
Density weighted average (kg/m³) lead to misinterpretations
Is there a big difference?

CHA-009-H CHA-107-A

➢ For the classical method, significant modification of the histogram


shape such as fluctuation of the primary and secondary porosity
mode
Is there a big difference?

MP-14 MP-11

➢ Using large pyroclasts (> 100 g, ~ 5-6 cm), similar effects with a lower
magnitude
➢ Frequency analysis degrades the results
How to determine if the sample size
is large enough?
● Sample size is primarly
dependant on the Frequency
dispersion of density Weighted by volume

data => standard


deviation
● Stability analysis can
help to quantify the
quality of the sampling
● Important: avoid
analytical skew
Stability analysis for the sample CHA-201-A
(n = 103)
How to determine if the sample size
is large enough?
● Sample size is primarly
dependant on the Frequency
dispersion of density Weighted by volume

data => standard


deviation
● Stability analysis can
help to quantify the
quality of the sampling
● Important: avoid
analytical skew
Stability analysis for the sample MP-01
(n = 30)
How many measurements are
required to stabilize the data?
3
Average slope (in %) below 5% (abs error)

2.5

1.5

0.5

0
10 20 30 40 50 60 70 80 90 100 110

Number of measurements

➢ For Chachimbiro data set, over 50 measurements, the density


average is stable (less than 1% every 2 measurements)
How many measurements are
required to stabilize the data?
3
Average slope (%) below 5% (abs error)

2.5

1.5

0.5

0
45 50 55 60 65 70 75 80

Total weight of the sample (kg)

➢ For Unzen data set (large pyroclasts), all samples have a good
stability
Conclusions
● Porosity and density data are critical to understand
volcanic processes
● Frequency analysis degrades the data and can
lead to misinterpretation
● This can be easily corrected using weighting
equation
● Stability analysis can be used to assess the quality
of the sampling and improve lab-time efficiency
● Using large pyroclasts allows a better reliability of
the results
Thanks for your attention!!

Explosion at Tungurahua volcano (08-02-2014)

Você também pode gostar