You are on page 1of 20

Geochemistry, Geophysics, Geosystems

RESEARCH ARTICLE Morphology, Effusion Rates, and Petrology of Postglacial Lavas


10.1029/2018GC007817
of Laguna del Maule Volcanic Field, Chilean Andes, and
Key Points:
• We combined analyses of
Implications for Their Plumbing System
morphology and petrography of Francisco Cáceres1,2,3 , Ángelo Castruccio1,2 , and Miguel A. Parada1,2
postglacial lavas of Laguna del Maule
Volcanic Field to estimate eruption 1
Department of Geology, University of Chile, Santiago, Chile, 2Andean Geothermal Centre of Excellence, Santiago, Chile,
parameters 3
• We estimated effusion rates Department of Earth and Environmental Sciences, Ludwig-Maximilians-Universität München, Munich, Germany
3
<100 m /s and durations between 1
and 8 years
• Magma chamber volumes range Abstract We analyzed the dimensions, morphology, and petrography of nine postglacial lava flows,
3
between 6 and 30 km , and conduit ranging from andesites to rhyolites, located at the Laguna del Maule Volcanic Field in the Southern Andes
radius estimations are between 20
and 70 m
of Chile. Our data were used in conjunction with theoretical models of magma ascent and lava flow advance
to estimate time scales of lava emplacement, effusion rates, magma chamber volume, and conduit
dimensions. The results indicate that the parameters of these past eruptions correspond to eruptions of
Supporting Information:
• Supporting Information S1 months to a few years in duration, with peak effusion rates <100 m3/s, and were fed by magma chambers of
• Data Set S1 6 to 30 km3 and conduits of radius 20 to 70 m. Thermobarometric results indicate crystallization pressures
• Data Set S2 between 260 and 450 MPa and temperatures of 780 to 1060 °C for the analyzed units. These results, taken
together with recent geophysical studies, suggest that the most likely scenario, in case of an eruption,
Correspondence to:
Á. Castruccio, corresponds to a medium size eruption (0.1–1 km3). Initial activity would be explosive, followed by the
acastruc@ing.uchile.cl extrusion of lava flows that can last several months.

Citation:
Cáceres, F., Castruccio, Á., & Parada,
M. A. (2018). Morphology, effusion
1. Introduction
rates, and petrology of postglacial Laguna del Maule Volcanic Field (LdMVF) is located in the northern part of the Southern Volcanic Zone
lavas of Laguna del Maule Volcanic
Field, Chilean Andes, and implications (Figure 1) of the Andes of Chile and has attracted the attention of the scientific community in recent years
for their plumbing system. due to the extremely high rate of surface deformation registered since 2007 (up to 20 cm/year of uplift; Le
Geochemistry, Geophysics, Geosystems, Mével et al., 2016; Singer et al., 2014) that could be indicating the incubation of a large rhyolitic eruption.
19. https://doi.org/10.1029/
2018GC007817 This inflation rate has been faster than deformation associated with other large calderas (Singer et al.,
2014) and has raised the concern of inhabitants and local authorities due to the socioeconomic impacts of
Received 6 JUL 2018 a potential eruption.
Accepted 15 NOV 2018
Accepted article online 20 NOV 2018 LdMVF covers an area of about 500 km2 with an erupted volume of ~350 km3 of lavas and pyroclastic depos-
its from 130 basaltic-to-rhyolitic eruptive vents around Laguna del Maule lake, emplaced since Pleistocene
(Frey et al., 1984; Hildreth et al., 2010), with a prevalence of dacitic-rhyolitic units (4 basalts, 28 basaltic-
andesites, 33 andesites, 11 dacites, 23 rhyodacites, 21 rhyolites, and 6 ignimbrites ranging from andesite to
rhyolite; Hildreth et al., 2010). During the postglacial period (<25 ky), 36 rhyolitic and rhyodacitic lava flows
and domes were erupted from 24 eruptive vents around the lake (Hildreth et al., 2010) and outside the
caldera margin (with the exception of the Los Espejos unit). Despite the exceptional nature of LdMVF and
the recent abnormal signals of unrest, there are few detailed studies of the volcanic products and lava
morphologies in this volcanic area. The main studies have been focused on geological mapping (Hildreth
et al., 2010), petrogenesis (Frey et al., 1984), geochronology (Singer et al., 2000), and tephrostratigraphy
(Fierstein et al., 2012), and only recently has geophysical research (e.g., Cardona et al., 2018; Cordell et al.,
2018; Feigl et al., 2014; Le Mével et al., 2016; Miller et al., 2017) and more detailed petrologic studies
(Andersen et al., 2017, 2018) focused on the chronology and origin of postglacial silicic magmas that fed
eruptions at LdMVF.
In this study we analyzed nine postglacial lava flows, focusing on the rhyolitic and rhyodacitic units surround-
ing the lake, and also including a dacitic and an andesitic lava located to the west of the lake (units ars and
dcr, Figure 1). We chose these flows as they are the most voluminous, youngest, and best preserved lavas of
the volcanic complex. We combined field work, morphometric analyses using digital elevation models
©2018. American Geophysical Union. (DEMs), petrologic studies on selected samples, and theoretical analyses of magma ascent in a conduit and
All Rights Reserved. lava flow advance to estimate effusion rates and time scales for lava flow emplacement. Our results

CÁCERES ET AL. 1
Geochemistry, Geophysics, Geosystems 10.1029/2018GC007817

Figure 1. Location of the Laguna del Maule Volcanic Field and lava flows analyzed in this work. Lava units are labeled following the notation of Hildreth et al. (2010).
rle = Los Espejos; rln = Las Nieblas; rcl = Cari-Launa; rcd = Colada divisorial; rcb = Cerro Barrancas; rdcn = Northwest Coulée; dcr = Cordón Rodríguez; ars = Río
Saso. Arrows indicate direction of flow, and stars indicate eruption vents. Suffixes in some units indicate direction of flow and that there is more than one flow
associated to the vent. Red line is the international highway between Chile and Argentina. We included the age and error range of the units (from Andersen et al.,
2017) in the lower part of the figure.

indicate that an exponential decay of the effusion rate is compatible with the observed dimensions of the
lava flows, and we suggest that emplacement time scales were of months to years for these postglacial
lavas. We also estimate conduit and reservoir dimensions for multiple-level storage regions. We identified
heating events induced by mafic intrusions into more evolved reservoirs, producing reequilibration of
mineral phases, which is consistent with the presence of mafic enclaves. Our results give insights into the
possible processes and nature of a future volcanic eruption in this area.

2. Geology of LdMVF
LdMVF overlies Tertiary deformed volcanoclastic rocks and undeformed Pliocene to early Quaternary mafic to
intermediate lavas of the Cola de Zorro Formation (Hildreth et al., 2010; Vergara & Muñoz, 1982). Silicic
volcanic activity has been dated from at least 3.7 Ma, including two major eruptions at 1.5 and 0.99 Ma,
associated with the formation of the 12 × 8-km Bobadilla caldera (Andersen et al., 2017; Hildreth et al.,
2010). The postglacial (<25 ky; Singer et al., 2000) volcanic activity has emitted products with compositions
ranging from basaltic to rhyolitic. Two main volcanic phases of silicic composition have been recognized in
the postglacial period (Singer et al., 2014), separated by 5–10 ky of relatively small and less frequent
eruptions. Phase 1 started soon after deglaciation, with the emplacement of rhyolitic and andesitic lava flows
and rhyodacitic domes mainly on the western part of the lake, and culminated 19 ky (Andersen et al., 2017)
with the eruption of a rhyolite coulee (rle unit, Figure 1). Phase 2 started in the Holocene and was character-
ized by an increased eruptive rate and numerous eruptions, mostly of rhyolitic composition, that culminated

CÁCERES ET AL. 2
Geochemistry, Geophysics, Geosystems 10.1029/2018GC007817

at about 2 ky (Andersen et al., 2017, 2018) with lavas erupted at the southwestern part of the lake (see
timeline in Figure 1). Few mafic eruptions have been recognized during the last 20 ky (Andersen et al.,
2017, 2018; Singer et al., 2014), although mafic enclaves are common in the early postglacial rhyodacites.

3. Methods
We analyzed and estimated the effusion rates of the following lava flows (we followed the notation of
Hildreth et al., 2010, Figure 1): Rhyolites Colada Las Nieblas (rln), Loma de Los Espejos (rle), Cari-Launa (rcl),
Colada Divisoria (rcd), two lava flows from Cerro Barrancas (rcb), Coulee Northwest Coulee (rdcn) rhyodacite,
Cordón Rodriguez (dcr) dacite, and Río Saso (ars) andesite. We selected these units as they are the best
exposed lavas of LdMVF, with well-preserved primary morphologies and dimensions, and because these
units are simple lava flows (as defined by Walker, 1971) that facilitate the analysis of their emplacement.

3.1. Sampling and Morphometric Parameters of the Lavas and Dome


We measured the length, width, and thickness of each flow at several control points from the vent (Figure 2),
using the Shuttle Radar Topography Mission DEM (downloaded from https://earthexplorer.usgs.gov/). This
DEM has a horizontal resolution of ~30 m. Additional field measurements were carried out at rln, rle, rdcn,
dcr, and ars units. At these lavas, we observed structures and measured the thickness of the flow using a laser
distanciometer (Bushnell Elite 1500, maximum range of 1,500 m, error of ±1 m) and clinometer (error of ±1°).
The main source of uncertainty of thicknesses and terrain slope measurements is that the paleosurface is
buried by the lava itself, and we are interpolating the surface between both sides of the lava. Thus, our values
of thicknesses and volumes should be taken as minimum values, and we believe that based on the present-
state topography, we may be underestimating the thickness by ~10% of its calculated value.

3.2. Analytical Techniques


We collected samples from units rln, rle, rdcn, dcr, and ars to estimate crystal modal content by mineral
counting on thin-section images and scanning electron microscope (SEM)- back-scattered electron detector
(BSED) images using JMicroVision software. We counted 1,000 points per sample image separating the popu-
lation of crystal size between phenocrysts (>150 μm), microphenocrysts (150–50 μm), and microlites
(<50 μm). In section 5, we referred to crystal and bubble contents as vol%, although measurements were
done in 2-D.
Mineralogical studies were carried out using the Hitachi S-3500N SEM in the School of Earth Sciences at the
University of Bristol, UK, and the FEI Quanta-250 SEM in the Department of Geology at the University of
Chile. The mineral chemistry data were obtained using the CAMECA SX100 electron microprobe analyzer
(EMPA) at the University of Bristol, UK, and the JEOL JXA-8230 EMPA at Laboratorio de Microscopía
Electrónica y Análisis por Rayos X (LAMARX)-National University of Cordoba, Argentina. The analytical condi-
tions for minerals consisted of an accelerating potential of 20 kV and electron beam current of 10 nA with
counting times between 6 and 30 s depending on the mobility of analyzed elements and the half times for
peaks and background. An accelerating potential of 20 kV and electron beam current of 4 nA were used for
glass compositions using the same counting times as minerals for peaks and backgrounds. All EMPA analyses
are listed in supporting information.

4. Theoretical Framework of Magma Ascent and Lava Flow Advance


In this section we present the equations of magma ascent in a shallow conduit and the advance of a lava flow
over terrain. The main aim is to use lava morphology to estimate advance and effusion rates that will be then
used to constrain some of the plumbing system parameters of LdMVF. As all methods used to study mag-
matic systems have different degrees of uncertainties, we believe that our results can be compared and used
in conjunction with recent petrological and geophysical works, giving a more complete insight into the mag-
matic system under LdMVF. We are analyzing the effusive stage of eruptions. Consequently, we believe that
our approach of assuming a laminar flow inside a conduit, without considering fragmentation is appropriate.
It would be interesting to analyze in future works the explosive stage of these eruption to give a more com-
plete analysis of past eruptions at LdMVF.

CÁCERES ET AL. 3
Geochemistry, Geophysics, Geosystems 10.1029/2018GC007817

Figure 2. Cartoon with methodology to estimate eruption rate trends of lava units. Thickness and width of lava flows are measured at different locations. These data
are used in conjunction with slope and viscosity estimations to calculate velocity versus distance using Jeffrey’s equation (V = velocity, h = flow thickness, ρ = lava
density, α = terrain slope, μ = lava viscosity; see equation (6)). The results are translated to distance and effusion rate versus time.

Table 1
List of Symbols Used in the Text
Variable Unit Description Range Reference
1
B s Exponential constant for erupted volume
C Constant of yield strength in the crust model 1 Castruccio et al. (2013)
C1 m Constant for viscosity calculation
C2 m Constant for viscosity calculation
2
g m/s Gravitational acceleration 9.8
h m Lava height 5–140
H m Magma chamber depth 5,000–15,000
1 2
k kg·m ·s Combined magma and host rock bulk modulus 5e9–5e10
L m Lava length
Lf m Final lava length 3,500–10,000
1 2
ΔP kg·m ·s agma chamber overpressure 2.50E+07 Blake (1981)
3
Q m /s Effusion rate
3
Qo m /s Initial effusion rate 2e0–7e3
r m Conduit radius 20–70
t s Time
v m/s Lava flow velocity
vcrit m/s Critical velocity for the explosive-effusive transition 0.01 Loewen et al. (2017)
3
V m Lava volume 0.05e9–1.16e9
3
ΔV m Volume of injected magma into the magma chamber
3
Vc m Magma chamber volume 4e9–35e9
3
Ve m Erupted volume
3
ΔVi m Initial volume of injected magma
w m Lava width 250–3,200
α ° Terrain slope 0.5–22
2
κ m /s Thermal diffusivity 1.00E06 Castruccio et al. (2013)
1 1
μ kg·m ·s Magma and lava viscosity 1e6–4.5e12
1 1
μf kg·m ·s Final lava viscosity 2.5e8–4.5e12
1 1
μi kg·m ·s Initial lava viscosity 1e8–3e10
3
ρl kg/m Lava density 2,500 Castruccio et al. (2013)
3
Δρ kg/m Density contrast between magma and crust 0–500 Miller et al. (2017)
1 2
σ kg·m ·s Lava yield strength 1e5–1e6 Castruccio et al. (2013)

CÁCERES ET AL. 4
Geochemistry, Geophysics, Geosystems 10.1029/2018GC007817

4.1. Magma Ascent in a Cylindrical Conduit


Under laminar conditions and assuming small vertical variations of density and viscosity of magma, we can
write the volumetric flow rate Q, inside a cylindrical volcanic conduit as (e.g., Jaupart & Allegre, 1991):
 
πr 4 ΔP
Q¼ Δρg þ (1)
8μ H

where r is conduit radius, μ is magma viscosity, Δρ is the difference between the mean density of the crust
above the magma chamber and magma density, g is the gravitational acceleration, ΔP is the overpressure
over lithostatic pressure inside the magma chamber, and H is the distance from the top of the magma
chamber to the surface. A list with all symbols used in calculations are presented in Table 1.
We assume that overpressure is built by the injection of magma inside the magma chamber (Blake, 1981):

ΔV k
ΔP ¼ (2)
VC

where ΔV is the injected volume of magma, k is the coupled bulk modulus of magma and wall rocks (Huppert
& Woods, 2002), and Vc is the volume of the magma chamber. If the overpressure decays as magma is
released from the magma chamber during an eruption, we can write

½ΔV i  V e ðtÞk
ΔPðtÞ ¼ (3)
VC

where Vi is the initial additional volume of magma inside the chamber and Ve(t) is the erupted volume as a
function of time. Combining (1), (2), and (3), we can write an expression for the evolution of the effusion
rate Q(t):

Qðt Þ ¼ Qo expðBtÞ (4)

where
 
πr 4 kΔV i
Qo ¼ Δρg þ (4a)
8μ V cH

πr 4 k
B¼ (4b)
8μ V c H

and the erupted volume is

Vc
V e ðt→∞Þ ¼ ðΔPi þ ΔρgHÞ: (5)
k

Thus, if we have data of effusion rate versus time from an eruption, we can fit this data by adjusting
parameters Qo and B and use these parameters to calculate eruption parameters such as r and Vc if we have
values of parameters Δρ, μ, H, and ΔPi, which can be estimated from independent methods.
4.2. Lava Flow Advance
If we consider that a lava flow has a Newtonian rheology and the width (w) is much larger than its thickness h
(as is the case with all lavas of LdMVF), we can use the well-known Jeffrey’s equation to estimate its velocity:

h2 ρl g sinα
v¼ (6)

where ρl is lava density and α is the terrain slope. As a first approximation, we can write the volumetric flow
rate of the lava, Qf, as the product of the velocity and the cross-sectional area of the flow (h * w):

h3 wρl g sinα
Qf ¼ (7)

CÁCERES ET AL. 5
Geochemistry, Geophysics, Geosystems 10.1029/2018GC007817

We justify the use of a Newtonian rheology because the lavas have a low crystal content (see next section). If
we assume that lava flow rate approximates the eruption rate, we can use the dimensions of the flow,
together with lava density and viscosity and measurements of terrain slope to estimate eruption rate values
at different control points along the flow, which can be expressed as variations of effusion rate with
time (Figure 2).
Lava flow behavior can depart from Newtonian due to high crystal content or the development of an external
crust due to cooling. Yield strength, which is the minimum stress needed to start advance, can develop in
these cases. If the flow is not very thick and there is enough time for cooling (see Castruccio et al., 2013;
Castruccio & Contreras, 2016), lava flow advance can be controlled by the yield strength of the crust (σ)
instead of the internal viscosity. In this case the advance of the flow can be written as follows (Castruccio
et al., 2013):

 1 =2
V 2 ρg sinα
L¼C pffiffiffiffiffi (8)
W 2 σ κt

where V is the volume of the lava flow, κ is the thermal diffusivity (~106 m2/s), t is the time since the start of
the flow advance, and C is a constant of order 1. Considering that V is approximately LWT, we can rearrange
equation (8):
!2
1 ðCHÞ2 ρl g sinα
t¼ (9)
k σ

Thus, we can use the thicknesses of a lava flow at a given distance to find the time, t, taken by the flow to
reach it, if we know the parameters at the right side of the equation. Again, we can use these calculations
to obtain the variation of effusion rate with time (Figure 2).

5. Results
5.1. Lava Flow Morphometry and Structures
All the analyzed lavas have a blocky morphology (Figure 3 and supporting information). Estimated volumes
vary between 0.05 and 1.16 km3. We assumed a density of 2,500 kg/m3 for lavas; thus, our volume estimations
are dense-rock equivalents. Although we measured vesicle contents up to 50%, these values are from sam-
ples from the outer part of the lavas, while the interior (and thus most of the lava) is usually denser, without
vesicles. As discussed before, these lava volumes are minimum estimates, and we believe that these values
can be up to 10% higher. Units rln and rle have flow fronts that are part of the lake shoreline, but we do
not recognize lava below the lake surface. Unit rcl has some undetermined proportion of lava at the front
end below Cari-Launa lake surface; thus, the volume estimation of this unit has a higher underestimation.
Morphometric ranges for all lava units are as follows: Maximum lengths are between 3.5 and 10 km, widths
vary between 250 m and 3.2 km, and flow thicknesses reach up to 140 m. Terrain slopes vary between 22° and
0.5° and were measured on uncovered ground adjacent to lava flows. Usually, the largest lava units increase
their thickness downflow as the terrain slope is lower. For smaller and more channelized flows, the thickness
increases toward the front end. This difference is attributed to the different behavior of the flows, where
larger, unconfined flows have a Newtonian behavior, while smaller, more channelized flows are controlled
by the crustal yield strength (Castruccio et al., 2013; Castruccio & Contreras, 2016). The main morphological
structures that can be found in these flows are levees, ogives, and spines. Most lavas are glass rich (mostly
>70 vol%).
5.1.1. Las Nieblas Rhyolitic Coulee (rln)
This lava is located in the southwestern side of the lake and has an estimated volume of 1.16 km3. The
estimated age is <1.8 ± 0.9 ky according to Andersen et al. (2018). There is a 100-m-tall pumice cone
associated with this lava unit. The lava flow has a maximum length of 5 km (Figure 3a), and it is composed
of two subparallel lobes (western and eastern lobes, Figure 3b), which have variable widths of up to 3 km.
The thickness of the flow increases from less than 20 m near the vent to 140 m at the western lobe front
and ~60 m in the eastern lobe front. At the flow base there is a welded breccia (Figure 3f) comprising nearly

CÁCERES ET AL. 6
Geochemistry, Geophysics, Geosystems 10.1029/2018GC007817

3
Figure 3. (a) SRTM shaded image of rln unit. (b) Google Earth® image of the ~1-km rln unit. Arrows indicate the direction of flow. (c) Approximately 60-m-thick sec-
tion of the rln-w unit. Letters indicate position of Figures 3d, 3e and 3f. (d) Pumice block with flow banding at the top of the rln. (e) Flow banding in an obsidian block
from rln. (f) Welded breccia texture at the base of rln unit. Green star indicates the location of Figure 3c. SRTM = Shuttle Radar Topography Mission.

10% of the thickness. The core is the thickest part (~70–80% of the total thickness, Figure 3c) of the flow and is
composed of massive or banded obsidian (Figure 3e). The upper level shows a brown coloration and
corresponds to a breccia composed of obsidian and highly vesicular pumiceous blocks (Figure 3d). This lava
flow has a 73.4–74.0 wt% SiO2 content (Hildreth et al., 2010). The phenocryst abundance (Figure 4) varies
between 8 and 13 vol% including plagioclases, biotite, and Fe-Ti oxides. Microphenocrysts constitute
<1 vol%, and microlite content varies between 1 and 7 vol%. Plagioclase phenocrysts are oligoclases and
show a normal zoning. Vesicular, vitrophiritic, and spherulitic textures are common. Vesicle contents vary
between 5 and 37 vol%.
5.1.2. Loma de Los Espejos Rhyolitic Coulee (rle)
This coulee is located in the northern side of the lake and has a total erupted volume of 0.82 km3. It is dated at
~19.0 ± 0.7 ka (Andersen et al., 2017) by the 39Ar/40Ar method. It forms two lobes with lengths of ~2.3 km
(northern lobe) and ~3.5 km (southern lobe). The maximum width of both lobes is about 2.0–2.5 km and
decreases to ~0.5 km at the front. Maximum thicknesses are ~140 m (northern flow) and 114 m (southern
flow). The vertical structure of the flow is similar to the rln unit. The whole-rock composition of the lava is
75.9–77.0 wt% SiO2 (Hildreth et al., 2010). Lava samples show 1–6 vol% phenocrysts of plagioclase, biotite,
and Fe-Ti oxides and <1 to 3 vol% microlites inside a hyaline groundmass. The volume fraction of vesicles
(obtained by point counting on SEM images) varies from 9 to 50 vol%.

CÁCERES ET AL. 7
Geochemistry, Geophysics, Geosystems 10.1029/2018GC007817

Figure 4. (a) Mineral phases present as phenocrysts and microphenocrysts in each unit. (b) Total crystal content separated into phenocrysts (>150 μm), microphe-
nocrysts (150–50 μm), and microlites (<50 μm). Percentages normalized on a vesicle-free basis. Dashed lines indicate mean content of phenocrysts in (a) and total
crystal content in (b).

5.1.3. Other Rhyolitic Units


We also analyzed the morphometry of the rcl, rcd, and 2 lava flows from the rcb unit using the Shuttle Radar
Topography Mission DEM (Figure 1).
The rcl unit is located in the eastern side of the lake. It is composed of two units that flowed to the northeast
and southwest of the vent. It has an estimated total volume of 0.83 km3 (0.66 and 0.17, respectively).
According to Andersen et al. (2017), it has an age of <3.3 ka by the 39Ar/40Ar method. The northeast flow
has a maximum length of 3.1 km and a maximum width of 2.4 km, and its thickness reaches up to 160 m.
The southwest flow has a maximum length of 3.5 km and a maximum width of 1.4 km, and its thickness
reaches up to 60 m. According to Hildreth et al. (2010), it has a whole-rock composition of 73.6–73.8 wt% SiO2.

CÁCERES ET AL. 8
Geochemistry, Geophysics, Geosystems 10.1029/2018GC007817

The rcd unit is also located in the eastern side of the lake and to the south of rcl. It is composed of two units
that flowed to the northwest and southeast and has an estimated total volume of 0.63 km3 (0.29 and 0.34,
respectively). Andersen et al. (2017) estimated that it has an age of 2.1 ± 1.3 ka by the 39Ar/40Ar method.
The northwest flow has a maximum length of 3 km and a maximum width of 1.7 km, and its thicknesses reach
up to 95 m. The southwest flow has a maximum length of 4 km and a maximum width of 1.4 km, and its
thicknesses reach up to 112 m. According to Hildreth et al. (2010) it has a whole-rock composition of
74.1–74.2 wt% SiO2.
The rcb unit is composed of multiple lava flows, coulees, and pyroclastic cones (Hildreth et al., 2010). Lavas
from this eruptive complex range in age from 11.4 to 1.9 ka (Andersen et al., 2017). Here we analyzed two flows
that flowed to the north and south of the vents (Figure 1). The north flow has a maximum length of 5 km and
a maximum width of 2 km, and its thicknesses reach up to 82 m. The estimated volume is 0.4 km3. According
to Andersen et al. (2017), it has an age of 5.6 ka. The south flow has a maximum length of 5.2 km and a
maximum width of 1.8 km, and its thicknesses reach up to 117 m. The estimated volume is 0.61 km3.
5.1.4. Northwest Coulee Rhyodacitic Lava (rdcn Unit)
This coulee is located in the northwestern part of the lake and has a total volume of 0.25 km3. It has been
dated at 3.5 ± 2.3 ky by 39Ar/40Ar in plagioclases (Andersen et al., 2017). It flowed toward the northeast
and reached a distance of 3.5 km from the vent. The width of the flow varies between 0.5 and 2 km, and
its thicknesses are 75 m near the vent and 95 m at the front. Flow structures are similar to rhyolitic units.
The flow is a glass-rich rhyodacite with 68.5–69.4 wt% SiO2 (Hildreth et al., 2010). Phenocryst content varies
between 9 and 16 vol% including plagioclases, amphiboles, biotite, and Fe-Ti oxides (Figure 4).
Microphenocrysts correspond to 0–5 vol% of the rock, and microlite abundance is 19–41 vol%, mainly plagi-
oclase, clinopyroxene, and Fe-Ti oxides. Vesicles vary between 3 and 24 vol%. Some plagioclase and amphi-
bole phenocrysts have resorption features and reaction rims, respectively. Mafic enclaves up to ~50 cm in
diameter are common.
5.1.5. Cordón Rodríguez Dacitic Lava Flow (dcr)
This unit has the smallest volume of the analyzed blocky flows, with an estimated total volume of 0.05 km3. It
has not been, dated but it is assumed to be postglacial based on its fresh morphology (Hildreth et al., 2010). It
has a length of 4.5 km. The average width is 300 m, while thicknesses increase from 10 to 41 m at the front.
Most surface blocks are of about 1–3 m with few vesicles (<5 vol%). Compositionally, this flow is a dacite
(65.4 wt% SiO2; Hildreth et al., 2010) with abundant mafic enclaves (53.2 wt% SiO2; Hildreth et al., 2010)
and lithic juvenile fragments. Crystal content varies between 18 and 22 vol% phenocrysts of plagioclase, pyr-
oxenes, amphiboles, and Fe-Ti oxides (Figure 4). Microphenocrysts and microlites show a similar mineralogy
and with abundances of 0–8 and 20–32 vol%, respectively. Vesicle contents vary between 10 and 22 vol%.
Compositional zoning and disequilibrium textures are observed in plagioclases. Some amphibole pheno-
crysts have thick reaction rims (10–20 μm).
5.1.6. Río Saso Andesitic Lava Flow (ars)
This is the longest studied LdMVF lava with 10-km length and a volume of 0.11 km3. The unit flowed
westward from a scoria cone on the southern wall of a valley. It has not been dated, but it is assumed that
it was emplaced in the postglacial period, based on its fresh morphology (Hildreth et al., 2010). Widths do
not exceed 400 m in the first kilometers but progressively increase downflow up to 0.5–1.5 km. Both sides
of the flow are limited by the northern and southern walls of the valley. The lava has an average thickness
of 10–20 m in the first 1–4 km and reaches up to 60 m at the front. The surface of the lava is composed of
blocks between 50 cm and 2 m in diameter. They are homogeneous in terms of texture, showing abundant
microcrystals, scarce vesicles (<10 vol%), and a small number of mafic enclaves. The lava has a composition
of 59.5–61.6 wt% SiO2 (Hildreth et al., 2010) and 10–24 vol% phenocryst contents of plagioclase, skeletal
olivines, orthopyroxenes and clinopyroxenes, amphiboles, and Fe-Ti oxides (Figure 5). Microphenocrysts
and microlites of plagioclases, Fe-Ti oxides, and pyroxenes occupy 3 and 29–44 vol. %, respectively.
Plagioclases are slightly zoned and exhibit sieve and patch textures. Reabsorbed edges in plagioclases and
thick reaction rims in amphiboles are commonly observed.

5.2. Mineral Phases and Glass Composition of Lava Flows


Lava sample textures range from aphanitic to porphyritic, with 1–25 vol% of phenocrysts and 3–70 vol%
of total crystal content (Figures 4a and 4b). The dominant phase is plagioclase (60 to 90 vol% of

CÁCERES ET AL. 9
Geochemistry, Geophysics, Geosystems 10.1029/2018GC007817

Figure 5. (a and b) Textures and chemical composition of plagioclases in the Laguna del Maule Volcanic Field studied units. (left) Back-scatter electron (BSE) detector
images and (right) chemical profiles of representative plagioclase crystals in (a) rln unit and (b) rdcn unit. Red line indicates analyzed profiles. (c) Amphibole
of the rdcn unit. Composition is uniform, and there are no reaction rims. (d) Amphiboles of the ars unit with reaction rims of less than 10 μm. (e) Skeletal olivine of the
ars unit.

phenocryst content) followed by biotite in rhyolites, amphibole in rhyodacites and dacites, and pyroxene
and olivine in the andesite unit (Figure 4a). There are minor amounts of Fe-Ti in all the analyzed
samples. The groundmass mineralogy is similar, with plagioclase as the dominant phase. Complete
data sets of mineral and glass compositions are provided in the supporting information. We also
included analyses of rhyodacitic units rdcd (Colada Dendriforme rhyodacitic coulee) and rdsp (Laguna
Sin Puerto rhyodacitic dome), which were not analyzed in terms of effusion rates because they are
not simple lava flows.
Plagioclases show a wide compositional range with progressively more albitic compositions with increasing
whole-rock silica content (Figures 5a and 5b). Most plagioclase phenocrysts exhibit compositional zonation
and disequilibrium features (sieve, patch, partially resorbed rims) with variable intensities among individual
crystals. Plagioclases in rhyodacitic units (Figure 5b) exhibit large compositional changes between core
and rim (An30 to An50) and show the most complex zonation patterns and disequilibrium features. On the
other hand, rhyolite plagioclase phenocrysts have a more homogeneous composition (An25–An30) and less
evidence of disequilibrium (Figure 5a).

CÁCERES ET AL. 10
Geochemistry, Geophysics, Geosystems 10.1029/2018GC007817

Amphiboles occur as euhedral to subhedral phenocrysts (Figures 5c and 5d) and as microlites of varied
length (30–2,000 μm) in andesite, dacite, and rhyodacite units. They correspond to Mg-hastingsite,
Mg-tschermakite, and Mg-hornblende. Usually, they have well-preserved edges, but reabsorption effects
and internal embayment are present in some cases. Reabsorbed edges are common where they are in
contact with other minerals, generally plagioclases and other amphiboles. Reaction rims of variable widths
(2–60 μm) are observed in several amphiboles of the andesite, dacite, and rhyodacite lavas.
Olivine crystals are only present in the ars unit where they are scarce and exhibit 250- to 300-μm length
and a skeletal texture (Figure 5e). They have compositions ranging from Fo71 to Fo78, showing in some
cases a subtle zoning with higher Fo contents at cores. Clinopyroxene (augite) crystals are present in all
but rhyolitic units. These mainly occur as microphenocrysts with homogeneous augitic composition and
scarce microlites. Few orthopyroxene crystals (enstatite) are found in ars and rdsp units. These are pheno-
crysts of ~1-mm length with rounded edges, magnetite chadacrysts, and, in some cases, forming clots
with magnetite and plagioclase crystals. Fe-Ti oxides are present in all the studied units, mainly as euhe-
dral to subhedral magnetite and ilmenite crystals. Small amounts of elongated ilmenite crystals of homo-
geneous composition are found in the two rhyolitic units. These crystals have the same length as
the magnetites.
Glass compositions range between 68.1 and 77.5 wt% SiO2. Rhyolitic lavas show glass compositions in the
range 74.7–77.5 wt% SiO2 for rle and 75.5–77.5 wt% SiO2 for rln. The rhyodacitic lavas show glass composi-
tions of 74.9–75.4 wt% SiO2 for rdcd, 71.6–74.8 wt% SiO2 for rdcn, and 72.5–73.1 wt% SiO2 for rdsp dome.
The dcr unit shows a range of 71.6–72.4 wt% SiO2.

5.3. Thermobarometric Conditions


Temperature, pressure, H2Omelt, and fO2 were determined in amphiboles using the single-crystal method of
Ridolfi and Renzulli (2012). We preferred this method over that of Ridolfi et al. (2010) because it has a more
accurate multivariable statistical adjustment and the high values (>0.5) of Mg# (Mg/Mg + Fe2+) in LdMVF
amphiboles. Results are shown in Figure 6. All calculated results have the uncertainties provided by the
method: 11.5% for pressure, ±23.5 °C for temperature, ±0.78% for H2Omelt, and ±0.37 log units for ΔNNO.
Because amphiboles could be formed from different melts of the LdMVF system, we calculate the equilibrium
composition of the amphibole-crystallizing melts from the Ridolfi and Renzulli (2012) chemometric
equations. The results indicate that most of the calculated SiO2melt differs from the whole-rock SiO2 content,
suggesting that the corresponding amphiboles are antecrysts.
The rdcn unit show P-T conditions of 275–450 MPa and 933–1008 °C taken from euhedral and large
(~400 μm) amphibole crystals, whereas smaller crystals gave values in the range of 265–340 MPa and
933–975 °C. These P-T values are higher than those obtained from amphibole phenocrysts in equilibrium
with the host melt (180–250 MPa and 828–933 °C; Andersen et al., 2017) and consistent with the antecrys-
tic nature of our studied amphiboles. The calculated H2Omelt and ΔNNO values are in a narrow range of
4.2–4.8 wt% and 1.4–2.7, respectively, despite the associated pressure and temperature variations. The
estimated values of H2Omelt are within the range 4.4–6.0 wt% obtained from plagioclase hygrometer by
Andersen et al. (2018).
Amphiboles from dcr, gave two ranges of P-T conditions. The first is between 415–455 MPa and 1015–1032 °C,
and the second between 305–330 MPa and 1005–1010 °C. Associated with the first P-T range, calculated
H2Omelt and oxygen fugacities are in the range of 3.8–3.9 wt% and 1.1–1.2 Δ;NNO, respectively, whereas
values of 3.5–3.6 wt% H2Omelt and 0.9–1.0 ΔNNO are associated with the second range.
Amphiboles from the ars unit give ranges of 345–420 MPa and 990–1013 °C. The oxygen fugacity conditions
correspond to 1.4–1.7 ΔNNO. H2Omelt contents are in the range of 3.9–4.4 wt%.
Based on the Shejwalkar and Coogan (2013) single-crystal method, it is possible to obtain the crystallization
conditions for olivine phenocrysts, which are only present in the andesite. Because of the high uncertainties
for pressure estimations, we only use the temperature estimation. Temperature values are in the range of
1026–1064 °C (±20 °C), with the highest temperatures being those calculated for the skeletal olivine pheno-
crysts. It is interesting to note that these temperatures are up to 70–50 °C higher than those obtained from
amphibole crystallization.

CÁCERES ET AL. 11
Geochemistry, Geophysics, Geosystems 10.1029/2018GC007817

Figure 6. Calculated thermodynamic storage conditions from mineral chemistry for studied units. (a) Pressure-versus-temperature diagram. P-T conditions were
calculated using the method of Ridolfi and Renzulli (2012) in amphiboles (rdcd, rdcn, rdsp, dcr, and ars units). Olivine temperature range are shown by the
double-headed arrow with the method of Shejwalkar and Coogan (2013; ars unit). Solidus lines according to Holtz et al. (2001). Stability lines for amphiboles in
dacitic-andesitic and primitive magmas from Kiss et al. (2014). Gray arrow shows the general path of rhyodacites. (b) fO2 versus temperature calculated with
amphiboles and Fe-Ti oxides (rln and rle units; Ghiorso & Evans, 2008); (c) melt water content versus temperature calculated with amphiboles (filled symbols) and
plagioclase-liquid (crosses; Lange et al., 2009). Fields represent the maximum error of the shown data. Dashed line encloses the maximum error for water content and
temperature using the plagioclase-liquid method.

Estimation of temperatures and oxygen fugacities for the equilibrium Fe-Ti oxide pair crystallization was
obtained from the two rhyolite lavas using the Ghiorso and Evans (2008) approach. Magnetite-ilmenite pairs
that passed the equilibrium filter (Bacon & Hirschmann, 1988) gave temperature ranges of 782–798 °C for rln
unit and 763–774 °C for rle unit, while similar fO2 conditions of 1.21–1.28 ΔNNO were obtained in both units.
These temperatures and oxygen fugacity conditions are basically the same as those obtained by Andersen
et al. (2018) from rhyolite Fe-Ti oxides.

5.4. Effusion Rates of Lava Flows and Estimation of Plumbing System Parameters
We used the dimensions of lava flows (Figure 2) in conjunction with slope measurements and viscosity
estimates (Figure 7) to calculate the effusion rates and time scales of lava extrusion. We used equation (7)

CÁCERES ET AL. 12
Geochemistry, Geophysics, Geosystems 10.1029/2018GC007817

Figure 7. Calculated variations of melt viscosity with water content and three temperatures for samples of units (a) rln,
(b) rdcn, (c) dcr, and (d) ars. Calculations used the equations of Giordano et al. (2008).

to get the effusion rate at different control points along the flow. Lava units dcr and ars were not well fitted
with the Newtonian rheology, and in these cases the yield strength in the crust model provided a better fit to
the data, and effusion rates were deduced from equation (9).
Lava viscosity can be estimated using glass composition, temperature, and water content (Giordano et al.,
2008) in conjunction with crystal content (e.g., Mueller et al., 2009). However, viscosity estimations are
sensitive to small variations of water content and temperature. Additionally, the lava front viscosity, which
controls lava advance (Castruccio & Contreras, 2016), can be much higher than the viscosity of lava from
the channel that is feeding it. For these reasons we calculated effusion rates using a range of viscosity values
(Table 2). Apparent viscosities of lava flows can increase by orders of magnitude with distance. Castruccio
et al. (2013) showed examples where the advance of lava flows of different compositions can be modeled
with two viscosities for proximal and distal parts of the flow. Here we propose and use the following formula-
tion for the viscosity of the lava flows, which models an abrupt viscosity increase:
μ  μ  
Lf  C1

μ¼ f i
1 þ erf þ μi (10)
2 C2

where μi and μf are the initial and final viscosities, respectively, Lf is the final length of the lava flow, and C1
and C2 are constants. Thus, to fit the thickness of a lava flow unit, it is necessary to find parameters μi and μf,
plus Qo and B. With estimations of parameters Qo and B we can infer plumbing system information as follows.
By rearranging equation (4b), we can estimate magma chamber volume Vc. Equation (4a) can also be
rearranged to estimate the conduit radius using estimations of parameters ΔPi, k, and Δρ. As these para-
meters are not well constrained, we performed a Monte Carlo simulation with 10,000 random selections of
these parameters to find the mean value and standard deviation of parameters r and Vc (supporting informa-
tion). An independent estimation of r can be made following the approach of Loewen et al. (2017). Numerous
authors indicate that the transition between explosive and effusive activity occurs when the ascent velocity
of magma drops below 0.01 m/s (e.g., Castro & Gardner, 2008; Eichelberger et al., 1986; Melnik & Sparks,
1999). Thus, we can estimate the conduit radius using the initial effusion rate of the lava flow:
rffiffiffiffiffiffiffiffiffiffiffiffi
  2 Qo
Qo ¼ v crit π r →r ¼ (11)
v crit  π
where vcrit is the critical threshold of the ascent velocity when the explosive-effusive transition occurs
(~1 cm/s).

CÁCERES ET AL. 13
Geochemistry, Geophysics, Geosystems 10.1029/2018GC007817

Table 2
Values of Effusion Rates, Eruption Duration, Conduit, and Magma Chamber Dimensions for the Units Analyzed in This Work
Parameters Fit Results

Lava Qo Eruption Erupted Mean eruption Conduit radius Conduit radius Magma chamber
3 1 3 3 3
unit Visc 1 (Pa·s) Visc 2 (Pa·s) (m /s) B (s ) duration (days) volume (km ) rate (m /s) (m; equation (11)) (m; equation (4a)) volume (km )

rle-n 3.00E+08 8.00E+09 500 1.10E06 11 0.29 310 126 85 (17) 13 (4)
3.00E+09 8.00E+10 60 1.25E07 100 0.32 37 40 50 (10) 14 (5)
3.00E+10 8.00E+11 5 1.10E08 1,118 0.30 3 13 26 (5) 13 (4)
rle-s 3.00E+08 2.00E+09 530 1.48E06 18 0.32 207 130 86 (17) 11 (3)
3.00E+09 2.00E+10 52 1.44E07 190 0.33 20 41 48 (10) 11 (3)
3.00E+10 2.00E+11 7 1.83E08 1,650 0.35 2 15 29 (6) 11 (4)
rln-w 1.00E+08 2.00E+09 733 7.70E07 67 0.94 163 153 94 (19) 28 (9)
1.00E+09 1.00E+10 160 1.56E07 318 1.01 37 71 64 (13) 30 (10)
1.00E+10 8.00E+10 26 2.35E08 2,406 1.10 5 29 41 (8) 33 (11)
rln-e 1.00E+08 2.00E+09 638 2.76E06 17 0.23 155 142 90 (18) 7 (2)
1.00E+09 1.00E+10 68 2.99E07 97 0.21 25 47 51 (10) 7 (2)
1.00E+10 8.00E+10 6 2.58E08 930 0.20 3 14 28 (6) 7 (2)
rcb-n 5.00E+08 1.00E+09 584 1.45E06 36 0.40 128 136 88 (18) 12 (4)
5.00E+09 1.00E+10 59 1.45E07 345 0.40 13 43 50 (10) 12 (4)
3.00E+10 9.00E+10 10 2.30E08 2,480 0.43 2 18 32 (6) 13 (4)
rcb-s 2.00E+08 3.00E+09 637 1.17E06 45 0.54 139 142 90 (18) 16 (5)
2.00E+09 3.00E+10 52 9.24E08 372 0.53 17 41 48 (10) 17 (5)
2.00E+10 3.00E+11 10 1.53E08 2,980 0.64 2 18 32 (6) 19 (6)
rcd-n 3.00E+08 4.00E+10 144 4.95E07 86 0.28 38 68 62 (13) 9 (3)
3.00E+09 4.50E+11 12 4.52E08 1,310 0.26 2 20 33 (7) 8 (3)
3.00E+10 4.50E+12 2 6.40E09 9,625 0.31 0 8 21 (4) 9 (3)
rcd-s 8.00E+08 1.00E+10 766 1.80E06 9 0.32 412 156 94 (19) 13 (4)
5.00E+09 4.50E+10 137 3.00E07 52 0.34 75 71 61 (12) 14 (4)
5.00E+10 5.00E+11 10 2.00E08 580 0.32 6 22 32 (6) 15 (5)
rcl-s 5.00E+08 2.50E+08 1,391 7.30E06 10 0.19 220 210 110 (22) 6 (2)
5.00E+09 2.50E+09 125 6.70E07 95 0.19 23 63 60 (12) 6 (2)
5.00E+10 2.50E+10 14 7.30E08 950 0.19 2 21 35 (7) 6 (2)
rcl-n 3.00E+08 5.00E+08 7,000 1.10E05 4 0.61 2,029 472 164 (33) 19 (6)
3.00E+09 5.00E+09 600 1.00E06 43 0.59 158 138 89 (18) 18 (6)
3.00E+10 5.00E+10 80 1.20E07 312 0.64 24 51 54 (11) 20 (7)
rdcn 3.00E+08 3.00E+10 513 1.60E06 23 0.31 155 128 63 (11) 32 (14)
3.00E+09 3.00E+11 52 1.60E07 239 0.31 15 41 36 (6) 33 (14)
3.00E+10 3.00E+12 6 1.70E08 2,321 0.34 2 13 21 (4) 35 (16)
Yield
strength
(Pa)
dcr 1.00E+05 0.5 1.00E08 347 0.01 0 4 11 (2) 4 (2)
5.00E+05 5 1.00E07 77 0.02 4 13 20 (4) 4 (2)
1.00E+06 50 1.00E06 5 0.02 41 40 36 (6) 4 (2)
ars 1.00E+05 13 9.61E08 347 0.13 4 20 14 (3) 17 (10)
3.00E+05 81 6.27E07 75 0.13 20 51 22 (5) 16 (10)
5.00E+05 280 2.48E06 26 0.11 50 94 30 (6) 14 (9)
1.00E+06 470 2.82E07 3 0.12 453 122 34 (7) 21 (13)
Note. Parentheses in conduit radius and magma chamber volume estimations indicate standard deviation. Grayish highlight indicates the best estimation of our
calculations.

CÁCERES ET AL. 14
Geochemistry, Geophysics, Geosystems 10.1029/2018GC007817

Full data on measured thicknesses, length, and width of the flows and terrain slope are listed in the
supporting information.
Table 2 shows the estimations of effusion rates, duration of eruptions, and calculations of r and Vc obtained
for all the analyzed units. Figures 8–10 show the results for some selected lavas (for the rest of the lava units,
see supporting information). For rhyolitic lava flow units, we used initial viscosities ranging from 108 to
1010 Pa·s and final viscosities from 109 to 1011 Pa·s. Lower viscosities give shorter eruption times. For example,
rln results (Figure 8) indicate that for an initial viscosity of 108 Pa·s, the eruption duration is ~67 days, but the
calculated initial effusion rate is >700 m3/s, which is a very high value for a rhyolitic lava eruption and implies
a ~153-m (equation (11)) or 94-m (equation (4a)) conduit radius. On the other hand, an initial viscosity of
1010 Pa·s results in an eruption duration of 6.6 years, with initial effusion rate of 26 m3/s and a conduit radius
of 29 m (equation (11)) or 41 m (equation (4a)). If we use an initial viscosity of the order of 109 Pa·s, the erup-
tion duration is ~300 days, with an initial effusion rate of 160 m3/s. Conduit radius estimations are similar with
both equations (71 m with equation (11) and 64 m with equation (4a)). Magma chamber volume is estimated
at 30 km3.
Other lava flow units show similar results, with initial viscosities in the range of 109–1010 Pa·s and eruption
durations of ~100–3,000 days. These results give magma chamber volumes of 6–30 km3. Results for the rdcn
unit are shown in Figure 9. These results are similar compared to those of the rhyolite units, with eruption
duration estimations ranging from 23 to ~2,300 days and initial eruption rates of 500 to 6 m3/s. Magma
chamber volume is estimated at ~30 km3.
In the case of the andesite Río Saso (ars, Figure 10), the Newtonian rheological model does not give satisfac-
tory results. A much better fit is obtained with the yield strength in the crust model. The results with a crustal
yield strength of 105–106 Pa give initial effusion rates of 13–470 m3/s and eruption durations of 3–350 days.
Calculated magma chamber volume is of the order of ~16 km3. Radius conduit values are between 14 and
34 m with the method of equation (4a). Results with equation (11) could be less accurate because ascent
velocity could not be the only control on the explosive-effusive transition for andesitic magmas. For these
magmas, bubbles can ascend much faster than the melt phase and the eruption mechanism can be
more complex.

6. Discussion
6.1. Time Scale Estimates of Effusive Phases of LdMVF Eruptions
Our results indicate that the most likely duration of most effusive phases of LdMVF were in the range of
~100–3,000 days, with peak effusion rates in the range of 5–100 m3/s. We favor results with initial viscosities
of the order of 109 Pa·s and durations of <1 year, as in these cases, conduit radius estimates coincide with
both methods (equations 4a and 11) and eruption durations are comparable with historical cases, although
lower effusion rates and longer durations cannot be discarded. There are few witnessed rhyolitic eruptions in
recorded history, but two recent eruptions in Southern Chile (Chaitén 2008 and Cordón Caulle 2011) can give
us some insights. Chaitén volcano started to erupt rhyolitic magma in May 2008, and after a week of mainly
explosive activity, a lava dome began to extrude inside the preeruption caldera and above an old dome. The
extrusion lasted almost 18 months, and the total extruded volume reached 0.8 km3. The average effusion rate
during the first 2 weeks was 66 m3/s (Pallister et al., 2013). At Cordón Caulle, another rhyolitic eruption started
in June 2011. After 2 weeks of explosive activity, a lava flow started to extrude and reached ~4 km from the
vent after a year. The total lava volume was ~0.6 km3 with initial effusion rates of ~50m3/s (Coppola et al.,
2017). These values are similar to our estimates of the LdMVF eruptions and suggest that these eruptions
had the same eruptive mechanisms as the historical cases of Chaitén and Cordón Caulle. In these two historic
eruptions, the ratio between the volume of pyroclastic and lava extruded is roughly 1. In LdMVF, field evi-
dence also suggests an initial explosive phase followed by the extrusion of lava, but it is difficult to measure
the volume of the pyroclastic deposits of the eruptions analyzed here, although fall deposits associated with
these eruptions seem to be of the subplinian type, probably with erupted volumes of less than 1 km3.
Prehistoric examples of silicic lava flows include the Chao dacite in northern Chile (De Silva et al., 1994), where
an extrusion duration of 100–150 years for the ~26-km3 lava flow was estimated, with peak effusion rates of
25 m3/s. Loewen et al. (2017) estimated eruption durations of 2–5 years for the Summit Lake lava flow
(volume of ~50 km3) in Yellowstone, with peak effusion rates of >100 m3/s. Manley (1996) estimated an

CÁCERES ET AL. 15
Geochemistry, Geophysics, Geosystems 10.1029/2018GC007817

Figure 8. Results of the rln-w unit. (a) Fitting of the thickness of the flow using different effusion rates. (b) Calculated dis-
tance versus time for effusion rates used in Figure 8a. (c) Time evolution of effusion rates. (d) variations of viscosity asso-
ciated with the different effusion rates used in Figures 8a–8c. The “sawtooth” character of the height estimates of Figure 8a
(and Figures 9a and 10a) is due to the discretization of terrain slope, which can change abruptly between points.

eruption duration of 2–16 years for the Badlands rhyolite lava flow (16 km3), Idaho, USA, with mean eruption
rates of 30–230 m3/s. All these prehistoric examples correspond to larger lava flows compared with the
LdMVF lavas analyzed in this work. A common assumption of these authors is a constant effusion rate in
contrast with our model where we used an exponential decrease. Both Chaitén and Cordón Caulle
eruptions showed a decrease in the eruption rate with time. It would be interesting to analyze if long-
lived, large-volume rhyolitic lava flows effectively are controlled by a different eruptive mechanism with a
nearly constant or intermittent effusion rate evolution as some historical long-lived eruptions suggest
(Nakada et al., 1999).

Figure 9. Results of the rdcn unit. (a) Fitting of the thickness of the flow using different effusion rates. (b) Calculated dis-
tance versus time for effusion rates used in Figure 8a. (c) Time evolution of effusion rates. (d) Variations of viscosity asso-
ciated with the different effusion rates used in Figures 8a–8c.

CÁCERES ET AL. 16
Geochemistry, Geophysics, Geosystems 10.1029/2018GC007817

Figure 10. Results of the ars unit. (a) Fitting of the thickness of the flow using different effusion rates. (b) Time evolution of
effusion rates. (c) Calculated distance versus time for effusion rates used in Figure 10a.

6.2. Plumbing System Parameters: Conduit, Depth, and Size of Reservoirs


Conduit dimension is one of the most difficult parameters to estimate because there are few direct observa-
tions and indirect methods such as geophysical imaging that do not have the spatial resolution to constrain
its value. Our results indicate a conduit radius of ~50 m for rhyolitic eruptions and ~10 m for the andesite unit
assuming a cylindrical conduit. There are few estimations of this parameter in other volcanic systems. For
example, for the 1980–1986 St. Helens eruption (dacitic composition), a conduit radius of ~10–20 m was
estimated by different approaches (Melnik et al., 2011). At Unzen volcano, Japan, the Unzen drilling project
(Nakada et al., 2005) found dike widths from different eruptions up to 40 m. Vogel et al. (1987) reported dike
widths of 34 m at Inyo domes, USA. Considering that according to Wada (1994) dike widths are related to
magma composition, with a range of ~1 m for basaltic magmas up to ~100 m for rhyolitic compositions,
our results seem reasonable. We used the approach of Loewen et al. (2017) based on the ascent velocity
criteria for the transition between effusive and explosive phases during silicic eruptions. This approach
should be used with caution as some authors (Castro & Gardner, 2008) suggest that other factors can also
contribute to the occurrence of the explosive-effusive transition.
Analysis of rhyolite units gave us magma chamber volumes in the range of 20–100 km3. For the rhyodacites
the calculated volumes are in the range of 8–40 km3. For the dacite and andesite units the volume range is
5–30 km3. These values agree with the estimation made by Miller et al. (2017), using gravity measurements,
for the volume of the magma reservoir responsible for the current inflation at LdMVF. Our estimates should
be considered as maximums as we are not considering volatile exsolution as a triggering mechanism that will
reduce these ranges (Bower & Woods, 1997). If we consider a mean volume of ~20 km3, this corresponds to a

CÁCERES ET AL. 17
Geochemistry, Geophysics, Geosystems 10.1029/2018GC007817

sphere of radius ~2 km or a cylinder of radius 4 km with a thickness of 1 km. Our thermobarometric results on
the rhyodacites units indicate a polybaric crystallization of amphibole (Figure 6a) that could suggest convec-
tive ascent in a vertical-oriented reservoir. Based on these considerations, we propose that the eruptions that
generated the lava flows analyzed in this work were driven by different, individual magma chambers or,
alternatively, by different compartments of eruptible magma inside a large mush-type reservoir as suggested
by Singer et al. (2014). These authors propose a magma system of LdMVF similar to the model of Hildreth
(2004) for the Mammoth Lake (USA) magmatic system. Gudmundsson (2012) proposed that magma cham-
bers are internally divided into several compartments that can erupt independently, and these compart-
ments can be connected only when the overpressure that triggers the eruption is sufficiently high.
6.3. Implications for Future Eruptions
Our results suggest that many of the youngest and largest lava flows units of LdMVF were generated by
eruptions that were similar in terms of erupted volume and duration to other rhyolitic eruptions such as at
Chaitén (2008–2009) and Cordón Caulle (2011–2012) volcanoes. These results in conjunction with recent
estimations of magma chamber volume and injected magma (Miller et al., 2017) support the idea that in case
of a volcanic eruption, as a result of the current deformation, the most likely scenario is an eruption with
an initial explosive phase (probably subplinian type) followed by the extrusion of a lava flow or dome of
~0.1–1 km3 with a total eruption duration of months to a year. It has been suggested previously that the
current deformation at LdMVF could be a precursory signal of a “super-eruption” (Singer et al., 2014).
There are some large past eruptions that generated large calderas in this volcanic system (Hildreth et al.,
2010), and consequently, a large event cannot be discarded as a result of the current unrest at LdMVF. It is
worth noticing that the eruptive activity of LdMVF has incremented its frequency during the Holocene
(Andersen et al., 2017, 2018; see timeline inset of Figure 1) and thus is critical to make a comprehensive
volcanic hazard assessment of the zone and develop contingency plans in case of a future eruption.
LdMVF is located high in the Andes, close to the border with Argentina, and there are very few inhabitants in
the area close to this volcanic complex. The main hazards would be the tephra dispersion to the east and
lahars that can be generated during the winter and spring months and can affect the international road that
crosses the area by the northern shoreline of Laguna del Maule lake (red line in Figure 1).

7. Conclusion
Our results indicate that postglacial lava flows generated by LdMVF were extruded over time scales on the
order of months to a few years, with peak effusion rates <100 m3/s. Magma chamber volume estimations
give results on the order of 6–30 km3 and conduit radius of 50–70 m for rhyolitic units and ~20 m for the
andesitic unit. Our thermobarometric results indicate deeper crystallization of amphibole for the rhyodacitic,
dacitic, and andesite units when compared with results of rhyolite units from previous work. It is worth noti-
cing that the trend of P-T conditions of the rhyodacitic units suggest a polybaric and polythermal crystalliza-
tion of amphibole that could be due to convection inside a vertically oriented reservoir. We suggest that the
large, mush-type reservoir proposed by Singer et al. (2014) and Andersen et al. (2017) that fed eruptions at
LdMVF is divided into individual, smaller reservoirs with the potential to erupt independently in the future.

Acknowledgments References
We thank CONICYT Master Scholarship
221320285, Department of Andersen, N. L., Singer, B., Costa, F., Fournelle, J., Herrin, J., & Fabbro, G. N. (2018). Petrochronologic perspective on rhyolite volcano unrest at
Postgraduate and Graduate grant of the Laguna del Maule, Chile. Earth and Planetary Science Letters, 493, 57–70. https://doi.org/10.1016/j.epsl.2018.03.043
University of Chile, FONDECYT project Andersen, N. L., Singer, B., Jicha, B., Beard, B., Johnson, C., & Licciardi, J. (2017). Pleistocene to Holocene growth of a large upper crustal
11121298, and FONDAP project rhyolitic magma reservoir beneath the active Laguna del Maule Volcanic Field, Central Chile. Journal of Petrology, 58(1), 85–114. https://
15090013 (CEGA), for financial support doi.org/10.1093/petrology/egx006
to develop this study. We thank G. Bacon, C. R., & Hirschmann, M. M. (1988). Mg/Mn partitioning as a test for equilibrium between coexisting Fe-Ti oxides. American Mineralogist,
Wadge and an anonymous reviewer for 73, 57–61.
their comments that greatly improved Blake, S. (1981). Volcanism and the dynamics of open magma chambers. Nature, 289(5800), 783–785. https://doi.org/10.1038/289783a0
the quality of the first draft. All data Bower, S., & Woods, A. W. (1997). Control of magma volatile content and chamber depth on the mass erupted during explosive volcanic
used in the text are shown in tables, eruptions. Journal of Geophysical Research, 102(B5), 10,273–10,290. https://doi.org/10.1029/96JB03176
figures, and supporting information. Cardona, C., Tassara, A., Gil-Cruz, F., Lara, L., Morales, S., Kohler, P., & Franco, L. (2018). Crustal seismicity associated to rapid surface uplift at
Laguna del Maule Volcanic Complex, Southern Volcanic Zone of the Andes. Journal of Volcanology and Geothermal Research, 353, 83–94.
https://doi.org/10.1016/j.jvolgeores.2018.01.009
Castro, J. M., & Gardner, J. E. (2008). Did magma ascent rate control the explosive–effusive transition at the Inyo volcanic chain, California.
Geology, 36(4), 279–282. https://doi.org/10.1130/G24453A.1

CÁCERES ET AL. 18
Geochemistry, Geophysics, Geosystems 10.1029/2018GC007817

Castruccio, A., & Contreras, M. (2016). The influence of effusion rate and rheology on lava flow dynamics and morphology: A case study from
the 1971 and 1988–1990 eruptions at Villarrica and Lonquimay volcanoes, Southern Andes of Chile. Journal of Volcanology and
Geothermal Research, 327, 469–483. https://doi.org/10.1016/j.jvolgeores.2016.09.015
Castruccio, A., Rust, A. C., & Sparks, R. S. J. (2013). Evolution of crust- and core-dominated lava flows using scaling analysis. Bulletin of
Volcanology, 75(1), 681. https://doi.org/10.1007/s00445-012-0681-2
Coppola, D., Laiolo, M., Franchi, A., Massimetti, F., Cigolini, C., & Lara, L. (2017). Measuring effusion rates of obsidian lava flows by
means of satellite thermal data. Journal of Volcanology and Geothermal Research, 347, 82–90. https://doi.org/10.1016/j.
jvolgeores.2017.09.003
Cordell, D., Unsworth, M. J., & Diaz, D. (2018). Imaging the Laguna del Maule Volcanic Field, central Chile using magnetotellurics: Evidence for
crustal melt regions laterally-offset from surface vents and lava flows. Earth and Planetary Science Letters, 488, 168–180. https://doi.org/
10.1016/j.epsl.2018.01.007
De Silva, S. L., Self, S., Francis, P. W., Drake, R. E., & Ramirez, C. (1994). Effusive silicic volcanism in the Central Andes: The Chao dacite and other
young lavas of the Altiplano-Puna Volcanic Complex. Journal of Geophysical Research, 99(B9), 17,805–17,825. https://doi.org/10.1029/
94JB00652
Eichelberger, J. C., Carrigan, C. R., Westrich, H. R., & Price, R. H. (1986). Non-explosive silicic volcanism. Nature, 323(6089), 598–602. https://doi.
org/10.1038/323598a0
Feigl, K. L., Le Mével, H., Ali, S. T., Córdova, L., Andersen, N. L., DeMets, C., & Singer, B. S. (2014). Rapid uplift in Laguna del Maule volcanic field of
the Andean Southern Volcanic Zone (Chile) 2007-2012. Geophysical Journal International, 196(2), 885–901. https://doi.org/10.1093/gji/ggt438
Fierstein, J., Sruoga, P., Amigo, A., Elissondo, M., & Rosas, M. (2012). Postglacial eruptive history of Laguna del Maule Volcanic Field in Chile,
from fallout stratigraphy in Argentina. AGU Fall Meeting Abstracts, V31F-03.
Frey, F. A., Gerlach, D. C., Hickey, R. L., López-Escobar, L., & Munizaga, F. (1984). Petrogenesis of the Laguna del Maule Volcanic Complex, Chile
(36°S). Contributions of Mineralogy and Petrology, 88(1-2), 133–149. https://doi.org/10.1007/BF00371418
Ghiorso, M. S., & Evans, B. W. (2008). Thermodynamics of rhombohedral oxide solid solutions and a revision of the Fe-Ti two-oxide
geothermometer and oxygen-barometer. American Journal of Science, 308(9), 957–1039. https://doi.org/10.2475/09.2008.01
Giordano, D., Russell, J. K., & Dingwell, D. B. (2008). Viscosity of magmatic liquids: A model. Earth and Planetary Science Letters, 271(1-4),
123–134. https://doi.org/10.1016/j.epsl.2008.03.038
Gudmundsson, A. (2012). Magma chambers: Formation, local stresses, excess pressures and compartments. Journal of Volcanology and
Geothermal Research, 237-238, 19–41. https://doi.org/10.1016/j.jvolgeores.2012.05.015
Hildreth, W. (2004). Volcanological perspectives on Long Valley, Mammoth Mountain and 5: Several contiguous but discrete systems. Journal
of Volcanology and Geothermal Research, 136(3-4), 169–198. https://doi.org/10.1016/j.jvolgeores.2004.05.019
Hildreth, W., Godoy, E., Fierstein, J., & Singer, B. (2010). Laguna del Maule Volcanic Field: Eruptive history of a Quaternary basalt-to-rhyolite
distributed volcanic field on the Andean rangecrest in central Chile (Vol. 63, p. 145). Santiago, Chile: Servicio Nacional de Geología y Minería,
Boletín.
Huppert, H., & Woods, A. W. (2002). The role of volatiles in magma chamber dynamics. Nature, 420, 493–495.
Holtz, F., Johannes, W., Tamic, N., & Behrens, H. (2001). Maximum and minimum water contents of granitic melts generated in the crust: A
reevaluation and implications. Lithos, 56(1), 1–14. https://doi.org/10.1016/S0024-4937(00)00056-6
Jaupart, C., & Allegre, C. (1991). Gas content, eruption rate and instabilities of eruption regime in silicic volcanoes. Earth and Planetary Science
Letters, 102(3-4), 413–429. https://doi.org/10.1016/0012-821X(91)90032-D
Kiss, B., Harangi, S., Ntaflos, T., Mason, P. R. D., & Pál-Molnár, E. (2014). Amphibole perspective to unravel pre-eruptive processes and con-
ditions in volcanic plumbing systems beneath intermediate arc volcanoes: A case study from Ciomadul volcano (SE Carpathians).
Contributions to Mineralogy and Petrology, 167(3), 986. https://doi.org/10.1007/s00410-014-0986-6
Lange, R., Frey, H., & Hector, J. (2009). A thermodynamic model for the plagioclase-liquid hygrometer/thermometer. American Mineralogist,
94(4), 494–506. https://doi.org/10.2138/am.2009.3011
Le Mével, H., Gregg, P., & Feigl, K. (2016). Magma injection into a long-lived reservoir to explain geodetically measured uplift: Application to
the 2007–2014 unrest episode at Laguna del Maule Volcanic Field, Chile. Journal of Geophysical Research: Solid Earth, 121, 6092–6108.
https://doi.org/10.1002/2016JB013066
Loewen, M., Bindeman, I., & Melnik, O. (2017). Eruption mechanisms and short duration of large rhyolitic lava flows of Yellowstone. Earth and
Planetary Science Letters, 458, 80–91. https://doi.org/10.1016/j.epsl.2016.10.034
Manley, C. R. (1996). Physical volcanology of a voluminous rhyolite lava flow: The Badlands lava, Owyhee Plateau, southwestern Idaho.
Journal of Volcanology and Geothermal Research, 71(2-4), 129–153. https://doi.org/10.1016/0377-0273(95)00066-6
Melnik, O., Blundy, J., Rust, A., & Muir, D. (2011). Subvolcanic plumbing systems imaged through crystal size distribution. Geology, 39(4),
403–406. https://doi.org/10.1130/G31691.1
Melnik, O., & Sparks, R. S. J. (1999). Nonlinear dynamics of lava dome extrusion. Nature, 402(6757), 37–41. https://doi.org/10.1038/46950
Miller, C. A., Williams-Jones, G., Fournier, D., & Witter, J. (2017). 3D gravity inversion and thermodynamic modeling reveal properties of
shallow silicic magma reservoir beneath Laguna del Maule, Chile. Earth and Planetary Science Letters, 459, 14–27. https://doi.org/10.1016/j.
epsl.2016.11.007
Mueller, S., Llewellin, E. W., & Mader, H. M. (2009). The rheology of suspensions of solid particles. Proceedings of the Royal Society, 466,
1201–1228.
Nakada, S., Shimizu, H., & Ohta, K. (1999). Overview of the 1990-1995 eruption at Unzen volcano. Journal of Volcanology and Geothermal
Research, 89(1-4), 1–22. https://doi.org/10.1016/S0377-0273(98)00118-8
Nakada, S., Uto, K., Sakuma, S., Eichelberger, J. C., & Shimizu, H. (2005). Scientific results of conduit drilling in the Unzen Scientific Drilling
Project (USDP). Scientific Drilling, 1, 18–22. https://doi.org/10.2204/iodp.sd.1.03.2005
Pallister, J. S., Diefenbach, A. K., Burton, W. C., Muñoz, J., Griswold, J. P., Lara, L., et al. (2013). The Chaitén rhyolite dome: Eruption sequence,
lava dome volumes, rapid effusion rates and source of the rhyolite magma. Andean Geology, 40(2), 277–294.
Ridolfi, F., & Renzulli, A. (2012). Calcic amphiboles in calc-alkaline and alkaline magmas: Thermobarometric and chemometric empirical
equations valid up to 1,130°C and 2.2 GPa. Contributions of Mineralogy and Petrology, 163(5), 877–895. https://doi.org/10.1007/s00410-
011-0704-6
Ridolfi, F., Renzulli, A., & Puerini, M. (2010). Stability and chemical equilibrium of amphibole in calc-alkaline magmas: An overview, new
thermobarometric formulations and application to subduction-related volcanoes. Contributions of Mineralogy and Petrology, 160(1),
45–66. https://doi.org/10.1007/s00410-009-0465-7
Shejwalkar, A., & Coogan, L. (2013). Experimental calibration of the roles of temperature and composition in the Ca-in-olivine
geothermometer at 0.1 MPa. Lithos, 177, 54–60. https://doi.org/10.1016/j.lithos.2013.06.013

CÁCERES ET AL. 19
Geochemistry, Geophysics, Geosystems 10.1029/2018GC007817

Singer, B. S., Andersen, N. L., Le Mével, H., Feigl, K., DeMets, C., Tikoff, B., et al. (2014). Dynamics of a large, restless, rhyolitic magma system at
Laguna del Maule, southern Andes, Chile. GSA Today, 24(12), 4–10. https://doi.org/10.1130/GSATG216A.1
Singer, B. S., Hildreth, W., & Vincze, Y. (2000). 40Ar/39Ar evidence for early deglaciation of the central Chilean Andes. Geophysical Research
Letters, 27(11), 1663–1666. https://doi.org/10.1029/1999GL011065
Vergara, M., & Muñoz, J. (1982). La Formación Cola de Zorro en la Alta Cordillera Andina Chilena (36°–39° Lat. S), sus características
petrográficas y petrológicas: Una revisión. Revista Geológica de Chile (now Andean Geology), 17, 31–46.
Vogel, T. A., Younker, L. W., & Schuraytz, B. C. (1987). Constrains on magma ascent, emplacement, and eruption: Geochemical and minera-
logical data from drill-core samples at Obsidian dome, Inyo chain, California. Geology, 15(5), 405–408. https://doi.org/10.1130/0091-
7613(1987)15<405:COMAEA>2.0.CO;2
Wada, Y. (1994). On the relationship between dike width and magma viscosity. Journal of Geophysical Research, 99(B9), 17,743–17,755.
https://doi.org/10.1029/94JB00929
Walker, G. P. L. (1971). Compound and simple lava flows and flood basalts. Bulletin of Volcanology, 35(3), 579–590. https://doi.org/10.1007/
BF02596829

CÁCERES ET AL. 20

You might also like