You are on page 1of 15

Analysis of the seepages through the Laja Lake barrier in central Chile

J.L. Arumi; G. Dietrich; I. Neuweiler, Enrique Muñoz, Daniel PAez

INTRODUCTION

Laja Lake is located in the Chilean Andes at 1360 m. a. s. l., next to Antuco Volcano in the
west and in the Biobío basin (Figure 1). The lake has an area of 87 km² and is located
between 36° 50’ and 37° 40’ S and 71° 7’ and 72° 45’ W (Mardones & Vargas 2005). The
Laja River is the main tributary of the Biobío River and springs from seepages through the
barrier made up of volcanic material that dams the lake and is the source of water that
supplies almost two million inhabitants of the Biobío Region, seven hydropower plants,
with 1150 MW of installed capacity, and almost 120,000 hectares of irrigation (Figure 1).

During the last decade, water resources management in the Laja system has become a
critical matter due to a long period of water scarcity that has affected Chile, as well as
increased citizen sensitivity to the protection of landscapes and ecosystem services, for
which Laja Falls has become a central icon. All of this has given rise to a series of research
and consulting projects on the Laja basin and the management of the system (Arumi et al.,
2012; Muñoz et al., 2014).

It bears noting that all of the operational models of the Laja system are based on solving the
water balance in Laja Lake using estimates of the runoff inputs from the contributing basin
to the lake, lake evaporation estimates (which in some cases are erroneously disregarded)
and the lake seepage law (Muñoz et al., 2014).

In this case we will focus on the Laja Lake seepage law, which consists of a series of
equations that relate the levels of Laja Lake to the streamflow of the seepages that give rise
to the river of the same name. The seepages of Laja Lake run through the barrier that was
formed by the collapse of the wall of Antuco Volcano and emerge in the Chilcas Falls
sector, located approximately 4 kilometers from the lake. Thus, according to the theory of
flow through porous media, using a direct relationship between the streamflow measured
on a given day and the lake level measured that day without taking into account the time it
takes the water to seep through the barrier proves contradictory.

Paradoxically, this relationship has been used since 1958 without its validity ever having
been discussed; therefore, this work is focused on understanding the dynamics of the flow
that connects Laja Lake to the Laja River via seepages through the barrier.

It is important to state that this work has inherent limitations regarding data availability,
budget and time. Therefore, it aims to improve our understanding of the hydrogeological
processes that control the seepages through the Laja Lake barrier through an analysis of
geological aspects of the lake and the application of various numerical and conceptual
models.
Figure 1. Study area showing the location of the Biobío basin; b) the Laja Lake, and a
Google Earth view of the barrier formed by a landslide from the Antuco Volcano.
ANTECEDENTS

The formation of lakes in mountainous areas through landslides is very common worldwide
(Bonnard, 2011, Korup and Tweed, 2007). According to Bonnard (2011), 12 of the 35
natural lakes that are located in Switzerland are considered to have resulted from landslide
dams. In New Zealand Korup (2004) made an inventory of 232 landslide dam occurrences,
of which 140 are still in existence. In Chile, there are several lakes resulting from
landslides, the most well-known of which are Laja, Caburga and Todos los Santos.

Costa and Schuster (1988) studied 183 landslide dams around the world, finding that most
of the landslides were caused by avalanches, slumps, slides and flows, with only 11 caused
by sensitive clay failures or falls. The triggering causes of the landslide dams studied by
Costa and Schuster were: rainstorms and snowmelt (52%), earthquakes (40%) and volcanic
eruptions (8%). From a geological point of view landslide dams can be classified
according to the relationships between the dam and the valley floor that is covered
following the classification defined by Swanson et al. (1986), which covers the six types of
dams described in Table 1 and Figure 2.

Because a natural dam is formed by a heterogeneous mix of earth material, it is very likely
that a failure will occur in the short term. In fact, Costa and Schuster (1988) studied the
failures of 73 landslide dams, finding that 85% of them broke after less than one year of
existence and that the principal failure mode is overtopping (71%), with piping (4%); slope
failure (3%) and human-controlled failure (22%) accounting for the remaining portion. The
longevity of landslide dams depends on factors such as the size and shape of the barrier, the
composition of the blocking material (size, shape and sorting of the material) and the rates
of seepage through the barrier (Costa and Schuster, 1988). Therefore, an understanding of
the seepage dynamics through the Laja Lake barrier is important not only for water
resources management, but also for the risk assessment of the two million people that live
downriver of the lake.
Table 1. Classification of landslide dams according to the relationship between the barrier
and the valley (modified from Costa and Schuster, 1988)

Type Description Percentage


I Small, does not span the entire 11
valley
II Large, spans the entire valley 44
III Fills the valley up- and 41
downstream
IV Contemporaneous failures at <1
both sides of the valley
V The same landslide has multiple <1
debris lobes
VI Involves multiple failure surfaces 3
and extends under the stream

Figure 2. Classification of landslide dams according to the relationship between the barrier
and the valley (modified from Costa and Schuster, 1988)
Laja Lake barrier

The formation of this lake is intimately related to the evolution of Antuco Volcano, located
at its far south, the activity of which began approximately 130,000 years ago.
Approximately 9,500 years BP, a large eruption of the original Antuco volcano, known as
Antuco I, which involved the collapse of the volcanic structure and the embankment of the
upper part of the valley, resulted in the elevation of Laja Lake being 100 m above its
current level. While there has not been a precise dating, in the works of Thiele et al. (1998)
and Moreno et al. (2000) a there is a detailed interpretation of Laja Lake barrier collapse
event based on the resulting avalanche deposits and debris flow that are distributed all
along the intermediate and lower sections of the Laja and Bíobío river valley. Nor is there
clarity regarding the degree to which the lake level fell following its discharge or how
many subsequent volcanic events contributed to the current embankment conditions. The
last large-magnitude volcanic event that altered the lake level occurred in 1853, when a
fissure eruption emitted a lava flow that raised the natural dam of the lake by about 20 m.
Therefore, Laja Lake was formed by a volcanic eruption and its barriers correspond to a
Type III barrier according to the classification made by Swanson et al. (1986).

Seepage law
In the agreement signed between the former Irrigation Directorate, the current Hydraulic
Works Directorate and Endesa in 1958, a series of measures was established for the
exploitation of Laja Lake, covering irrigation and electricity generation uses (DOH, 2007.
Under the framework of this agreement what was called the “Laja Lake Seepage Law” was
established, allowing the contributing flows to the Albanico hydropower plant that
originate in the seepages that give rise to Laja Lake in the Chilcas and Torbellino falls
sector to be estimated. The relationship between the Laja Lake water level and the seepage
streamflows was established as three equations that cover three lake water levels (Figure 3):

Level between 1,362 and 1,369 m. a. s. l.


𝑄𝑆 = 8.44 ∗ 10−3 ∗ ( 𝑙𝑒𝑣𝑒𝑙 – 1,290 )2,000 [1]

Level between 1,340 and 1,362 m. a. s. l.


𝑄𝑆 = 9.74 ∗ 10−2 ∗ ( 𝑙𝑒𝑣𝑒𝑙 – 1,275 )1,370 [2]

Level between 1,220 and 1,340 m. a. s. l.


𝑄𝑆 = 9.26 ∗ 10−3 ∗ ( 𝑙𝑒𝑣𝑒𝑙 – 1,220 )1,684 [3]

With:
QS ∶ Seepage streamflow [m3 /s]
Level: Elevation of the water level of Laja Lake [m. a. s. l. ].

When taking into account that the speed of groundwater in highly permeable media has an
order of magnitude of meters per day and that in the case of the Laja barrier the water runs
a distance of approximately 4000 meters, the theoretical travel time should be on the order
of a year, but the seepage law proposes an almost instantaneous connection.
60
Seepage streamflow (m3/s) Streamflow for level between 1,362 and 1,369 m. a. s. l
50

40 Streamflow for level between 1,340 and 1,362 m. a. s. l

30

20 Streamflow for level between 1,220 and 1,340 m. a. s. l

10

0
1220 1245 1270 1295 1320 1345 1370
Laja Lake water level elevation (m.a.s.l.)

Figure 3. Streamflow curve of the “seepage law” as a function of the Laja Lake water
level by section.
METHODS

This analysis was performed using secondary information available on the internet. The
topography of the study area was obtained from a Digital Elevation Model (DEM) based on
the Advanced Space-borne Thermal Emission and Reflection Radiometer (ASTER) of 1
arc-second satellite stereo images. The ASTER data analysis was done using ArcGis 10.2
for surface modeling and Global Mapper 16 to generate the Laja barrier profile for
groundwater modeling (Figure 4)

A first analysis of the seepage law used the statistics available from the General Water
Directorate (www.dga.cl) on Laja Lake levels and the streamflows measured at the Abanico
Canal at km 49 station, approximately 9 km down-water of the springs. As the streamflow
statistics include both seepage inputs from Laja Lake and that of the contributing basin
indicated in Figure 1, it was necessary to estimate these inputs using a hydrological model.

Leinemann (2014) and later Lerch (2015) applied the SWAT model to generate synthetic
streamflow series for the intermediate basin, which were subtracted from the streamflow
statistics measured at the Abanico Canal at km 49 station in order to estimate the seepage
streamflows (QS). Subsequently, the lake elevation – QS relationship was analyzed
considering time gaps of 0, 1, 2, 3, 7, 14, 21, 100 and 300 days.

The data needed to carry out the simulation using SWAT are: i) soil series mapping
obtained from the Natural Resource Information Center (CIREN), which was verified in the
field, b) precipitation data generated from information available from the DGA and c)
climate information obtained from the Research Data Archive of the National Center for
Atmospheric Research (NCAR) in Boulder, Colorado, USA (http://rda.ucar.edu). The use
of climate information in the Laja watershed was validated by Muñoz et al. (2014), who
found that internet-based data sets reflect climate variability, but that a scale correction is
needed to match the water balance in a mountain watershed.

The flow of water through the Laja Lake barrier was studied using two simulation models:
the first model employed was Groundwater Vistas and the second was HydroGeoSphere.
Groundwater Vistas is a Modflow-based model that allows the simulation of water flow
and solute transport in different porous media (http://www.groundwatermodels.com/). The
core of the model is the MODFLOW algorithm that was developed in 1988 by McDonald
and Harbaugh, solving the water balance and Darcy equation in a block-centered modular
domain using a using a finite difference numerical scheme. The flow through the Laja Lake
barrier was simulated using mudflow assuming that the barrier is composed of a
homogeneous granular material (for example, blocks of rocks, gravel and sandy soils) that
was deposited after the avalanche.

HydroGeoSphere (HGS) is a three-dimensional finite element model that simultaneously


solves the 2D diffusive wave equation and the 3D form of Richards’ equation
(https://www.aquanty.com/hydrogeosphere). The model has the ability to simulate surface
and subsurface flow interaction and was derived from the algorithm created by Therrien
(1992) called FRAC3DVS.
Figure 4. Conceptual scheme of the Laja Lake barrier, obtained from interpretation of
digital terrain elevation models and available geological information.

RESULTS AND DISCUSSION

Hydrological analysis of the seepage law

Using SWAT an estimate of the streamflows generated by the catchement located


downwater of the Barrier, which were subtracted from the streamflows recorded by the
DGA at the Abanico Canal at km 49 gauging station, was obtained. Analysis of the results
of Leinemann established that the best fit occurred for three-day gaps, which is consistent
with the seepage law (Figure 5).

Figure 5: Relationship between the lake level and the seepage law, with a three-day gap
(Leinemann, 2014)
Modflow model application

An initial approximation of groundwater flow was performed using Modflow. To this end,
a simplified barrier was simulated by assigning it a constant width and a shape given by the
digital terrain elevation model (Figure 4).

The flows conditions were assessed considering three Laja Lake levels (elevations of 1305,
1335 and 1350 m.a.s.l.). Each height was tested with different hydraulic conductivities in
order to obtain flows similar to those given by the seepage law. Iterating the hydraulic
conductivity for each lake level with the Modflow model and comparing the water flow in
these conditions with the values of the seepage law produced the results in Table 1, where
the conditions for each modeled lake level are indicated (Conejeros, 2015).

With these results it is noticed that very high hydraulic conductivities are required in order
to reproduce the flow of water through the barrier, which would be equivalent to infills like
those shown in Figure 2. These conductivities are consistent with the seepage law, but
according to flow in a saturated porous medium, the location of the springs should vary at
different lake levels as is shown in figure 6. That situation does not occur, since the location
of Chilcas Falls has been constant over time (Figure 7). This indicates that the runoff within
the Laja barrier does not occur through a saturated porous medium, but rather probably a
fractured or double-porous medium, similar to the illustrated in figure 8.

Table 2. Results of the groundwater flow model in Modflow, compared with the seepage
law (Conejeros, 2015)
K Water flow Seepage law
Height [m.a.s.l.] Water flow [m3/s]
[m/day] [m3/day] [m3/s]

1,305 325 1,447,765 16.76 16.43

1,335 450 2,432,150 28.15 27.34

1,350 500 2,956,605 34.22 36.09


1400 a)

Elevation (meters) 1300


b)

c)
1200

1100

1000
0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000
Distance (meters)

Figure 6. Scheme showing the results of modeling groundwater flow through the barrier
using MODFLOW at different Laja Lake levels

Figure 7. Chilcas Falls: a) Condition prior to 1981 (Endesa, 1981) and in the present
(2014)
Figure 8. Macro-porous medium formed by large particles of volcanic origin

Application using HydroGeoSphere

Spitzenberg (2014) studied a second approximation using the HydroGeoSphere model. This
model allows variable-saturation flow conditions to be simulated in a domain that takes into
account the existence of fractures to the extent that they are not deformable. Spitzenberg
(2014) simulated the flow under the barrier considering the existence of an unfractured
porous medium and one with fractures of 5 and 10 millimeters. Figure 9 presents the results
of the simulations involving fractures, which allowed flows equivalent to those determined
by the seepage law, and more interestingly, the existence of a fixed point where the springs
occur, to be obtained.
Figure 9. Results of the modeling using a fractured medium (Spitzenberg, 2014)

Discussion

The results obtained from these academic exercises allow the following hypothesis
regarding the flow through the Laja Lake barrier to be proposed:

This barrier was formed by a large landslide of material that made up the northern wall of
Antuco Volcano, approximately 9,500 years ago. As the water flowed through the interior
of the barrier, the fine material was washed away, leaving a macro-porous medium such as
that shown in Figure 5. This medium generates preferential flow conditions that conduct
most of the water that emerges from the seepages.

The Laja Lake seepages connect to the barrier, forming an immediate drop, similar to those
shown in the left of Figures 7a and 7b, which should be associated with preferential flow within
the barrier, generating free surface flow conditions in thousands of small channels that
make up the streamflow that connects Laja Lake to the river of the same name.

A study done by the International Atomic Energy Agency (IAEA), the Hydraulic Works
Directorate and Endesa (Plata et al., 1989) sought a way to minimize the seepages through
the barrier. To this end, a series of field campaigns was carried out that included
measurements of water velocity in the lake, a series of borings throughout the barrier and a
series of analyses of environmental tracers such as environmental isotopes, temperature and
electrical conductivity.

The IAEA results demonstrate that the water emerges from the seepages from Laja Lake
and not from another hydrogeological system (for example, that of Antuco Volanco); in
addition, the borings performed indicated the presence of permeable strata that generate
flow conditions. Plata et al. (1989) reported the results of a tracer test that demonstrated the
existence of two flow types within the barrier: a fast flow that has a transit time of
approximately 2 to 3 days and a slower flow that has a transit time on the order of 20 days.
In accord with Plata et al. (1989), the fast flow contributes approximately 80% of the
streamflow that emerges from the seepages.

It bears mentioning that the authors of this article had access to the IAEA study only
recently, and courtesy of that organization. Therefore, the results of the numerical tests
using the SWAT, MODFLOW and HydroGeoSphere models were not influenced by the
results presented by Plata et al. (1989), which in turn confirm our conclusions. This
highlights the possibility of achieving robust results based on a conceptual work, even in
situations with very limited field information.

Conclusions

Laja Lake was formed by the collapse of the northwest wall of Antuco Volcano, which
created a volcanic barrier that obstructed the original course of the river around 10,000
years ago.

The current Laja River rises from a series of seepages that emerge at the foot of the
volcanic barrier at two points known as the Chilcas and Torbellino sectors. These seepages
result from a water flow that comes from the lake and crosses the volcanic barrier through
preferential flow systems that in practice generate free flow conditions within the barrier.
This explains the reduced residence time of the water within the barrier and the almost
instantaneous response of the seepage law.

Despite the originally expressed doubts, for purposes of managing the Laja system water
balance, the seepage law can be considered valid.

Acknowledgments

The authors wish to thank Conicyt for its support through the Fondecyt-BMBF-PCCI1-
2031 and Conicyt/Fondap/15130015 projects, as well as the IAEA for its support through
the RLA 7021 project.

References
• Arumí, J. L., D. Rivera, A. Rougier, R. Díaz. 2012. Estimación de pérdidas de agua
en tramos de ríos del sistema Laja-Diguillín en la zona central de Chile. Tecnología
y Ciencias del Agua, antes Ingeniería Hidráulica en México. Volumen III, núm. 3,
julio-septiembre de 2012, pp 135-141
• Bonnard, 2011. Technical and Human Aspects of Historic Rockslide-Dammed
Lakes and Landslide Dam Breaches. In S. G. Evans, R. L. Hermanns, A. Strom and
G. Scarascia-Mugnozza (Eds). Natural and Artificial Rockslide Dams. Springer
Berlin Heidelberg", Berlin, Germany. pp 101-122. ISBN: 978-3-642-04764-0; doi
10.1007/978-3-642-04764-0_3
• Conejeros A., 2015. Estudio conceptual de las filtraciones del Lago Laja. Memoria
para optar al título de Ingeniero Ambiental. Universidad de Concepción
• Costa, J., R. Schuster. 1988. The formation and failure of natural dams. Geological
Society of America Bulletin, v. 100, p. 1054-1068,13 figs., 7 tables, July 1988. doi:
10.1130/0016-7606(1988)
• DOH. Guía Para la Explotación del Lago Laja. Gobierno de Chile. Ministerio de
Obras Públicas. Dirección de Obras Hidráulicas – Endesa S.A. Santiago, Chile.
2007. 17 pp.
• Endesa. 1981. Central Hidroelectrica Antuco. Empresa Nacional de Electricidad,
Santiago, Chile. 54 pp.
• Korup O. 2004. Geomorphometric characteristics of New Zealand landslide dams.
Engineering Geology. 73 (2004) 13-35. DOI: 10.1016/j.enggeo.2003.11.003
• Korup, O., F. Tweed. 2007. Ice, moraine, and landslide dams in mountainous
terrain. Quaternary Science Reviews, Volume 26, Issue 25-28, p. 3406-3422. DOI:
10.1016/j.quascirev.2007.10.012
• Leinemann J., 2014. Investigations about water losses from a natural volcanic
reservoir in the Andes mountains (Laja Lake, Chile). Universidad de Leibniz
Hannover. Septiembre 2014
• Kerch H. 2015. Hydrological simulation of a volcanic catchment in the Chilean
Andean Mountain as a tool for water use conflict resolution. Master thesis at
Institute of Water Resources Management, Hydrology and Agricultural Hydraulic
Engineering, Leibniz University Hannover Germany. 98 pp.
• Mardones M., and Vargas, J. Efectos hidrológicos de los usos eléctricos y agrícola
en la cuenca del río Laja (Chile Centro-Sur). Revista de Geografía Norte Grande.
N° 33 pp 89-102
• McDonald M. G. and A. W. Harbaugh. 1988. A modular three-dimensional finite
difference ground-water flow model. MODFLOW. U. S. Geological Survey. Open
file Report 83-875.
• Moreno, H.; Lohmar, S.; López Escobar, L; Petit-Breuilh, M. E. 2000. Contribución
a la evolución geológica, geoquímica e impacto ambiental del volcán Antuco
(Andes del Sur, 37°25´S). IX Congreso Geológico Chileno, Actas Vol. 2, pp. 44-48,
Puerto Varas.
• Muñoz, E., J. L. Arumí, D. Rivera, A. Montecinos, M. Billib and C. Álvarez. 2014
Gridded data for a hydrological model in a scarce-data basin. Proceedings of the
Institution of Civil Engineers, Water Management 167 May 2014 Issue WM5.
Pages 249–258 http://dx.doi.org/10.1680/wama.12.00086
• Plata A., Traguen E., Silva A., Gonzalez F., Verni F. Mery H. 1989. Estudio de las
Filtraciones del Lago Laja. Informe proyecto CHI/8/016. Organismo Internacional
de Energía Atómica (OIEA). Viena, Austria 231 pp.
• Spitzenberg, Frank. 2014. Modelación del flujo bajo la barrera del Laja usando
HydroGeoSphere. Universidad de Leibniz Hannover, Agosto 2014
• Swanson, F. J. Oyagu, Norio, and Tominaga, Masaki, 1986, Landslide dams in
Japan, in Schuster, R. L., ed, Landslides dams – Processes, risks and mitigations:
American Society of Civil Engineers Geotechnical Special Publications 3, pp 131-
145
• Therrien, René (1992). Three-dimensional analysis of variably saturated flow and
solute transport in discretely-fractured porous media (Ph.D.). University of
Waterloo, Waterloo, Ontario
• Thiele, R., Moreno, H., Elgueta, S., Lahsen, A., Rebolledo, S. y Petit-Breuilh M. E.
Evolución geológico-geomorfológica cuaternaria del tramo superior del valle del río
Laja. Revista geológica de Chile. Vol.25, núm. 2, diciembre 1998, pp. 229-253.

You might also like