Written statement of the experts of Chile

INTERNATIONAL COURT OF JUSTICE
DISPUTE OVER THE STATUS AND USE OF THE
WATERS OF THE SILALA
(CHILE v. BOLIVIA)
WRITTEN STATEMENT OF THE
EXPERTS OF THE REPUBLIC OF CHILE
DRS. HOWARD WHEATER AND DENIS PEACH
14 JANUARY 2022

ABOUT THE AUTHORS
Dr. Howard Wheater
Dr. Howard Wheater is Canada Excellence Research Chair Laureate in Water
Security at the University of Saskatchewan, Canada, where he was founding
Director of the Global Institute for Water Security, and Emeritus Professor of
Hydrology at Imperial College London, where he held a full-time academic
appointment for 32 years. A leading expert in hydrological science and modelling,
he has published more than 240 refereed articles and 6 books. He is a Fellow of the
Royal Society of Canada, the UK's Royal Academy of Engineering and the
American Geophysical Union. He was awarded the 2018 International Prize for
Water (Dooge Medal) by UNESCO, the World Meteorological Organisation and
the International Association for Hydrological Sciences and the 2006 Prince Sultan
bin Abdulaziz International Prize for Water. He has initiated and led national and
international research programmes in the UK and Canada, and has advised states,
provinces and national governments on flood, water resource and water quality
issues. He sat on the Court of Arbitration concerning the Indus Waters Treaty and
has been instructed by Hungary and Argentina in cases before the International
Court of Justice. He was, until 2014, vice-chair of the World Climate Research
Programme's Global Energy and Water Cycle Exchange (GEWEX) project and led
UNESCO' s GW ADI arid zone water program. In Canada, he led the Changing Cold
Regions Network, focused on the analysis and prediction of hydrological change in
western Canada, and the Global Water Futures Program, focused on managing
water futures in Canada and other cold regions where global warming is changing
landscapes, ecosystems, and the water environment. In 2018 he was the only nonUS
member of a US National Academies panel that reported on Future Water
Priorities for the Nation.
Dr. Denis Peach
Dr. Denis Peach spent nine years as the Manager of the British Geological Survey
(BGS) Groundwater Programme, then 6 years as BGS Chief Scientist. He is a
hydrogeologist with broad scientific interests and 49 years of experience, which
includes work for a UK water authority, overseas work in tropical hydrogeological
environments, small island hydrogeology and work for international consultants in
arid zone hydrogeology. His particular scientific interests include groundwater
modelling which he developed in BGS, arid zone hydrogeology, and Chalk and
Karst hydrogeology. He has been a Vice President of the Geological Society of
London (GSL), is a Visiting Professor at Imperial College, London and University
of Birmingham and he gave the Ineson Distinguished lecture at the Geological
Society of London in 2009. He has led numerous national geological and
ll
hydrogeological research programmes in the UK and has sat on many national
research programme boards and national water resource strategic committees. He
currently carries out research with BGS and Imperial College and recently provided
advice to the University of Saskatchewan and UK engineering consultants.
11
TABLE OF CONTENTS
LIST OF FIGURES ................................................................................................. V
1 INTRODUCTION ........................................................................................... 1
1.1 Background ............................................................................................... 1
1.2 Summary of the key areas of agreement and disagreement ..................... 1
1.3 Our five reports to the Court - background and summary ....................... 2
2 THE SILALA RIVER ...................................................................................... 3
2 .1 Introduction .............................................................................................. 3
2.2 Catchment definition and hydrollogical functioning ................................. 3
2.3 Geology and hydrogeology ...................................................................... 8
2.4 The historical channelization of the Silala River in Bolivia ................... 11
3 KEY AREAS OF AGREEMENT ................................................................. 15
4 KEY AREAS OF DISAGREEMENT ........................................................... 15
4 .1 Introduction ............................................................................................ 15
4.2 DHI's Water Balance and Near Field models ........................................ 17
4.3 Key areas of disagreement.. .................................................................... 19
4.3.1 Near Field Model boundary conditions .................................... 19
4.3.2 Model inconsistencies, inaccuracies, and instabilities .............. 20
4.3.3 Errors in geological and hydrogeological interpretation .......... 22
4.3.4 Wetland degradation ................................................................. 23
4.3.5 Could the flow from groundwater fed springs in the Cajones
and Orientales springs have been significantly enhanced by
the use of explosives? ............................................................... 25
4.4 Summary discussion ............................................................................... 26
5 CONCLUSIONS ............................................................................................ 26
6 REFERENCES .............................................................................................. 28
llll
LIST OF ]FIGURES
Figure 1. 3D topography with contour lines delimiting the surface water drainage
basin of the Silala River basin. International boundary (red line) and watershed
boundary (black line) as also shown in Figure 2 (top panel of Munoz et al., 2017,
Figure 3-3, at CM, Vol. 5, p. 182) ...... ...... ....... .. ...... .. ...... .. ...... .. ...... .. ...... .. ...... .. ...... 4
Figure 2. Longitudinal profile of the Silala River and main tributaries (Wheater and
Peach, 2017, Figure 4, at CM, Vol. 1, p. 143) .. ...... .. ...... .. ...... .. ...... .. ...... .. ...... .. ...... 5
Figure 3. Silala River topographic and groundwater catchments (Wheater and
Peach, 2019a, Figure 1, at CR, Vol. 1, p. 105) ... ... .... ..... .... ... .... ..... .... ... .... ..... .... ... .. 7
Figure 4. FCAB former Intake in Bolivia, FCAB Intake in Chile and pipelines
constructed and used by FCAB. The FCAB former Intake in Bolivia and Pipeline
N°1 (orange line) conducted water from Bolivian Territory to the FCAB reservoirs
at San Pedro Station (and on to Antofagasta). FCAB Intake and Pipeline N°2 (green
line) conducted water from Chilean territory, also to the San Pedro reservoirs
(Munoz et al. , 2017; Wheater and Peach, 2017, Figure 6, at CM, Vol. 1, p. 146) .
.. .. ...... .. ....... ....... ....... ......... ....... ......... ....... ........ .. ...... .. ...... .. ...... .. ...... .. ...... .. ...... .. .... 12
Figure 5. Domains covered by the difforent DHI models (Munoz et al., 2019;
Wheater and Peach, 2019a, Figure 2, at CAP, Vol. 1, p. 91) .. .. ...... .. ...... .. ...... .. .... 18
1 INTRODUCTION
1.1 Background
We, Drs. Howard Wheater and Denis Peach, have produced this written statement
at the request of the International Court of Justice, as expressed in a letter of
15 October 2021 from the Registrar, M. Philippe Gautier, to the Agent of the
Republic of Chile, Ms. Ximena Fuentes Torrijo. M. Gautier has asked for a
summary of our five reports, previously presented to the Court in Chile's Memorial
(CM), Reply (CR) and Additional Pleading (CAP).
In this introduction, we briefly summarise the key areas of agreement and
disagreement concerning the expert issues on which we have reported in this case
and present the background of our previous submissions to the Court. In section 2,
we provide a description of the Silala River and its hydrology and hydrogeology,
and in sections 3 and 4 we present the key areas of agreement and disagreement
between the experts. Our conclusions are presented in section 5.
1.2 Summary of the key areas of agreement and disagreement
Both sets of experts (i .e., ourselves and Bolivia's consultants, the Danish Hydraulic
Institute (DHI)) agree that the Silala River exhibits the properties of an international
watercourse. Whether as surface water or as groundwater, the waters of the Silala
River flow naturally down-gradient across the international border into Chile.
Channelization works carried out on the Bolivian side of the border in 1928 will
have had some limited effect on the flow of the Silala River. The experts disagree
as to the magnitude of this impact. We consider that impact to be very small. The
DHI experts, by contrast, consider the impact to be an increase in trans-border
surface flow in the region of 11 % to 33%. However, it is agreed by all the experts
that the channelization has not impacted the direction of flow of the Silala River
and, apart from the very small effects of channelization on evaporation, any increase
in surface flow in the river will be accompanied by a decrease in groundwater flows
across the border, and vice versa.
In other words, even though there is disagreement as to the precise impacts of the
channelization on surface water flows ,, it is agreed that notwithstanding the
channelization (and subject to the very small effects of channelization on
evaporation), all the waters of the Silala River continue to flow down-gradient
across the international border into Chile. Thus, the channelization has not, and
could not have, materially affected the quantity of water flowing into Chile.
ll
1.3 Our five reports to the Court - background and summary
In the pleadings before the Court, we have set out, in a series of joint reports, the
growing scientific evidence concerning the hydrological functioning of the Silala
River, together with our independent expert opinion on Bolivia's various
submissions to the Court.1 In these reports we address a series of questions posed
to each of us by Chile. The answers to questions addressed to Dr. Wheater have
therefore been drafted by Dr. Wheater, as first author, and similarly the answers
addressed to Dr. Peach, by Dr. Peach as first author. However, the joint authorship
of these reports reflects the fact that they represent our joint opinion.
Together with Chile's Memorial of July 2017 we submitted two reports, namely
The Silala River Today - Functioning of the Fluvial System ("Wheater and Peach
2017"), which focussed on the hydrological functioning of the river, and The
Evolution of the Silala River, Catchment and Ravine ("Peach and Wheater, 2017"),
which focussed on the geological and geomorphological history of the basin. In
both reports, a central question was whether the Silala River satisfies the criteria of
an international watercourse, which was answered affirmatively by both Dr.
Wheater and Dr. Peach, from a hydrological and a hydrogeological perspective,
respectively. In addition, Drs. Wheater and Peach provided a first opinion on the
impacts of the historical channelization, namely that the effects on surface flows at
the border would be insignificant, no more than 2%.
In Chile's Reply (February 2019) to Bolivia's Counter-Memorial (BCM) we
updated the scientific evidence and provided our independent expert opinions on
Bolivia's scientific evidence, in particular concerning the modelling by Bolivia's
international consultants, DHI, of the impacts of channelization of the Bolivian
wetlands on the surface water flow (DHI, 2018a). In Impacts of Channelization of
the Silala River in Bolivia on the Hydrology of the Silala River Basin ("Wheater
and Peach, 2019a"), we summarised the points of technical agreement between the
parties and explained our concerns about Bolivia's modelling. In addition to our
technical issues with the modelling, we identified Bolivia's erroneous interpretation
of the geology and hydro geology of the Silala River basin, which was addressed in
our second report, Concerning the Geology, Hydrogeology and Hydrochemistry of
the Silala River Basin ("Peach and Wheater, 2019").
Finally, in Chile's Additional Pleading (September 2019), following receipt of
Bolivia's digital data and in response to Bolivia's Rejoinder (BR), we reported on
our further analysis of Bolivia's modelling of the impacts of the channelization in
Bolivia (Impacts of Channelization of the Silala River System in Bolivia on the
Hydrology of the Silala River Basin - an Updated Analysis ("Wheater and Peach,
1 Under the joint direction of the authors of this Written Statement, a team of Chilean experts under
the leadership of Dr. Jose Mufioz, an expert on groundwater hydrology, conducted a series of
intensive studies, together with enhanced monitoring, which continues to the present time.
2019b" or "Updated Analysis, 2019"). This Updated Analysis, based on inspection
of the data files used to run Bolivia's models, showed further and very significant
modelling errors. We also reported an updated analysis of Bolivia's interpretation
of the geology and hydrogeology of the Silala River surface and groundwater
catchments upon which Bolivia's modelling was based. We concluded that
Bolivia's modelling was wholly unreliable and should be disregarded by the Court.
2 THE SILALA RIVER
2.1 Introduction
After initial site reconnaissance and review of ex1stmg scientific studies, a
substantial program of hydrological and hydrogeological studies was put in place
by Chile, on our recommendation, to better understand the hydrological functioning
of the Silala River, including surface water-groundwater interactions, and its
geological and geomorphological evolution. The groundwater flow regime was
investigated with drilling and pump testing, and detailed geological mapping and
hydrochemical surveys were carried out Much of this we reported in our two
reports of 2017 (Wheater and Peach, 2017; Peach and Wheater, 2017). However,
Bolivia's Counter-Memorial made a number of erroneous claims about the geology
and hydro geology of the Silala River basin and the impacts of channelization on the
Bolivian wetlands (the Cajones and Orientales wetlands)2 from which the Silala
River flows originate. As we are unable to observe the Bolivian wetlands directly,
we developed detailed studies of a similar wetland in the Silala River basin in Chile,
the Quebrada Negra wetland, which allowed us to better understand wetland
functioning in the basin and undertake comparative analysis of the Bolivian
headwater wetlands based on remote sensing data. This we reported in Chile's
Reply, together with a detailed analysis of Bolivia's geological and hydrogeological
interpretations. While some uncertainties remain, we now have a much-improved
understanding of the basin, which we summarise below.
2.2 Catchment definition and hydrological functioning
The Silala River is a typical groundwater-fed river.3 The perennial river flows
originate in groundwater springs in Bolivia, associated with the Cajones and
Orientales wetlands, at more than 4323 metres above sea level (m.a.s.1.), but the
river interacts with groundwater along its flow path. It receives substantial inputs
from groundwater springs that emerge from the wall of the Silala River ravine that
2 Denoted 'Northern' and ' Southern' respectively by Bolivia.
3 Many major rivers originate in perennial or ephemeral groundwater springs. The River Thames
(UK) is a notable example (British Geological Survey, 1996).
3
crosses the international border (at approximately 4277 m.a.s.1.) and loses water
from the flowing channel to an underlying fluvial aquifer (CM, Vol. 1, pp. 135,
168-169). A deeper groundwater system has also been identified, which currently
contributes flow to the river in Chile via discharge from an artesian wel14 (CM,
Vol. 1, pp. 135, 171-173).
The topography is such that natural drainage will flow from Bolivia to Chile from
a topographic catchment, shown in Figure 1. Figure 2 shows the longitudinal profile
of the river. The difference in elevation between the spring sources in Bolivia and
the river channel at the border is more than 45 metres, and the gradient of the natural
river channel is relatively steep (approximately 4-5%). In the vicinity of the border,
the river channel flows within a ravine that has been created by fluvial processes.
In Peach and Wheater (2017) we showed that the ravine provides evidence that a
river has flowed across what is now the international border, at this location, for
more than 8,400 years (CM, Vol. 1, pp. 218-223).
Figure 1. 3D topography with contour lines delimiting the surface water drainage basin
of the Silala River basin. International boundary (red line) and watershed boundary
(black line) as also shown in Figure 2 (top panel of Munoz et al., 201 7, Figure 3-3, at
CM, Vol. 5, p. 182).
4 An artesian well is one in which water pressure in the aquifer generates surface flows from the
well.
4
Cerro lnacaliri .t1
ode1Caj6n
5628m
lnacaliri Police
Station
CODELCO Intake
f Kilometers
Mercator Projection,WGS84
4450
LongitJdinal prt>file of R(o Silala
4400
Hilo 16-LXXl'I
Clo, de SIio/a
4850m
I
Les v~¥-
4350 ~ [',IP
]:
E
"~'
~
E
0
.)::
C
0
:,::;
~
(l) w
4300
4250
4200
km6,008 ~ ,.
,rn·s-:-204-1: (O' -~,f ~1
.,/~ km3.221 ~ 1:111
4150 ~ .,.----r- s !c. -o
:o'o~ ,,10\0 ... r-
4100
4050
7 ?
A/ _/
rrm 0.165
4000
0 1000 2000
I'll ·::
~
3000 4000 5000 6000
Distance from lnacaliri police station (m)
7000
-
Orientales
8000 9000 10000
Figure 2. Longitudinal profile of the Silala River and main tributaries (Wheater and
Peach, 2017, Figure 4, at CM, Vol. 1, p. 143).
5
The climate of the basin was described in Wheater and Peach (2017) (CM, Vol. 1,
pp. 154-161). Our estimate of average annual precipitation over the topographic
basin is 165 mm. In such a dry climate, evaporation (mainly from open water
surfaces and plant transpiration) is limited, over most of the area, by the available
precipitation. However, for wetland areas, springs provide water that can support
high rates of evaporation. We estimated that 78 mm of the annual precipitation is
discharged as river flow, and 87 mm is lost as evaporation from the catchment as a
whole. Using remote sensing methods, we estimated that the evaporation from
Bolivia's wetlands was equivalent to 0.7% of the river flow, but recognizing the
considerable uncertainty in this estimate, we suggested that 2% of the average flow
was an upper bound to this estimate.
Analysis of the water balance5 showed that the surface flows in the Silala River
cannot be supported by precipitation over the topographic catchment alone. In
Bolivia's Counter-Memorial, DHI identified a larger groundwater catchment
(BCM, Vol. 2, p. 275, Figure 5), with which we largely agree. Our best estimate of
this area is shown in Figure 3.
The groundwater that provides the spring flows that feed the Bolivian Orientales
and Cajones wetlands emerges from the Volcanic and Alluvial deposits that
underlie these wetlands. Many of these deposits are aquifers, which are supplied
with recharge waters derived from the precipitation, less evaporation, over the large
groundwater catchment (Figure 3). These aquifers are extensive throughout the
catchment in Bolivia and Chile. Hence groundwater will flow down-gradient across
the international border from Bolivia to Chile either as surface water, from the
springs, or as groundwater within the aquifers (CM, Vol. 1, pp. 167 and 168, Figures
20 and 21).
5 Simply stated, the difference between precipitation and evaporation provides the water available
for surface water and groundwater flows within and leaving the basin (including any abstraction for
public or industrial uses), neglecting seasonal and inter-annual changes in storage.
6
Kilometers
MercatorProjection,WGS84
I
""SILALA RIVER BASIN,
GROUNOWATER
Military
0 0 Post
""1entales
. d,Sllolo
Om
CATCHMENT
Figure 3. Silala River topographic and groundwater catchments (Wheater and Peach,
2019a, Figure 1, at CR, Vol. 1, p. 105).
As noted above, a central concern of Chile, articulated in its Application to the
Court of June 2016, was that the Court should recognize the Silala River as an
international watercourse. We concluded in Wheater and Peach (2017) ( CM, Vol. 1,
pp. 135-137), from our expert point of view, that the Silala River is without doubt
'a system of surface waters and groundwaters constituting by virtue of their
physical relationship a unitary whole and normally flowing into a common
terminus' , and that it is 'a watercourse, parts of which are situated in different states'
( which we understand to be the relevant definition from the 1997 UN Watercourses
Convention). The natural direction of flow is across the international border, from
Bolivia to Chile, and the 'common terminus' element is satisfied by the discharge
of Silala waters into the San Pedro River, and ultimately via the Loa River into the
Pacific Ocean.
We noted that in Bolivia's Counter-Memorial, DHI agree with this understanding
(Wheater and Peach, 2019a, at CR, Vol. 1, pp. 103-106). They confirm that the
Silala River is 'a coupled groundwater-surface water system[ ... ] extending across
the border' (BCM, Vol. 2, p. 266). They also note that numerous additional springs
occur downstream of the Orientales wetlands and add to the river flow (BCM,
7
Vol. 2, pp. 368-369). Further they note that 'groundwater level gradients and
hydrogeological properties clearly indicate groundwater flow from Bolivia to
Chile' (BCM, Vol. 5, p. 84). Whilethemagnitudeofcross-bordergroundwaterflow
remains uncertain, DHI estimate that ' the: groundwater flow across the border is at
least of the same order of magnitude as surface water discharge at the border'
(BCM, Vol. 5, p. 84).
2.3 Geology and hydrogeology
To understand the functioning of a groundwater-dominated catchment, it is
necessary to understand the underlying geology, which determines the
hydrogeological characteristics, e.g., the extent and properties of the aquifer
systems, as well as their inter-connectedness. Experts from the Chilean Geological
Survey, SERNAGEOMIN, working with Dr. Peach, have conducted extensive
studies, as reported in Chile's Memorial (SERNAGEOMIN, 2017), Reply
(SERNAGEOMIN, 2019a) and Additional Pleading (SERNAGEOMIN, 2019b). In
addition, various comments by Bolivia raised a doubt as to the nature and formation
of the ravine in which the river flows from Bolivia to Chile (e.g., CM, Vol. 3, p. 375;
BCM, Vol. 5, p. 119). Hence studies of the geomorphology of the river were carried
out by the Chilean expert team (Mao, 2017; Latorre and Frugone, 2017).
The current geology is a result of a long history of geological activity, summarised
below:
i) During the period from about 6 million to about 1.5 million years ago (Ma)
the area now occupied by the catchment of the Silala River was subject to
episodes of volcanism associated with the collision of the ocean tectonic plate
to the west (beneath the Pacific Ocean) and the South American continental
tectonic plate. This resulted in volcanic activity that shaped the landscape,
including the building of the Cerro Inacaliri, Cerrito de Silala and the Volcan
Apagado (CM, Vol. 1, pp. 199-200), which are all dominant features of the
catchment morphology (Figure 1 ).
ii) The foundations of these edifices are formed of largely volcanic domes and
lavas dated at around 6.6-5.8 Ma. Upon these older basal rocks, which can be
found beneath the Silala River ravine, are deposits called Ignimbrites. These
were emplaced by explosive volcanic eruptions extruding flows of rock
fragments, molten rock droplets and hot gases, which flowed down the
existing topographic gradient at great speed (CM, Vol. 1, pp. 199-200). The
first of these (Cabana Ignimbrite, ca 4.12 Ma) was a very extensive and
voluminous event affecting a large area of the Chilean Altiplano. This was
followed by a first period of fluvial activity, which eroded a valley in the
ignimbrite and left fluvial sediments (CR, Vol. 3, pp. 208-209). On top of
these early fluvial deposits a further ignimbrite (Silala Ignimbrite ca 1. 61 Ma)
8
was deposited, probably filling the valley. Subsequently further volcanism
led to a massive lava flow being erupted from the Inacaliri volcano (1.48 Ma)
which flowed into the headwater area of the Silala River. This lava flow
truncated the then-existing drainage network of the Silala River ( CM, Vol. 1,
pp. 208-215; CR, Vol. 1, pp. 179-183; SERNAGEOMIN, 2017;
SERNAGEOMIN, 2019a).
iii) There appears to have been a hiatus in volcanic activity in the catchment after
1.48 Ma, and the next events to impact the catchment morphology were
associated with the glaciation of the high peaks, above 4400 m.a.s.1. There is
no evidence of glacial erosion or glacial deposits to be found at the level of
the current Silala River ravine or in the ravine. The cutting of the Silala River
ravine, as we know it today, was caused by fluvial processes. It began in the
period ca 12,000-8,400 years ago and continues today. Radio-carbon dating
has shown that there are sediments deposited by the current Silala River
system in the ravine that are more than 8,400 years old. The river began
cutting the ravine before that, probably as a result of the melting of the
glaciers about 12,000 years ago that caused significant runoff and increased
flow in the river and continues in a cycle of erosion and deposition in response
to climatic regime changes (CM, Vol. 1, pp. 218-223; Latorre and Frugone,
2017).
Features of fluvial erosion are common in the sides of the ravine. There are four
water-cut river terrace surfaces and four sedimentary sequences of deposits several
metres thick (CM, Vol. 1, pp. 218-223; Arcadis, 2017). These deposits include
sands, gravels, silts and organic remains of wetlands. The sides of the ravine contain
minor wind erosional features, and there are some windblown sand deposits to be
found, but these are minor features, and would have had no significant impact on
the ravine formation (CM, Vol. 1, pp. 227-233; SERNAGEOMIN, 2017).
Archaeological surveys have found artefacts and shelters or temporary dwellings
along the course of the river, mainly on the upper three terraces (CM, Vol. 1,
pp. 224-225; McRostie, 2017). These testify to the human use of the river and its
course over the past at least 1,500 years. There is no doubt that the geological,
geomorphological, and other evidence points definitively to the historical existence
of a fluvial system in the Silala River catchment. The modem ravine, created by
fluvial action, has existed for more than 8 millennia (CM, Vol. 1, pp. 218-225;
Latorre and Frugone, 2017).
These geological processes and events have formed the landscape of the Silala
River catchment and ravine as we know it today ( a schematic cross-section through
the Silala River ravine and Cerro Inacaliri and V olcan Apagado can be found in
CR, Vol. 1, p. 190, Figure 3-6). We noted that the hydrological regime is not only
a reflection of the climate and meteorology, but of the nature and topography of the
land surface and the rocks found in the subsurface. The current topography
(Figure 1) and river profile (Figure 2) are a direct result of the interaction of the
9
atmospheric processes, solid earth processes and biological processes and their
variability over the last 6 million years .. The natural gradients of the landscape
topography and the river channel are such that the river must flow naturally from
Bolivia to Chile. Similarly, the current groundwater level gradient indicates a
natural flow from Bolivia to Chile (CM, Vol. 1, p. 167, Figure 20), as agreed by
DHI (BCM, Vol. 2, p. 266).
We also note, based on studies of the fluvial geomorphology (Mao, 2017), that the
current fluvial system continues to be geomorphologically active; we have observed
size-selective transport of fine and coarse sediments and bed armouring, 6 and the
current channel morphology of steps and pools is consistent with that needed to
transport the current flow and sediment loads. The river also maintains flourishing
populations of fish and invertebrates, an indicator of aquatic ecosystem health
(Mao, 2017).
It is clear from our investigations reported in Chile's Memorial and Chile 's Reply
that the hydrogeology of the groundwater catchment is highly complex, but we have
found three distinct aquifer systems that are active in Chile (Arcadis, 2017):
i) A fluvial aquifer that is found beneath the bed of the Silala River and within
the ravine (CM, Vol. 1, pp. 166--169). These deposits are composed of
sediments laid down by the river and associated riparian wetlands. They
support minor groundwater flows but display a distinct groundwater level
different from the perched and regional aquifers described in (ii) and (iii)
below.
ii) A perched aquifer system that is present in alluvial deposits that overlie the
bedrock volcanic formations found in the Silala River basin, as evidenced
from geophysical investigations, spring flows into the Silala River, in
particular from the northern side of the Silala River ravine (CM, Vol. 1,
pp. 168-169), and confirmed by hydrochemical analyses that show the
distinctly different nature of the water from deeper groundwaters (Herrera
and Aravena, 2017; Herrera and Aravena 2019).
iii) A regional aquifer system that was formed by a succession of ignimbrite
deposits of variable permeability, which are interbedded with fluvial deposits
(providing high permeability). The groundwater found in this aquifer has a
distinctly different hydrochemical signature to those of the perched aquifer
(CM, Vol. 1, pp. 171-172). This aquifer is recharged from the extensive
groundwater catchment (Arcadis, 2017; BCM, Vol. 2, p. 275, Figure 5) (See
Figure 3 above).
It is also clear that recharge to these aquifers in the groundwater catchment, most
of which lies in Bolivia, either emerges at the Bolivian wetland springs or the
6 Gravel-river beds typically have an 'armoured' layer of coarse grains on the surface, which acts to
protect finer particles underneath from erosion.
Chilean springs downstream of the international border, or flows within the regional
ignimbrite aquifer down gradient through Chile to the southwest. The vertical
variability of permeability in the ignimbrites is demonstrated by the artesian
overflowing well, SPW-DQN, and implies a confining, low permeability layer (CR,
Vol. 1, p. 216).
The differences in hydrochemistry and carbon isotope content between the Cajones
and Orientales wetland spring waters are marked and indicate different origins for
the groundwater issuing from the two sets of springs. The Cajones waters are
probably derived from recharge more locally and show similarities to the
groundwaters emerging from the perched aquifer springs emerging from the ravine
wall in Chile. However, the groundwaters emerging from the Orientales springs
show close similarity to the groundwater flows found at depth such as those that
enter the Silala River from the artesian borehole SPW-DQN (CR, Vol. 1, pp. 201-
213; Herrera and Aravena, 2017; Herrera and Aravena, 2019).
Geological mapping by SERNAGEOMTI\f, in Chile, has found no evidence of the
' Silala fault ' as proposed by Bolivia (BCM, Vol. 4, pp. 69-81 , and p. 75, Figure 27)
in the Silala River ravine, but several faults downstream in Chile indicate that the
regional aquifer is only found at depth beneath low permeability Pliocene lavas
(CAP, Vol. 2, pp. 214-217), and it is likely that groundwater flow further downgradient
would be limited.
2.4 The historical channelization of the Silala River in Bolivia
Although, as noted above and in Section 3 below, there is agreement between
ourselves and Bolivia's experts that the Silala River has the characteristics of an
international watercourse, and broad agreement about the nature and functioning of
the catchment, including cross-border surface and groundwater flows, there are
remaining scientific differences between the experts. Apart from the interpretation
of the geology and hydrogeology, discussed above, a key difference is focussed on
the effects of historical channelization of the river system in Bolivia, undertaken
for sanitary reasons in the context of water supply (CM, Vol. 1, p. 98).
In the context of the social and economic development of a hyper-arid region, the
Silala River has historically been an important regional water source for Chile. In
1906 a concession was granted by Chile to a British company, the Antofagasta
(Chile) and Bolivia Railway Company Ltd. (FCAB) to supply drinking water to the
port city of Antofagasta (CM, Vol. 1, p. 40). Two years later, in 1908, FCAB
secured rights to use the waters of the Silala River from Bolivia. We understand
that engineering works were constructed during the period 1909-1910 to enable
flow diversion into a pipeline from the Silala River in Bolivia. The concession
continued until terminated by Bolivia in 1997 (CM, Vol. 1, p. 42). A second intake
11
and pipeline were constructed by FCAB in 1942 on Chilean territory. These points
of water withdrawal were just downstream and just upstream of the international
border, as shown in Figure 4. The figure also shows the location of a further
withdrawal point, some distance downstream of the FCAB pipeline intakes, which
was implemented in 1956 by the Chilean state-owned mining company CO DELCO,
for domestic water supply to one of its copper mines. Further details can be found
in Wheater and Peach (2017) (CM, Vol. l, pp. 145-147).
CODELCO Intake
<- .
; ·Ar
f 1
600 (\ 1200
Meters
Mercator Projection,WGS84
1800
Q""6 ,.,,c1,
'!(Vegra -·-._-·--· -
J ~ .
(~( ~ ""'''"''"' ( - ""'""' ,. '
Figure 4. FCAB former Intake in Bolivia, FCAB Intake in Chile and pipelines constructed
and used by FCAB. The FCAB former Intake in Bolivia and Pipeline N° 1 (orange line)
conducted water from Bolivian Territory to the FCAB reservoirs at San Pedro Station
(and on to Antofagasta). FCAB Intake and Pipeline N°2 (green line) conducted water
from Chilean territory, also to the San Pedro reservoirs (Munoz et al., 2017; Wheater
and Peach, 2017, Figure 6, at CM, Vol. l,p. 146).
12
In 1928, FCAB constructed a network of small channels (0 .6 m wide x 0.6 m deep)
in the Bolivian Orientales and Cajones wetlands, together with some additional
channelization of the main river in Bolivia (CM, Vol. 1, p. 42). Constructed as earth
channels, and lined with stone, these would act as drains, allowing ingress of water
from the wetlands, and loss of water to adjacent soils. The purpose of the channels
was to avoid contamination of the water with eggs of green flies that were breeding
in the vegetation through which the river was flowing (CM, Vol. 1, p. 98). The
history of maintenance of these channels is unclear, though DHI notes (BCM,
Vol. 2, pp. 281-282) that in recent years, in parts of the Southern (Orientales)
wetland, the canal and drains have been removed, filled in or blocked, in partial
attempts at wetland restoration.
Bolivia's experts agree with us that the channelling of flow on Bolivian territory
has not influenced the river flow direction, which follows the natural topographic
gradients (BCM, Vol. 2, p. 267). Flow across the border in the present ravine has
occurred for at least the last 8,400 years and long predated the concessions to FCAB
and the later construction of a system of small channels. However, it can reasonably
be expected that the channelization willl have had some limited effect on the
generation of surface flows in Bolivia's wetlands, with potential effects on the
extent and health of the wetland vegetation, and on the downstream transmission of
surface flows in channelized sections of the river. It should be noted, however, that
any effects on surface flow, other than by increased or decreased evaporation would
be accompanied by a compensating effect on groundwater flows down-gradient to
Chile. In lay terms, whether as surface or as groundwater, all the waters of the Silala
River inevitably flow downhill from Bolivia into Chile. Any increase in surface
water flow, whether minor (according to us) or significant (according to DHI) could
not somehow lead to a material increase in the overall quantity of water flowing
into Chile.
The principal hydrological effect of the drainage channels is to reduce the elevation
of the groundwater water table in the vicinity of the channels. Instead of
groundwater emerging at the wetland surface, it will flow into the drain. This means
that at the drain location, the water table will be drawn down to just above the base
of the drainage channels (0 .6 m), instead of at the ground surface. With increasing
lateral distance from the channel, the water table will of course have a higher
elevation. 7 Evaporation of water from an open water surface will typically be higher
than water transpired by vegetation. However, since in the wetlands the water table,
even at a depth of 0.6 m, is relatively close to the surface, evapotranspiration rates
are expected to be close to those from open water, and hence any differences in
evaporation will be small.
7 Bolivia's soils data show water table depths ranging from 0.1 to 0.4 min the Northern wetland and
from 0.15 to 0.45 in the Southern wetland (BCM, Vol. 3, pp. 12-13).
13
Our preliminary calculations (Wheater and Peach, 2017, at CM, Vol. 1, pp. 161-
164) showed that even under the most conservative assumptions, evaporation from
these wetlands is a very small component: of the water balance of the Silala River. 8
From remote sensing data, we estimated this to be equivalent to 0.7% (1.3 1/s) of
the river flow at the border but, recognizing the uncertainty in this estimate, we
suggested an upper limit of2% (3.41/s) of the river flow at the border. Clearly, even
if some reduction of this evaporation had occurred due to channelization (it should
be noted that we show in later reports that that has not been the case), minor changes
to this very small element of the catchment water balance would have had no
significant effect on flows at the border. Further, the channels, as we understand,
have (until very recently) not been maintained since 1997 (CM, Vol. 1, p. 42), and
we see no evidence of a change in flow regime at the border. In fact, satellite data
shows the wetland extent to be dominated by large natural seasonal and inter-annual
variability (CAP, Vol. 1, pp. 137-140).
In Bolivia's Counter Memorial, Bolivia's: consultants, DHI, raised further possible
effects of the channelization. They agreed with us that changes in evaporation could
be expected due to drainage of the wetlands, but also suggested (BCM, Vol. 2,
p. 276) that the channels would increase surface water discharge 'due to lowering
of the hydraulic head loss by removal of peat or constraining rock cover' . It was
stated that at the spring discharge points, 'the soil and any underlying layers of
coarser material or rocks have been completely removed' (BCM, Vol. 2, p. 276).
These effects were conceptualised for modelling using two scenarios, one in which
the channels were removed (the 'No Canals' scenario), and a second (the 'Wetland
Restoration' scenario) in which assumed long-term peat accumulation was
simulated. They also suggested that the channelization would affect the interactions
between flowing surface water and groundwater, reducing seepage losses from the
channels to underlying groundwater.
We agreed (Wheater and Peach, 2019a, at CR, Vol. 1, pp.106-109) that the
reduction of water table elevation due to the installation of drainage will increase
the gradient of groundwater spring flow to the stream, and hence increase the
groundwater discharge to the river, and that further accumulation of peat, which has
a relatively low hydraulic conductivity, could, in the long term, produce an
additional resistance to groundwater flow due to the peat cover, thereby reducing
the surface flow. We also agreed that there could be changes to stream-groundwater
interactions. However, as we stated in Wheater and Peach (2019a) in our opinion,
any changes in surface flow due to these effects will be very small. We return to
these issues in section 4, below.
8 Although wetland evaporation rates are high, they are associated with relatively small areas.
14
3 KEY AREAS OF AGREEMENT
We, and Bolivia's experts, agree on several key points concerning the nature and
functioning of the Silala River. In summary, these are:
i) The Silala River flows naturally from Bolivia to Chile. The river rises in two
sets of springs in Bolivia, which maintain the Cajones and Orientales
wetlands (BCM, Vol. 2, p. 266; CM, Vol. 1, p. 177).
ii) The river is primarily fed by groundwater and interacts with groundwater
along its course to the border and beyond (BCM, Vol. 2, pp. 368-369; CM,
Vol. 1, p. 177).
iii) In addition, there are substantial groundwater flows from Bolivia to Chile,
possibly of an equivalent magnitude to the surface water flows (BCM, Vol. 2,
p. 266; CR, Vol. 1, p. 104).
iv) In summary, the Silala River is a coupled groundwater-surface water system,
extending across the border (BCM,, Vol. 2, p. 266; CM, Vol. 1, p. 177) and
hence it appears to be accepted that it is an international watercourse.
v) Construction of the channels in the 1920s on Bolivian territory has not
influenced the river flow direction, which follows the natural topographic
gradients (BCM, Vol. 2, p. 267; CM, Vol. 1, p. 178).
vi) This channelization will have had some effect on the surface water flow of
the Silala River. An increase in river flow due to these works would be
expected (BCM, Vol. 2, p. 266; CM, Vol. 1, p. 178). (As discussed further in
section 4 below, we consider that this impact will be very small).
vii) Some impact of the drainage channels on evaporation from the wetlands
would be expected but is small (BCM, Vol. 2, p. 303; CM, Vol. 1, p. 178).
(As discussed further in section 4 below, we consider that this impact will be
very small).
viii) Apart from the effects of channellization on evaporation, any increase in
surface flow in the river will be accompanied by a decrease in groundwater
flows across the border, and vice versa (BR, Vol. 5, p. 30; CR, Vol. 1, p. 108).
4 KEY AREAS OF DISAGREEMENT
4.1 Introduction
There is one over-riding point of disagreement between ourselves and DHI, which
has been a central issue for Bolivia in its Pleadings before the Court. This concerns
the magnitude of impacts of channelization on surface water flows.
We had noted (Wheater and Peach, 2017, at CM, Vol. 1, p. 134) that the
channelization in Bolivia could have an effect in reducing evaporation, and hence
increasing surface water flows across the border, but calculated that this effect
15
would be very small. DHI agreed, both that this was a likely effect, and that it would
be small.9
We agreed (Wheater and Peach, 2019a, at CR, Vol. 1, pp. 106-109) that the
installation of drainage will increase the gradient of groundwater spring flow to the
stream, and hence increase the groundwater discharge to the river, and that further
accumulation of peat could, in the long term, produce additional resistance to
groundwater flow, thereby reducing the surface flow. We also agreed that there
could be changes to stream-groundwater interactions. However, as stated in
Wheater and Peach (2019a), in our opinion, any changes in surface flow due to
these effects will be very small.
DHI's surprising conclusion was that the overall effects of channelization on
surface flows would be large. DHI stated in Bolivia's Counter-Memorial (BCM,
Vol. 2, pp. 266-267), 'Without canals[ ... ] [a] reduction of surface flows of30-40%
is estimated compared to current conditions'. 1° Following our critique of their
modelling in Wheater and Peach (2019a), DHI revised their estimates in Bolivia's
Rejoinder, but nevertheless continued to assert very large effects: ' .. .. the simulated
range of decrease in transborder surface flow when removing the canals is 11 %-
33 % ' (BR, Vol. 5, p. 56). 11 We have consistently stated that these estimates are
wholly implausible and that, given the relatively small reductions in groundwater
table depths associated with the channelization of the wetlands and of the main
river, any effects will be very small.
Most importantly, both we and DHI agree that any increase in surface flows due to
the channelization would have been accompanied by a decrease in groundwater
flows from Bolivia to Chile, and vice versa (CR, Vol. 1, p. 107-108). In later
pleadings, Bolivia agreed that 'with no canals, less water enters the surface water
system and more enters the groundwater' (BR, Vol. 5, p. 30). Recharge to the
groundwater system occurs over the large catchment shown in Figure 3 and is not
affected by changes in downstream surface or groundwater flows. Since the
groundwater flows from Bolivia to Chile., any difference in the combined flows of
surface water and groundwater from Bolivia to Chile will be mainly due to the
difference in wetland evaporation losses, 12 which as noted above, both sides have
agreed is small.
9 Our original best estimate was a maximum increase of 1.3 1/s in surface water flow (CM, Vol. 1,
p. 161), with an upper bound of 3.4 1/s (CM, Vol. 1, p. 164). DHI estimated this effect to be
equivalent to 2-3 1/s of river flow (BCM, Vol. 2, p. 303).
10 30% is the effect of a scenario of channels removed, 40% of a scenario of channels removed and
an assumed regrowth of peat soils.
11 33% and 11 % are DHI's upper and lower bounds for channel removal only, with no assumed peat
regrowth.
12 Other changes to evaporation may arise due to the channelization of the main river channel, and
the associated interconnection between surface water and groundwater, but these will be minor.
16
Bolivia's estimates depend on simulations carried out by DHI using a widely used
and respected series of models. The question thus arose for us - how could these
models produce such unrealistic effects? In Wheater and Peach (2019a), based on
the limited information provided in Bolivia's Counter-Memorial, we noted errors
in the modelling, some associated with technical issues, particularly the small scale
of the simulations and the associated model boundary conditions, and others
concerning the underlying geology on which the model was based. Subsequently,
we were provided with the digital data used by DHI to run their models, and were
able to see multiple errors and unexplained assumptions.
In the sections below we first introduce DHI's models, and then explain why DHI's
simulations are incorrect. We summarise the series of very serious errors we found
in the modelling, including errors in the geology, and address incorrect assertions
made by Bolivia concerning wetland shrinkage and degradation. We also include a
short comment on Bolivia's unfounded assertions concerning the use of explosives
to increase the yields of the Bolivian springs.
4.2 DHl's Water Balance and Near F'ield models
Bolivia's experts established a suite of models to simulate the Silala River system
(CAP, Vol. 1, pp. 89-91) (Figure 5). A Water Balance Model was used, based on
the MIKE-SHE hydrological modelling software, to simulate the water balance of
the topographic catchment and a larger groundwater catchment, estimated to be
234.2 km2
. However, the modelling results used to estimate the effects of
channelization and peat accumulation were based on the simulation of a very small
area (2 .56 km2
) around the river and wetland spring areas in Bolivia, named by DHI
as the Near Field. This Near Field modelling was carried out using a combination
of two models: (i) the MIKE-SHE hydrological model was used alone for the two
scenarios without channelization ('No Canals' , and 'Wetland Restoration', as
discussed above), and (ii) for the scenario representing channelization ( the
'Baseline' scenario), the MIKE-SHE hydrological model was linked to the
MIKE-11 hydraulic model, which represented the detail of flow in the river
channels (BCM, Vol. 5, p. 11).
17
Cerro I caliri
ode/ rj6n
562 m
lnacaliri Police
Station
~ CODELCO Intake
Laguna
Blanca
Laguna
Chica
NEAR BORDER MODEL (NBM)
WATER
BALANCE
MODEL
(WBM)
Figure 5. Domains covered by the different DH! models (Munoz et al., 2019; Wheater
and Peach, 2019a, Figure 2, at CAP, Vol. 1, p. 91).
It is relevant to note that the results of the Water Balance Model (BCM, Vol. 3,
Annex E), which calculated recharge from the natural input variables of rainfall less
evaporation, were not used in the Near Field Model (BCM, Vol. 5, Annexes G and
H). Rather, assumptions were made by DHI concerning the groundwater elevations
and/or flows at the boundaries of the Near Field Model (known as the model
boundary conditions), as discussed below. We note that the recharge calculated
using the resulting flow into the Near Field Model was different from the Water
Balance Model results, and that the different Near Field Model scenarios had
different recharge values ( see section 4 .3 .1 below). That cannot, of course, be the
case. In reality, all the recharge to the Ignimbrite aquifer(s) from the extended
groundwater catchment flows into the Near Field Area. 13 The precipitation and
groundwater catchment areas are essentially unchanged between scenarios, and any
changes to evaporation are acknowledged by DHI to be small.
13 DHI's incorrect interpretation of the geology (see CR, Vol. 1, pp. 179- 201 , in particular Figures
3-6 and 3-7) allows groundwater flows to bypass the Near Field, thereby providing DHI with an
erroneous justification for the changing inflows (BR, Vol. 5, pp. 28-30).
18
4.3 Key areas of disagreement
4.3.1 Near Field Model boundary coinditions
The choice to model such a small part of the catchment area of the river system as
the Near Field means that the modelled flows will be mainly determined by the
assumed boundary conditions for the model: the assumed boundary conditions used
by DHI were inappropriate. In particular, water table conditions at the model
upslope boundary were fixed, whereas in reality, water table conditions are not at
all fixed. The changes due to the removal of channels and the hypothetical longterm
accumulation of peat cover that were proposed by DHI to have such large
effects on the groundwater discharge to the stream would also affect the conditions
at the model boundary, given its close proximity to the stream.
One obvious effect of the inappropriate boundary conditions is that the inflows to
the model changed significantly for the different scenarios investigated by DHI.
And clearly, because the inflows to the model change, the model outputs change,
too. The inflow to the Near Field Model was 253 1/s for the Baseline (with
channelization) scenario, but 216 1/s for the Restored wetlands scenario with
channels removed and assumed peat regrowth (BCM, Vol. 5, p. 67, Table 1). A
difference in combined surface water and groundwater outflows of 49 1/s was
reported for the different scenarios, of which 37 1/s was generated solely by
inappropriate changes to boundary inflows. As noted above, in reality, the recharge
from the groundwater catchment will be essentially unchanged for the three
scenarios and can only flow to Chile - either as surface water or as groundwater.
In Wheater and Peach (2019a) (CR, Vol. 1, pp. 114-125), we demonstrated, using
simple calculations as an example, that the erroneous boundary assumption will
exaggerate the effects of water table rise and peat cover, and perhaps explain DHI's
exaggerated estimates. In our opinion, in the context of the very large groundwater
elevation differences that determine the groundwater flow (150 metres, according
to DHI's modelling (BCM, Vol. 3, p. 488, Figure 11)), the effects of lowering the
water table by less than 0.6 metres and long term growth of peat cover (assumed by
DHI to be up to 0.6 metres (BCM, Vol. 5, p. 70)) will be minor, a few per cent at
most of the cross-border surface flow. We estimated the order of magnitude of these
combined effects to be a 1.2% change in river flow (CR, Vol. 1, p. 124). Although
based on a major simplification ofreality (a 2-dimensional hillslope segment), this
analysis nevertheless indicated the likely order of magnitude of the effects.
Also importantly, as noted above, any increase in surface water flow would be
accompanied by a corresponding decrease in groundwater flow, the former flowing
down the topographic gradient, and the latter down the groundwater hydraulic
gradient to Chile.
19
In their report (DHI, 2019) attached to the Bolivian Rejoinder (15 May 2019), DHI
accepted our criticism of the model boundary conditions used to simulate the effects
of channelization and accepted that its calculations had overestimated the effects
(BR, Vol. 5, p. 55).
DHI presented revised results, in which 1the previous results were described as an
upper limit. A different approach was taken to the Near Field boundary condition
to define a lower limit, and hence a new range of impacts was specified for the
effect of channelization. It was said: 'if the channels and drainage mechanisms were
removed, cross-border surface flows in the Silala River would decrease by 11 % to
33% of current conditions [ ... ] evapotranspiration from wetlands without canals
will increase by 28% to 34% of the reference values, i.e. between 2.8 and 3.4 1/s,
while groundwater flows across the[ ... ] border will increase between 4% and 10%
as compared to current conditions ' (BR, Vol. 1, p. 35). However, even the lower
limit ( an 11 % decrease in surface water flows) gave an implausibly high estimate
of the effect of channelization in our opinion. We reiterate that recharge into the
groundwater catchment has remained the same and losses to cross-border flows are
confined to the effects of increased evaporation, so these new DHI numbers are
impossible. It therefore seemed that other errors were likely to be present in DHI's
modelling.
We noted a further point on our concern for the boundary conditions in Wheater
and Peach (2019a) (CR, Vol. 1, p. 125). The field observations reported by DHI
(BCM, Vol. 5, p. 49, Figure 35) were inconsistent with the assumed Near Field
Model lateral boundary conditions. This was consistent with other concerns for the
accuracy of the geology used by DHI to define their Near Field Model, which we
explain in section 4.3 .3 below.
4.3.2 Model inconsistencies, inaccuracies, and instabilities
In Wheater and Peach (2019a), we noted various other inconsistencies in DHI's
Counter-Memorial results (CR, Vol. 1, pp. 126-127). However, the digital data
provided by Bolivia in February 2019 (after a repeated request) allowed more
detailed evaluation of DHI's modelling. Analysis of model configurations,
parameters, input data and simulation results showed that there were many aspects
of the modelling that gave rise to serious concern for the reliability of the results,
in particular, for the modelling of the 2.56 km2 Near Field area on which Bolivia's
estimates of the effects of channelization were based.
Inspection of the digital data by Chilean hydrologists (Mufioz et al., 2019) revealed
many unreported differences between the DHI models used for the intercomparison
of scenarios, and in the MIKE-SHE model's boundary conditions and
initial conditions. These unreported differences were compounded by unexplained
methodology, and incorrect assumptions. We mention a few of these below; a more
20
comprehensive explanation can be found in Mufioz et al. (2019) and Wheater and
Peach (2019b), at CAP, Vol. 1, pp. 100-118.
i) Perhaps of greatest impact was the fact that we found that different
topographies had been used for modelling the different scenarios, including
different topographies used in the Baseline Scenario for the modelling of
catchment processes ( the MIKE-SHE model) and the modelling of channel
flow (the MIKE-11 model). These differences in topography, of up to
7 metres, were far greater than the small changes in channel depth and peat
growth that the models were being used to evaluate, and in themselves would
generate large differences between the scenarios (CAP, Vol. 1, pp. 103-104,
Figures 5 and 6). It follows that the differences in topography used were
clearly not warranted.
ii) We also found unexplained additions of water. In a physically-based model
of the Near Field area, we would expect the groundwater inputs to reflect the
physical reality that the springs are fed from groundwater inflows at the model
boundaries. However, from the DHI model files it was clear that, in addition
to the boundary groundwater inflows, extra water had been introduced into
the model as an external input, with no explanation or justification. Some
42 1/s was introduced to the Baseline Scenario (i.e., the situation with
channelization) as so-called 'spring recharge ', whereas only 31 1/s was input
to the two scenarios representing no channelization. Clearly a difference of
11 1/s had been introduced into the scenario comparisons, an amount that
accounts for more than half of the DHI reported simulated changes in surface
flows due to the channelization (CAP, Vol. 1, pp. 110-111). These
introductions of unaccounted-for water amount to the invention of that water,
with no physical justification. By introducing this invented water, DHI
artificially increased the simulated effect of the channelization.
iii) Very large differences were also found in the assumed initial conditions, i.e.,
the initial groundwater elevations, for the different scenarios. The differences
varied between -18 m and +16.5m (CAP, Vol. 1, p. 110, Figure 9). The Near
Field Model is a dynamic (time-varying) model, and while it was run to
approximate a steady-state condition, the model shows large transient
instabilities, so that such large differences in initial conditions would be
expected to affect the simulation results.
iv) The results reported to the Court were exaggerated due to instabilities in the
DHI model outputs, illustrated in VVheater and Peach (2019b) (CAP, Vol. 1,
p. 115, Figure 12). These instabilities were mainly associated with the MIKE-
11 model and arose partly due to numerical errors in the DHI modelling, and
partly due to inconsistencies in DHI's representation of channel topography.
v) Additionally, very high channel roughness values were used by DHI in the
hydraulic modelling, which gave rise to slower velocities than expected, and
21
larger flow depths, which is perhaps why the model erroneously simulated
water flows outside the main channel in places (CAP, Vol. 1, p. 111).
While the reported model errors and inaccuracies for the Near Field Model were of
a similar magnitude to the effects being simulated, which in itself casts doubt on
the validity of the conclusions from the modelling, we conclude that the large
effects proposed by DHI are mainly an artefact of these unreported differences
between the modelled scenarios. We note that the largest numerical errors were
associated with the MIKE-11 hydraulic model, and that different topographies were
used for the MIKE-SHE and MIKE-11 modelling for the same scenarios. The fact
that MIKE-11 was used for the Baseline simulation (i.e., with channels), but not for
the 'No Channel' and 'Restored Wetland' scenarios adds a further major
inconsistency to the scenario inter-comparisons. Indeed, we have subsequently
observed that, when the DHI models are run with more realistic data with respect
to topography, and when the numerical errors in the MIKE-11 model are addressed
and the two models are used consistently for all scenarios, the results are in line
with our estimates.
4.3.3 Errors in geological and hydro,geological interpretation
A large number of errors and inconsistencies have been found in Bolivia's
geological mapping and structural geology analysis (SERNAGEOMIN 2019a;
SERNAGEOMIN, 2019b). These were incorporated by DHI in their own
conceptual understanding of the hydrogeology and hence in the Near Field Model.
Consequently, DHI's interpretation of the hydrogeology and its implementation in
the Near Field Model contains many errors, as has been detailed in Wheater and
Peach (2019b ), at CAP, Vol. 1, pp. 119-13 7, the most important of which are listed
below:
i) An error in the assignment of a radiometric date to establish the age range of
the Ignimbritas Silala (Bolivian name) leading to an incorrect interpretation
of the stratigraphy (the rock layering). This has important impacts on aquifer
geometry and the distribution of permeability in the Near Field Model, the
ignimbrite aquifer having a much more restricted areal extent than proposed
by Bolivia (CAP, Vol. 1, pp. 122-128).
ii) Bolivia has ignored the existence of the Silala and Cabana Ignimbrites in their
establishment of the Ignimbrite stratigraphy. The Silala Ignimbrite is highly
welded, and outcrops unconformably over much older Ignimbrites in the
Orientales wetland. The Cabana Ignimbrite is highly permeable. Both have a
limited lateral extent and are constrained between two hills of Miocene low
permeability volcanics in Bolivia, which limits the flow of groundwater
through this region. This impacts on the Near Field Model parameterization
and the aquifer geometry incorporated into the Near Field Model (CAP,
22
Vol. 1, pp. 122-128). This means, for example, that DHI's incorrect
interpretation of the geology allows groundwater to bypass the Near Field,
whereas in reality it must all flow through this area.
iii) The Silala Fault, invoked as a high-permeability groundwater pathway by
DHI, does not exist, could not be related to tectonic events that took place
millions of years before the ignimbrites or the Miocene Volcanics were
deposited and cannot be used to specify narrow high-permeability zones
running down the Cajones, Orientales and Silala River ravines in an
impossible sinuous manner (CAP, Vol. 1, pp. 128-130; CAP, Vol. 2, pp. 214-
221).
iv) The Bolivian structural analysis is flawed. This has led to erroneous
interpretations in the structural geology, which has then led to the false
assumption of the presence and location of open fractures able to conduct
groundwater, so there is a likelihood of incorrect assignment of aquifer
properties in both conceptual and numerical modelling. (CAP, Vol. 2,
pp. 212-235).
v) DHI has ignored Chilean evidence of a shallow aquifer system, which is
supported by geophysical and hydrochemical evidence. Although DHI has
acknowledged two sources of groundwater supplying the Bolivian wetland
springs, these have been ignored in the construction of the Near Field Model,
leading to incorrect interpretation of the groundwater water table distribution
and groundwater flowpaths (CAP, Vol. 1, pp. 132-133; Peach and Wheater,
2019; Arcadis, 2017; SERNAGEOMIN, 2019a; Herrera and Aravena, 2017;
Herrera and Aravena, 2019).
vi) The DHI conceptual model of groundwater flow and potentiometric contours
used for the Near Field Model (BCM, Vol. 4, p. 97) are in conflict and
represent different interpretations of the groundwater flow regime (Wheater
and Peach, 2019b, at CAP, Vol. 1, p. 108, Figure 8).
All of these listed issues affect the representation of groundwater/surface water
interaction in the Near Field Model and in tum affect the estimation of the impact
of the channelization on surface and groundwater flows .
This list is disturbing and leads to the conclusion that the modelling which has been
used to support and justify the DHI estimates of the impact of channelization on the
surface and groundwater flows from the Bolivian wetlands at the headwaters of the
Silala River is highly flawed. In short, the Near Field Model models developed by
DHI as Bolivia's expert advisors are based on an incorrect understanding of the
geology and hydro geology of the Silala River surface and groundwater catchments.
4.3.4 Wetland degradation
Studies by Bolivia's experts, including 1two reports by FUNDECO (BR, Vol. 3,
Annexes 23 .3 and 23.4), considered the effect of historical channelization on
23
observed changes in the wetlands. Bolivia's studies shed light on some of the
changes that have taken place in the wetlands, but they are flawed in several
important respects, as discussed by us in \Vheater and Peach (2019b) (CAP, Vol. 1,
pp. 137-140), and by Bolivia's own consultant (DHI, 2018b) in Bolivia's Rejoinder
(BR, Vol. 2, pp. 65-122).
Bolivia repeats a very serious error in the reporting of current wetland areas ( as
0.6 ha), which DHI note is flawed.14 Bolivia also asserts that large reductions in
wetland area are solely due to the historical channelization.15 However, this is based
on their FUND ECO (2018) report, in which, using geochemical evidence, they state
that wetland desiccation ' ... began around 1908, which is a clear sign of the effects
that canalization had on the Silala springs ' (BR, Vol. 3, p. 142), and note, from
pollen analysis, 'From 1908 onwards, a gradual desiccation process took place'
(BR, Vol. 3, p. 142). It follows that, on Bolivia's evidence, this desiccation predates
the construction of the channelization (installed in 1928) by some 20 years. They
also note that 'this desiccation process reached its climax around 1950 ... ' (BR,
Vol. 3, p. 142), which to our knowledge does not coincide with any channel
changes. Other strands of their evidence, from soil analysis, indicated major
changes between 680 and 862 years ago, and between 1960 and 1980 (BR, Vol. 3,
p. 155). Given that the dates ofreported changes bear no relationship to the date of
channelization, it must be concluded that other factors are playing a significant role.
We agree with Bolivia's experts, DHI, that climate changes could have been the
cause of some of these changes (BR, Vol. 2, p. 99).
In Wheater and Peach (2019a) (CR, Vol. 1, pp. 127-138), we reported on detailed
monitoring of an undisturbed Chilean wetland within the Silala River basin,
coupled with high resolution remote sensing data of the Bolivian wetlands, to
investigate whether there was evidence of degradation of the Bolivian wetlands.
Bolivia had asserted that 'the artificial channels and drainage network of the Silala
River substantially affected and degraded the bofedales and caused the wetlands to
recede and decline' (BCM, Vol. 1, p. 102). Our results showed that both the
Bolivian and Chilean wetlands continue to fully occupy the valley floor, and
seasonally extend up the base of adjacent hillslopes (CR, Vol. 1, pp. 132-136). It
therefore appears that channelization in Bolivia has not affected the area of active
wetland in the valley floors, where the drainage channels are located.
The hydrometeorological functioning of the wetland vegetation, as indicated by
remote sensing, is similar in all three wetlands, and associated estimates of actual
evaporation suggest that the highest evaporation rates are observed from Bolivia's
Cajones and Orientales wetlands, some 10% greater than that of the undisturbed
14 'It seems that the areas in the Ramsar report aire not reflecting the full wetland' (BR, Vol. 5, p.
41).
15 'The scientific evidence shows that the hydraulic works generated the fragmentation of the
bofedals ' (BR, Vol. 1, p. 50).
24
Quebrada Negra wetland (CR, Vol. 1, p. 137). This indicates that, with respect to
evaporation, the Bolivian wetlands are functioning at least as well as the
undisturbed Chilean wetland. Thus, from the satellite data, it appears that there has
been no significant reduction in evaporation associated with the channelization of
the Bolivian wetlands, and the small reductions in water table elevations associated
with the drainage of the Bolivian wetlands have not inhibited evaporation from the
wetland vegetation.
The important conclusion, that no effects of channelization on wetland evaporation
have been detectable, has significant implications. As discussed above, changes to
wetland evaporation losses are seen by both ourselves and DHI to be the primary
cause of potential changes in the total cross border flow of water from Bolivia to
Chile, as both surface water and groundwater flow from Bolivia into Chile (BR,
Vol. 5, p . 30; CR, Vol. 1, p. 108).
We note in passing that Bolivia's confusion concerning the wetlands extends to
their criticism of our analysis of high-resolution remote sensing data of wetland
extent (BR, Vol. 1, p. 46). Our analysis showed strong seasonal variability in the
spatial extent of active wetland vegetation, which according to Bolivia 'reveals
flawed calculation that cannot be reasonably accepted' . Bolivia thus ignores the
evidence of its own experts, Torrez Soria et al. (2017) (BCM, Vol. 3, p. 73) and
Castel (2017), who also confirm a large expansion and contraction of areas of active
wetland vegetation as the seasons progress.
4.3.5 Could the flow from groundwater fed springs in the Cajones and
Orientales springs have been significantly enhanced by the use of
explosives?
It was suggested (BCM, Vol. 1, p. 47) that the groundwater-fed springs of the
Cajones and Orientales wetland had been enhanced by explosives. However, as we
discussed in Peach and Wheater (2019), the evidence for this is very flimsy and a
reference cited by Bolivia concerning development of deep borehole yields from
very low permeability rocks by explosive methods is inapplicable. The BCM cites
Driscoll (1978) (BCM, Vol. 1, p. 47) as evidence that blasting can enhance water
flows by a factor of 6 to 20. However, this article is in no way applicable to Bolivia's
situation. Firstly, it concerns the development of deep ( over 100 metres depth)
borehole water supplies, not springs. Secondly the boreholes were located in poorly
fractured granites, quartzites and slates. These rocks are metamorphic, have
undergone considerable changes due to very high temperatures and pressures, and
hence are normally very poorly permeable, unlike the permeable rocks feeding
Bolivia's springs. Thirdly, the deep boreholes were plugged with sand to direct the
blast horizontally, which is clearly inapplicable to Bolivia's situation. Bolivia's
springs could not have been developed significantly to increase yields by the
explosive methods they suggest (CR, Vol. 1, pp. 217-218).
25
4.4 Summary discussion
In short, the basic reasons for the disagreement between Bolivia's experts, DHI,
and ourselves concerning the impacts of historical channelization concern the poor
use of well-established modelling software by DHI and inaccurate understanding
of geology. Our initial concerns with respect to the DHI modelling associated with
errors in boundary conditions were acknowledged by DHI, but when DHI's digital
data were made available, further very serious errors and unexplained assumptions
became apparent. In our opinion, the effects of channelization are primarily due to
changes in wetland evaporation. However, DHI and we agree that these will, at
most, account for a very small(< 2%) increase in the river's surface water flow,
and our remote sensing analysis shows no material difference in wetland
evaporation when an undisturbed wetland in Chile is compared to the channelized
wetland in Bolivia.
DHI has proposed additional effects, arising from changes in groundwater elevation
gradient due to channelization and increased hydraulic resistance to groundwater
flow associated with hypothetical peat accumulation. While we accept that these
effects are feasible, our own analysis showed that these effects were very small ( a
result that has subsequently been confimled when the DHI models were run by us
with errors partially corrected).
We noted in our Updated Analysis (Wheater and Peach, 2019b at CAP, Vol. 1,
p. 142) that DHI refer to a historical estimate of flow, made in 1922 prior to the
channelization, to support their simulations and conclusions. However, in our
opinion, a single estimate, made at a location that is uncertain, and in a difficult
environment where contemporary measurements have had large errors, cannot be
considered reliable. DHI (BR, Vol. 5, p. 56) reported that the single historical flow
measurement was 18% lower than current flows (at a location that they assumed),
but noted (BCM, Vol. 2, p. 392) that even under nearly ideal flow gauging
conditions, based on a specially-constructed flume, errors in flow rate measurement
in the Silala River can be expected to be of the order of 25-30%.
Bolivia's claims of wetland shrinkage and degradation are not supported by our
remote sensing and ground-based data, and are disputed by their own consultants,
DHI. Similarly, Bolivia's assertion that wetland degradation has been proved to be
due to channelization have been shown to be incorrect by ourselves and by DHI.
And Bolivia's claims concerning the use of explosives are implausible and
unsupported by any reliable evidence.
5 CONCLUSIONS
It has been encouraging to note that there is general agreement between Bolivia's
consultants, DHI, and ourselves concerning the hydrological functioning of the
26
Silala River basin. The Silala River flows from Bolivia to Chile and is a system of
surface waters and groundwaters constituting a unitary whole and flowing as both
surface water and groundwater across the international border. It is therefore
unequivocally an international watercourse.
Important differences remain concerning the interpretation of the hydrogeology, but
of most significance for the case is the continued assertion by Bolivia, based on
advice from its consultants, DHI, that there are large effects on the surface water
river flows associated with the historical channelization of the Bolivian wetlands,
when in our opinion, these are very small. In our reports accompanying Chile's
Reply, we showed that there were errors in the geology used in DHI's modelling
and a fundamental flaw in the treatment of the associated model boundary
conditions. In our Updated Analysis attached to Chile's Additional Pleading, we
further demonstrated that much of Bolivia's geological interpretation was wrong.
In addition, with access to the digital data used for the modelling, we showed
conclusively that DHI's modelling was fatally flawed. As explained above, the
large, simulated effects were in large part artefacts of DHI modelling errors.
In our opinion, the potential effect of the channelization is mainly a reduction in
evaporation from the wetlands. Such a reduction could increase the water available
for surface flow across the border. However, both DHI and we agree that such an
effect would be very small, at maximum 2% of the annual flow. In fact, our remote
sensing analysis of Bolivia's wetlands suggests that their lateral extent and seasonal
dynamics have not been significantly affocted by the channelization, nor has their
evaporation been significantly reduced.
Bolivia's consultants, DHI, invoke additional mechanisms to explain their large,
modelled effects, represented in their scenarios that compare the 'Baseline'
(channelized) situation with a 'No Canals:' and a 'Restored Wetland' scenario. Our
simplified calculation, while approximate, suggested that these mechanisms might
generate a 1 % change in surface flows. Additionally, it is important to note that any
increase in surface flows will be accompanied by a corresponding decrease in
groundwater flows across the border and vice versa. Any net change in water flow
across the border will primarily be due 1to change in wetland evaporation, which
DHI agree is small, and our remote sensing analysis suggests is negligible.
In conclusion, Bolivia's modelling of the impacts of channelization has been shown
to be flawed in many respects. It is wholly unreliable and should not be relied on
by the Court. We have consistently stated our expert opinion that these impacts will
be small. Indeed, the effects of the historic channelization on flows across the
border from Bolivia to Chile are so small that they are unlikely to be detectable.
27
6 REFERENCES
Arcadis, 2017. Detailed Hydrogeological Study of the Sil ala River. (CM, Vol. 4,
Annex 11).
Alcayaga, H., 2017. Characterization of the Drainage Patterns and River Network
of the Silala River and Preliminary Assessment of Vegetation Dynamics Using
Remote Sensing. (CM, Vol. 4, Annex I).
British Geological Survey, 1996. British Regional Geology, London and the
Thames Valley, 4th ed., 173 pp.
Castel, A.P., 2017. Analisis Multitemporal mediante imagenes de satelite de las
Bode/ales de las Manantiales del Silala, Potosi -Bolivia. DIREMAR. La Paz. (CR,
Vol. 2, Annex 98).
Danish Hydraulic Institute (DHI), 2018a. Study of the Flows in the Silala Wetlands
and Springs System. (BCM, Annex 17).
Danish Hydraulic Institute (DHI), 2018b. Technical Analysis and Independent
Validation Opinion of Supplementary Technical Studies Concerning the Silala
Springs. (BR, Vol. 2, Annex 23).
Danish Hydraulic Institute (DHI), 2:019. Updating of the Mathematical
Hydrological Model Scenarios of the Silala Spring Waters with: Sensitivity
Analysis of the Model Boundaries. (BR, Vol. 5, Annex 25).
Driscoll, F.G., 1978. Blasting- it turns dry holes into wet ones, Johnson Drillers'
Jnl, Nov/Dec, Johnson Division UOP, Inc., St. Paul, MN, p. 3.
Fundaci6n para el Desarrollo de la ecologia (FUNDECO), 2018. Study of
Evaluation of Environmental Impacts in the Silala, Palynology. (BR, Vol. 3,
Annex 23.4).
Herrera, C. and Aravena, R., 2017. Chemical and Isotopic Characterization of
Surface Water and Groundwater of the Silala Transboundary Basin, Second
Region, Chile. (CM, Vol. 4, Annex 111).
Herrera, C. and Aravena, R., 2019. Chemical and Isotopic Characterization of
Surface Water and Groundwater of the Si/ala River Transboundary Basin, Second
Region, Chile. (CR, Vol. 3, Annex XI).
Latorre, C. and Frugone, M., 2017. Holocene Sedimentary History of the Rio Silala
(Antofagasta Region, Chile). (CM, Vol. 5, Annex IV).
Mao, L., 2017. Fluvial Geomorphology of the Silala River, Second Region, Chile.
(CM, Vol. 5, Annex V).
28
McRostie, 2017. Archaeological First Baseline Study for the Silala River, Chile.
(CM, Vol. 5, Annex VI).
Mufioz, J.F., Suarez, F., Fernandez, B., Maass, T., 2017. Hydrology of the Silala
River Basin. (CM, Vol. 5, Annex VII).
Mufioz, J.F., Suarez, F., Sanzana, P. and Taylor, A. , 2019. Assessment ~f the Silala
River Basin Hydrological Models Developed by DHI. (CAP, Vol. 2, Annex XV).
Peach, D.W. and Wheater, H.S., 2017. The Evolution of the Silala River, Catchment
and Ravine. (CM, Vol. 1).
Peach, D.W. and Wheater, H.S. , 2019. Concerning the Geology, Hydrogeology and
Hydrochemistry of the Silala River Basin. (CR, Vol. 1).
Ramsar Convention Secretariat, 2018. Report Ramsar Advisory Mission N° 84,
Ramsar Site Los Lipez, Bolivia. (BCM, Vol. 5, Annex 18).
SERNAGEOMIN (Chile), 2017. Geology of the Silala River Basin. (CM, Vol. 5,
Annex VIII).
SERNAGEOMIN (Chile), 2019a. Geology of the Silala River Basin: An Updated
Interpretation. (CR, Vol. 3, Annex XIV).
SERNAGEOMIN (Chile), 2019b. A Review of the Geology Presented in Annexes
of the Rejoinder of the Plurinational State of Bolivia. (CAP, Vol. 2, Annex XVI).
Torrez Soria et al. , 2017. Characterization of the Soils of the Silala Bofedals and
its Vicinities. (BCM, Vol. 3, Annex 17, Annex D, Appendix Al).
Wheater, H.S. and Peach, D.W., 2017. The Silala River Today-Functioning of the
Fluvial System. (CM, Vol. 1).
Wheater, H.S. and Peach, D.W., 2019a. Impacts of Channelization of the Silala
River in Bolivia on the Hydrology of the Silala River Basin. (CR, Vol. 1).
Wheater, H.S. and Peach, D.W., 2019b. Impacts of Channelization of the Silala
River System in Bolivia on the Hydrology of the Silala River Basin - an Updated
Analysis. (CAP, Vol. 1).
29
Statement of IndeJJiendence and Truth
1. The opinions I have expressed in my Reports and Written Statement represent my
true and independent professional opinion. Where I have relied on the observational and
monitoring studies under my supervision by the Chilean scientific experts, or data
supplied to me by the Republic of Chile, I have noted that in my Reports and Written
Statement.
2. I understand that my overriding duty is to the Court, both in preparing the Expert
Reports that accompany the written presentations of the Republic of Chile, this Written
Statement and in giving oral evidence, if required to give such evidence. I have complied
and will continue to comply with that duty.
3. I have done my best, in preparing the Written Statement, to be accurate and
complete in answering the request of the International Court, as expressed in a letter of 15
October 2021 from the Registrar of the Court to the Agent of the Republic of Chile. I
consider that all the matters on which I have expressed an opinion are within my field of
expertise.
4. In preparing my Reports and Written Statement, I am not aware of any conflict of
interest actual or potential which might impact upon my ability to provide an independent
expert opinion.
5. I confirm that I have not entered into any arrangement where the amount or
payment of my fees is in any way dependent on the outcome of this proceeding.
6. In respect of facts referred to which are not within my personal knowledge, I have
indicated the source of such information.
7. I have not, without forming an independent view, included anything which has
been suggested to me by others, including the technical team and those instructing me.
Dr. Howard Wheater
Hydrological Engineer
10 January 2022
30
Statement of Inde1pendence and Truth
1. The opinions I have expressed in my Reports and Written Statement represent my
true and independent professional opinion. Where I have relied on the observational and
monitoring studies under my supervision by the Chilean scientific experts, or data supplied
to me by the Republic of Chile, I have noted that in my Reports and Written Statement.
2. I understand that my overriding duty is to the Court, both in preparing the Expert
Reports that accompany the written presentations of the Republic of Chile, this Written
Statement and in giving oral evidence, if required to give such evidence. I have complied
and will continue to comply with that duty.
3. I have done my best, in preparing the Written Statement, to be accurate and
complete in answering the request of the International Court, as expressed in a letter of 15
October 2021 from the Registrar of the Court to the Agent of the Republic of Chile. I
consider that all the matters on which I have expressed an opinion are within my field of
expertise.
4. In preparing my Reports and Written Statement, I am not aware of any conflict of
interest actual or potential which might impact upon my ability to provide an independent
expert opinion.
5. I confirm that I have not entered into any arrangement where the amount or
payment of my fees is in any way dependent on the outcome of this proceeding.
6. In respect of facts referred to which are not within my personal knowledge, I have
indicated the source of such information.
7. I have not, without forming an independent view, included anything which has been
suggested to me by others, including the technical team and those instructing me.
Dr. Denis Peach
Hydrogeologist
10 January 2022
31