i
W O R L D B A N K S T U D Y
Assessment of the Impacts of Climate
Change on Mountain Hydrology
Development of a Methodology through a Case Study in Peru
September 28, 2010
Authors
Walter Vergara
Alejandro Deeb
Irene Leino
Megan Hansen
Energy (LCSEG)
Sustainable Development Department
Latin America and the Caribbean Region
The World Bank
Pub
lic D
iscl
osur
e A
utho
rized
Pub
lic D
iscl
osur
e A
utho
rized
Pub
lic D
iscl
osur
e A
utho
rized
Pub
lic D
iscl
osur
e A
utho
rized
Pub
lic D
iscl
osur
e A
utho
rized
Pub
lic D
iscl
osur
e A
utho
rized
Pub
lic D
iscl
osur
e A
utho
rized
Pub
lic D
iscl
osur
e A
utho
rized
ii
Acknowledgements
The task was undertaken with the collaboration of several institutions with considerable
experience and skills relevant to the proposed analysis. These include the Meteorological
Research Institute of Japan (team led by A. Kitoh), the Stockholm Environmental Institute (team
led by D. Purkey and M. Escobar), the Institut de Recherche pour le Développement of France
(team led by B. Francou), the National Center for Atmospheric Research (L. Buja), the PNNL
(team led by S. Ghan), and SENAMHI of Peru (team led by E. Silvestre, and including J.
Ordóñez, C. Oria, O. Felipe, W. Lavado, W. Suárez, V. Rodríguez and A. Llacza).
The authors would like to express their gratitude for the support and inputs provided by P.Benoit,
S. Bogach, V. Alavian, M. Kerf, A. Valencia and E. Crousillat.
iii
Acronyms and Abbreviations
AMIP Atmospheric Model Intercomparison Project
AOGCM Atmosphere-ocean General Circulation Model
CAM3 Community Atmospheric Model
CCSM Community Climate System Model
CCSR Center for Climate System Research (of the University of Tokyo)
CDD Consecutive Dry Days
CLM Community Land Model
COES Comité de Operación Económica del Sistema Interconectado Nacional
GCM General Circulation Model
GHG Greenhouse Gases
GIS Geographic Information System
IPCC Intergovernmental Panel on Climate Change
IRD Institut de Recherche pour le Développement
JMA Japan Meteorological Agency
MINEM Ministry of Energy and Mines (of Peru)
MRI Meteorological Research Institute (of Japan)
NCAR United States National Center for Atmospheric Research
NIES National Institute for Environmental Sciences (of Japan)
PNNL Pacific Northwest National Laboratory
PRAA Regional Adaptation to Glacier Retreat Project
RMC Regional Circulation Model
RMSE Root Mean Square Error
RX5D Maximum 5-day Precipitation total in millimeters
SDII Simple Daily Intensity Index
SEI Stockholm Environmental Institute
SENAMHI Servicio Nacional de Meteorología e Hidrología del Perú
SRES Special Report on Emission Scenarios
UNFCCC United Nations Framework Convention on Climate Change
WEAP Water Evaluation and Planning Tool
iv
Contents
ACKNOWLEDGEMENTS ........................................................................................................................................... ii ACRONYMS AND ABBREVIATIONS ......................................................................................................................... iii EXECUTIVE SUMMARY ........................................................................................................................................... ix 1. INTRODUCTION ..................................................................................................................................................... 1
1.1. OBJECTIVE OF THE STUDY .......................................................................................................................... 1
1.2. METHODOLOGY ........................................................................................................................................ 1 1.2.1. CLIMATE ANALYSIS ............................................................................................................................ 1 1.2.2. HYDROLOGY ANALYSIS ...................................................................................................................... 2
1.3. STRUCTURE OF THE REPORT ................................................................................................................................. 2
1.4. RIVER BASINS USED AS CASE STUDIES ......................................................................................................... 3
1.4.1. SANTA RIVER ................................................................................................................................................ 3 1.4.2. MANTARO RIVER .......................................................................................................................................... 4 1.4.3. RIMAC RIVER ................................................................................................................................................ 4 2. CONTEXT ................................................................................................................................................................ 6 2.1. CLIMATE IMPACTS ON WATER REGULATION ........................................................................................................ 8 2.2. CLIMATE IMPACTS ON GLACIERS .......................................................................................................................... 8 2.3. CLIMATE IMPACTS ON MOUNTAIN WETLANDS .................................................................................................... 9 3. CLIMATE ANALYSIS .............................................................................................................................................. 11 3.1. HIGH RESOLUTION CLIMATE PROJECTION OVER PERU FOR 21
ST CENTURY ................................................ ……..12
3.1.1. RESULTS FOR AVERAGE PRECIPITATION .................................................................................................... 12 3.1.2. EXTREMES PRECIPITATION EVENT RESULTS .............................................................................................. 14 3.1.3. SURFACE HYDROLOGY RESULTS: SOIL WETNESS AND RUNOFF ................................................................. 16 3.2. ENSEMBLES TO SIMULATE FUTURE CLIMATE AT A BASIN LEVEL ........................................................................ 18 3.3. RAINFALL TREND ANALYSIS ................................................................................................................................. 21 3.4. LIMITATIONS IN CLIMATE PROJECTIONS............................................................................................................. 23 4. DEVELOPING A TOOL FOR THE HYDROLOGY ANALYSIS ..................................................................................... 25 4.1. INTRODUCTION OF THE WATER EVALUATION AND PLANNING (WEAP) MODEL ............................................... 25 4.2. DEVELOPMENT OF A GLACIER MODULE ............................................................................................................. 25 4.3. THE PÁRAMO MODULE ....................................................................................................................................... 28 5. TESTING THE HYDROLOGY TOOL AT A BASIN LEVEL .......................................................................................... 29 5.1. CALIBRATION OF NONGLACIATED SUB-BASINS .................................................................................................. 29 5.2. CALIBRATION AND VALIDATION: THE SANTA RIVER BASIN ................................................................................ 31 5.3. CALIBRATION AND VALIDATION: RIMAC-MANTARO RIVER BASINS ................................................................... 35 5.3.1. PARAMETER SETTING ................................................................................................................................. 35 5.3.2. THE MANTARO RIVER BASIN ...................................................................................................................... 36 5.3.3. THE RIMAC RIVER BASIN ............................................................................................................................ 40 5.3.4. CONCLUSIONS OF THE CALIBRATION-VALIDATION IN THE RIMAC-MANTARO SYSTEM ............................ 42 6. RESULTS FROM THE HYDROLOGY ANALYSIS ...................................................................................................... 43 6.1. VISUALIZATION OF CLIMATE CHANGE IN THE SANTA RIVER BASIN .................................................................... 43 6.1.1. Simulation of future climate dynamics ...................................................................................................... 43 6.1.2. Expected hydrologic response to observed trends .................................................................................... 45 6.2. VISUALIZATION OF CLIMATE CHANGE IN THE RÍMAC AND MANTARO BASINS ................................................... 47 6.3. RESULTS ............................................................................................................................................................... 51 7. CONCLUSIONS ...................................................................................................................................................... 53 7.1. KEY TOOLS ............................................................................................................................................................ 53 7.2. RESULTS ............................................................................................................................................................... 54
v
7.3. LIMITATIONS ........................................................................................................................................................ 54 7.4. PRODUCTS ............................................................................................................................................................ 54 REFERENCES ......................................................................................................................................................... 56
vi
Annexes
ANNEX 1: IPCC – Emission Scenarios......................................................................................................58
ANNEX 2: Subgrid treatment and its advantages and weaknesses……………………………….………62
ANNEX 3: Verification of the ability of the simulated dataset to reproduce observed precipitation
behavior (draft) …………………………………………………………………………………………...68
ANNEX 4: Trend analysis………………………………………………………………………………...72
ANNEX 5: Technical report on glacier and high elevation wetlands model selection and
parameterization…………………………………………………………………………………………...74
ANNEX 6: An approach for modeling the hydrologic role of glaciers in WEAP.....................................107
ANNEX 7: Calibration of nonglaciated sub-basins…………………………………………….………..118
ANNEX 8: Final calibration-validation of the Santa River model............................................................129
ANNEX 9: Data on calibration-validation of the Mantaro-Rímac model.................................................144
vii
Figures
FIGURE 1.1: The process of developing the methodology to address the hydrologic response to climate
change at the basin level……………………………………………………………………………………3
FIGURE 1.2: Schematic of Santa River model………………………………………………………….....4
FIGURE 1.3: Schematic of the Rímac and Mantaro system ………………………………………………5
FIGURE 2.1: Anticipated temperature anomalies in the American Cordillera..…………………………...6
FIGURE 2.2: Reduction in surface area of Glacier Santa Isabel in Colombia………..………...………....8
FIGURE 2.3: Cumulative loss in length for selected glaciers in the Andes since 1870………...………....9
FIGURE 2.4: Observed pattern of temperature as it has changed at Paramo and other high altitudes during
the period 1950-2007 in the Valley of Rio Claro in the Northern Andes………...……………………….10
FIGURE 3.1: Annual mean rainfall over Peru for 1979-2003…………………...…………………..........13
FIGURE 3.2: Annual mean precipitation changes for the near future (2035-49) and end of the 21st century
for 60-km and 20-km resolution………………………….……………………….………………....……14
FIGURE 3.3: Changes in maximum 5-day precipitation for the near future and the end of century, for 60
km and 20 km……..……………………..……………………………….……………………………......15
FIGURE 3.4: Changes in maximum consecutive dry days for the near future and the end of century, for
60 km and 20 km……………………………………………………………………………………...…...16
FIGURE 3.5: Change in wetness index of top layer of soil for the near future and the end of century, for
60 km and 20 km…………...………………………………..…..…………………...……….……..…….17
FIGURE 3.6: Changes in river flows: current annual flows and change between the present and the end of
century…………………………….………………………………………………………….....................18
FIGURE 3.7: Potential changes in temperature and precipitation in the Santa River basin………………19
FIGURE 3.8: Precipitation and temperature time-series associated with two sets of conditions in the Santa
River basin…………………..…………………………………..……………………..……….………....20
FIGURE 3.9: Trend analysis: Spatial distribution of areas exhibiting similar trends………….…………21
FIGURE 3.10: Precipitation patterns in the Mantaro River basin, average for 1990–99 and 2030-39...…23
FIGURE 4.1: Illustrating sub-basins and elevation bands………………………………….......................27
FIGURE 5.1: Calibration and validation of model in the Corongo basin for two historical periods: 1967–
83 and 1984–99..…………………………………………………......……………………………...…….30
FIGURE 5.2: Observed and simulated streamflow in Llanganuco, Paron and La Balsa in the Santa River
basin………………………………………………………………………………………………...……..31
FIGURE 5.3: Comparison of the simulated and observed Q for 1970–99……………...…….…………..32
FIGURE 5.4: Comparison of streamflow at La Balsa with and without glaciers…………………………34
viii
FIGURE 5.5 Another comparison of streamflow at La Balsa with and without glaciers...….....................34
FIGURE 5.6: Observed and simulated monthly average streamflows for the Mantaro River basin during
the 1970–81 calibration period…..…….………………………………………………........…………….37
FIGURE 5.7: Observed and simulated flow rates for the 1966–96 reference period…...…...…...…..…...38
FIGURE 5.8: Correlation between observed and simulated flow rates for the reference period….……...39
FIGURE 5.9: Observed and simulated monthly average streamflows for the Rímac River basin during the
1970–81 calibration period …….………………………...…..…………….……………………………..40
FIGURE 5.10: Observed and simulated streamflows at the Chosica station in the Rímac River basin…..41
FIGURE 5.11: Observed and simulated streamflows at the Surco station in the Rímac River basin….…41
FIGURE 6.1: Simulated remaining glacier ice coverage in 2036 under the fast warming climate
projection………………………………………………………………………………………………….44
FIGURE 6.2: Flow through Cañón del Pato for historic conditions, marginally warmer, and fast warming
scenarios………………………………………………………...……………………………....................45
FIGURE 6.3: Observed and simulated discharges using trend analysis at selected sites on the Santa River,
1966–96…..……….……………………………………….………………………………………………46
FIGURE 6.4: Results of the trend analysis for mid-century in the Santa River……..……………...….....46
FIGURE 6.5: Comparison of average monthly discharges in the Santa River between observed and mid-
century values, based on trend analysis…………………………………………..………………….........47
FIGURE 6.6: Observed and simulated discharges using trend analysis at selected sites on the Rímac and
Mantaro Rivers, 1966–96………………….………………………………………………………………48
FIGURE 6.7: Results of the trend analysis for the mid-century in the Rímac and Mantaro Rivers…........49
FIGURE 6.8: Comparison of average monthly discharges in Mantaro River for observed, mid-century and
end-of-century values, based on trend analysis..…………………………………………….....................50
FIGURE 6.9: Comparison of average monthly discharges in the Rímac River for observed, mid-century
and end-of-century values, based on trend analysis……………………………….………..…………..…51
ix
Tables
TABLE 2.1: Number of stations in Peruvian Andes with significant (at 5% level) trends for seasonal
temperature indices during the 1960–2000 period…………….……………………………………………7
TABLE 3.2: Observed annual trends for precipitation in the Mantaro River basin………………………22
TABLE 4.1: Principal selection criteria of glacier modeling……………………………………………..26
TABLE 4.2: Statistics of glacier model results for the Artesón sub-basin in the Santa River basin..….…28
TABLE 5.1: Calibration and validation statistics for the Santa River…………………………...………..33
TABLE 5.2: Land use parameter values for the glacier module……………………………………...…..36
TABLE 5.3: Criteria for the calibration and validation periods for the Mantaro sub-basins………...…...40
TABLE 6.1: Simulated reductions in glaciated area between 2006 and 2036 under two climate
projections…………………………………………………………………………………………………43
x
Executive Summary
Climate change is projected to result in larger temperature increases (anomalies) at high
altitudes than in surrounding lowlands. In the Andes, this may lead to the accelerated retreat
of tropical glaciers, the drying of unique neo-tropical alpine wetland ecosystems locally known
as ―páramos” or ―bofedales”, as well as increased weather variability and weather extremes, all
of which will affect water regulation. These impacts in turn may affect ecosystem integrity and
the economics of power and water supply in the region. Peru is one of the countries that could be
affected by these changes, as it relies on its mountain basins for the provision of over 50 percent
of its power, and discharges from upper basins feed water supply and agricultural systems.
This report seeks to develop techniques to assess the impacts of climate change on
mountain hydrology through case studies on selected water basins in Peru. This is done by first
projecting climate conditions in Peru (the climate analysis) and then using these results in a more
detailed water planning model to estimate how the changes in climate will affect future
hydrology conditions (the hydrology analysis). To ensure that the projected hydrology conditions
reflect the situation in upper watersheds of the Andes, an existing water planning model is
modified to include modules that depict the dynamics of glaciers and mountain wetlands.
Throughout the study the methodology is applied to three basins, the Santa, Mantaro and
Rimac river basins, which were selected with the help of the Government of Peru. These three
basins were chosen based on their perceived vulnerability to climate change and their relevance
to the energy, water supply and agricultural sectors.
Climate analysis
The first step in the assessment of climate change’s impact on mountain hydrology
consisted of a climate projection for the Peruvian Andes. This was done with the help of
three different tools, 1) a single General Circulation Model (GCM), which runs in a super
computer called the Earth Simulator-2; 2) data from an ensemble of 16 GCMs used in the Fourth
Assessment Report of the Intergovernmental Panel on Climate Change (IPCC); and 3) a trend
analysis based on historical observational data at a basin level.
With the help of the Earth Simulator-2, the MRI-GCM (Meteorological Research Institute
General Circulation Model)1 was run for the entire country at a very high resolution (20 km),
resulting in a detailed picture of how Peru’s climate would evolve during the 21st century under
the A1B Greenhouse Gas emission trajectory.2 Data on future temperature, soil moisture,
1 The code for the model run by the Earth Simulator was developed jointly by the Center for Climate System
Research (CCSR) of the University of Tokyo and the Japanese National Institute for Environmental Sciences
(NIES). This particular version of the CCSR/NIES Atmosphere GCM has been used for several international
modeling efforts, including future projections for the IPCC Special Report on Emission Reductions (SRES) and the
Atmospheric Model Inter-comparison Project (AMIP). 2 The 16 GCM ensemble run was done based on emission trajectories A1B and B1. A1B is a middle of the road
trajectory that nevertheless must be seen as a very conservative scenario, given the current rate of global greenhouse
gas emissions. A summary description of the emission trajectories drawn by the IPCC can be found in Annex 1.
xi
precipitation, and runoffs was generated. To strengthen the robustness of the results, these
projections were complemented with ensemble simulations using the same GCM, but modeling
multiple climate scenarios over the same area at lower resolutions (60 km and above).
The projections depict a much warmer climate than presently exists. Most notably, much
higher temperature anomalies are projected in the Peruvian Andes than in surrounding areas,
confirming prior analysis done in cooperation with the University of Massachusetts (Bradley et.
al., 2006). The average temperature change reaches 200 percent of the global average toward the
end of the century. This level of change is likely to translate into considerable alterations of the
ecology and the functioning of mountain habitats.
The model results also show a considerable intensification of the water cycle, with more
intense rains and longer dry periods toward the end of the century but with discernible changes
already by 2020. In particular, the upper layer of soil is projected to become gradually drier.
This and other changes would result in considerable modifications in stream-flows, with
significant projected reductions in the northeastern part of the country and on the southern coast.
To complement the projections from the MRI-GCM, the study used the outputs from a set of 16
GCM runs from the Fourth Assessment Report of the IPCC. Based on these results, two local
climate change scenarios through year 2040 were designed for use at a basin-level: marginal
warming with a 15 percent increase in precipitation and a 0.5oC increase in temperature; and
faster warming with a 10 percent decrease in precipitation and a 2oC increase in temperature.
These scenarios represent a range of possible future rainfall patterns. Finally, complementing the
use of GCMs, a rainfall trend analysis was used to gain confidence on the likelihood of future
rainfall patterns.
Hydrology analysis
GCMs are one way to project changes in temperature and rainfall under various future climate
change scenarios. However, they are more skillful at predicting temperature than precipitation
and often fail to incorporate all aspects of the hydrologic cycle in a satisfactory manner.
Therefore to assess the hydrologic impacts of climate change with sufficient accuracy a
separate hydrological model is typically needed. Furthermore, due to the complex topography
of mountainous regions in general and in Peru, it is not clear that a generic hydrological model
can adequately represent the local conditions. Thus an existing hydrological model was modified
to incorporate glacier and páramo modules to better represent the mountainous hydrological
system.
This study chose the Water Evaluation Planning Tool (WEAP) hydrologic model due to its
flexibility and suitability for Andean countries. The addition of the glacier module strengthens
the WEAP model’s applicability to high mountain conditions in the northern Andean region by
enhancing its ability to model the dynamic behavior of glaciers and their contribution to the
water cycle. As a result of this modification, when the model is applied to glaciated basins, it can
reflect net changes in hydrology induced by the disappearance of glaciated areas. The second
module added allows WEAP to deal with the hydrologic behavior of páramos3, which are
mountain wetlands in the Andean Cordillera. Although the páramo module was developed as
3 Páramo is a Neotropical ecosystem. It is located in the high elevations, between the upper forest line and the
permanent snow line. The ecosystem consists of mostly glacier formed valleys and plains with a large variety of
lakes, peat bogs and wet grasslands intermingled with shrublands and forest patches.
xii
part of this study, it was not used in the case studies because there are no extensive páramo
landscapes present in the three chosen river basins in Peru. However, efforts are underway to
apply the technique in Colombia.
The WEAP model had to be calibrated and validated for the three selected river basins of this
study. This involved a calibration in non-glaciated basins in Peru to obtain parameters that could
be thereafter applied in the selected three glaciated basins. The initial calibration and validation
of non-glaciated basins gave satisfactory results. Also the calibration and validation of the model
in the three selected river basins gave reasonable statistics. In the case of Santa River, the model
captured well conditions during low-flow period, but more fine-tuning is needed for high-flow
periods. It also represented larger glaciers better than small ones. All in all, it gave a reasonable
representation of the hydrologic behavior of the basin under current climate conditions. Since
Mantaro-Rimac River basin represents a bit more complex case, the application of the final
calibration-validation process was less precise. In general, the results show good correlations
from the calibration and validation, but there are some exception cases. Therefore there is a need
for further bias correction before the data of all the stations are used to guide decisions. After the
calibration-validation exercise, the tool was used to estimate the basin-level hydrologic response
to climate changes including variations in temperature, rainfall, and runoffs.
Results of the Case Study Analysis
While the purpose of this study is to produce a useful methodology to assess the impacts of
on mountain hydrology, the analysis nonetheless provides interesting results that can be used as
a basis for further analysis of climate impacts in the three analyzed basins and other similar areas
throughout the region.
For example, in the Santa basin, the analysis projects lower mean runoffs by mid-century,
including decreased year-round monthly runoffs at the station of La Balsa, the diversion point for
Cañón del Pato, an important power generation facility. The mean reduction is projected to be 21
percent by 2050-2059 compared with the present. Similarly, average flows at the Condocerro
station, in the lower part of the Santa River basin, are projected to decrease six percent. The
minimum flows at this station are projected to decrease by 18 percent.
The Mantaro-Rimac basins are more complex to simulate given the high level of man-made
infrastructure for water storage and run-off regulation, including dams and channels, though
significant results were obtained. The expected response to future climate conditions in the
Mantaro basin indicates a shift in the distribution of runoffs, as well as a reduction in peak flows.
Overall, discharges at key points in the basin seem to decrease. Conversely, at Rimac, projected
conditions indicate no significant changes and at the most suggest the possibility for a slight
reduction during the dry season. Most relevant for the objective of the study, the results of the
simulation of the glacier evolution in the Santa and Mantaro/Rímac River systems were
satisfactory, as the simulated evolution is consistent with historical records.
Conclusion
On the basis of the obtained results, it seems that the combination of the climate and
hydrology analysis can simulate current conditions at a regional and basin-level and
project future hydrologic conditions. The methods employed could be of use to predict future
impacts of climate change on hydrology for other mountain basins in the Andes.
1
CHAPTER 1
Introduction
1.1. OBJECTIVE OF THE STUDY
The objective of the study is to develop a methodology to assess the net impacts of climate
change on mountain hydrology. The development of this methodology would allow planners and
policy makers to better plan for adaptation measures to address the climate change impacts on
the power and water supply sectors.
1.2. METHODOLOGY
Since the objective of this study is to develop a methodology to assess the hydrological impacts
caused by climate change on mountain hydrology, the report seeks to find a solution for this
challenging task in the Andes. In order to assess climate change impacts on surface hydrology in
mountainous areas in Peru, a three-step process was followed. First, the study made projections
of future climate conditions for Peru at a national and a basin level by using GCMs as well as
conducting a local rainfall trend-analysis. In parallel, a suitable hydrologic model was chosen
and modified to better represent the mountainous hydrological system. Finally, these two pieces
(climate change analysis and hydrology analysis) were put together to study the hydrological
impacts of climate change in three pre-selected basins in Peru.
1.2.1. Climate analysis
The climate analysis focused on defining projected future climate conditions that could be used
as inputs to a river basin hydrologic model. Three complementary approaches were used for the
climate analysis.
GCMs are one option to project changes in temperature and rainfall under various future climate
change scenarios. GCMs are powerful tools for representing global processes. GCMs were used
in two ways throughout this study.
a) Use of high-resolution GCM and its river component. The Earth Simulator provided
data at a resolution of 20 km by 20 km for near future and end of the century. This model
also includes a river component which was used to assess country-wide river flow
changes even without a separate hydrologic tool.
b) Use of ensemble GCM outputs. Data from an ensemble of results from 16 GCMs was
available from the IPCC. This data was used to identify two future scenarios which were
representative of the range of anticipated changes in temperature and precipitation and
applicable to the local basins in this study.
2
In addition, a trend analysis was conducted. A trend analysis takes historical trends from the past
and projects them into the future. This was chosen as the complementary tool for GCMs because
it provides statistics by which to judge the adequacy of GCMs in simulating local conditions.
c) Trend analysis. Complementing the use of GCMs, a rainfall trend analysis was used to
gain confidence on the likelihood of the future climate conditions projected by the
GCMs.
1.2.2. Hydrology Analysis
GCMs are more skillful at predicting future temperature ranges than precipitation changes. As a
result, even if GCMs offer realistic simulations of the global and regional temperatures in most
cases they fail to accurately model the full hydrologic cycle at a basin level. Thus, additional
tools specialized in estimating hydrology patterns, a separate hydrology model, are required.
The hydrology analysis sought to translate the predicted climate conditions into changes in
runoffs. This was done through the use of a rainfall runoff model (WEAP) which was modified
to properly simulate the dynamic hydrologic behavior in the Andes and represent mountainous
conditions.4 The modification was made by complementing the WEAP model with glacier and
páramo modules. This is by itself an important contribution to help bridge an existing knowledge
gap in understanding the impacts of climate change on mountain hydrology.
However, even if a local region’s climate data is applied to a separate hydrological model, it still
may not represent the local conditions well enough. This is because current hydrologic models
are simplified, conceptual representations of a part of the hydrologic cycle and are not developed
for areas with complex morphology and topography like mountain river basins. Overall,
mountainous environments are particularly sensitive hydrological systems that are particularly
vulnerable to climatic variations. Therefore the assessment of the hydrological impacts of
climate change in the Andes is especially challenging. Local and basin-level projections are
required in addition to a separate mountain-adapted hydrological model.
1.3. STRUCTURE OF THE REPORT
This report presents the methodology used to assess impacts of climate change on mountain
hydrology in Peru, which is modeled in Figure 1.1, and some of the results obtained from
employing this methodology. Chapter 2 presents the method and results of the climate
projections (climate analysis). Chapter 3 presents the hydrologic model (WEAP) and describes
how it is modified to take into account the complex mountain hydrological system (hydrology
analysis). In Chapter 4 the results of the climate analysis are fed into the modified hydrologic
tool to assess future hydrology conditions in selected basins in Peru. Finally, Chapter 5 presents
the main conclusions of the study.
4 The WEAP model can evaluate the hydrologic feasibility of water management options related to the storage,
distribution, use and conservation of regional water supplies (Sieber et al. 2004; Yates et al. 2004). WEAP is a
microcomputer tool for integrated water resources planning. It provides a comprehensive, flexible and user-friendly
framework for policy analysis. WEAP is distinguished by its integrated approach to simulating water systems and by
its policy orientation.
3
Figure 1.1. The process of developing the methodology to address the hydrologic response
to climate change at a basin level
Climate analysis Hydrology analysis
Running a GCM in
the Earth Simulator-2
•high-resolution
climate projections
•Direct answers to
changes in river
flows
Ensembles of 16
Global Circulation
Model runs to assess
the range of change
at the selected
stations
Rainfall trend
analysis to
simulate basin-
level conditionsWater
Evaluation and
Planning Tool
Glacier
module
Hydrology response at a basin level•Santa River
•Rimac River
•Mantaro River
1.4. RIVER BASINS USED AS CASE STUDIES
Due to the challenges encountered with this type of modeling, throughout the report the
methodology described above is tested on three river basins in Peru: the Santa, Rímac and
Mantaro River basins. The basins were selected on the basis of their perceived vulnerability to
climate impacts and their relative economic relevance. They are vulnerable to climate change
given their relative altitude and the presence of glaciers. Thus, the impacts are relatively easy to
demonstrate. They also house large populations, provide water to urban centers, and are major
producers of agricultural products. The country’s main hydropower plants are located in these
river basins. These plants provided over 43 percent of the hydropower production in Peru in
2009 (COES 2009) and according to the Comité de Operación Económica del Sistema
Interconectado Nacional (COES) estimates, the future hydropower potential in these basins is
also significant. It is estimated that that these three river basins would total 42 percent of new
hydro capacity and 47 percent of added generation in the future.5 The three basins are briefly
described below.
1.4.1. Santa River
The Santa River basin has a total area of about 11,600 km2. The Santa river is fed by the glaciers
of the Cordillera Blanca which define the basin’s eastern boundary. The river flows north along a
central valley guarded by the cordilleras on both sides, known as the Callejón del Huaylas. The
5 These basins were also selected due to the different characteristics they included, including the important role of glaciers in the
Santa basin, Mantaro’s size and distinct regions, and Rímac’s typical dry and steep coastal basin. Due to their characteristics (and
the unique characteristics of each basin), applying and testing the models in these river basins provide useful experiences, but the
results cannot be generalized directly to other basins.
4
river basin is home to the Cañón del Pato hydropower plant, the second largest in the country.
On the coastal delta the Santa River feeds the Chavimochic irrigation district, which provides
water to the Chao, Virú, Moche and Chicama valleys. Nearly one million people live in the
basin. Figure 1.2 below offers a schematic representation of the Santa River basin and its
hydrologic representation in WEAP.
Figure 1.2. Schematic of Santa River model
1.4.2. Mantaro River
The Mantaro River basin covers an area of 34,550 km2 and it is divided in 23 sub-catchment
areas. The basin has great socioeconomic relevance in Peru. Located in the center of Peru, it
houses several important cities and it is the most densely populated basin in the sierra with over
700,000 inhabitants. It houses hydropower plants that supply over 34 percent of the energy
required by the national interconnected grid. It is also the food basket for Lima. Its agricultural
production not only feeds the major urban centers, but it is also a major exporter of
nontraditional products.
1.4.3. Rimac River
The Rímac River, located in western Peru, is part of the Pacific basin and has a length of 160 km.
The river begins in the highlands of Huarochirí Province in the Lima region at an elevation of
5,706 meters above sea level, fed initially by glaciated sub catchments and its mouth is located in
Callao. The Rímac basin is the most important source of potable water for the Lima and Callao
Metropolitan Area, serving a population of over seven million people. This basin’s very large
slope (over 3.5 percent) makes its hydraulic behavior mostly critical and supercritical, with great
capacity to transport sediments and large boulders. It also makes the basin attractive for
hydropower development. Precipitation in the Rímac basin ranges from values close to 800
5
mm/year in the high mountains to close to zero (less than 2 mm/year) on the Pacific coast. The
Rímac basin is also listed as a basin that is highly vulnerable to the impacts of climate change.
Because there is a major water transfer from the Mantaro River to the Rímac River basin, it was
decided after the analysis was started that the simulation runs should include both basins (the
Montaro-Rimac system).
Figure 1.3. Schematic of the Rímac and Mantaro system
6
C H A P T E R 2
Context
While potential climate change impacts have been studied quite extensively in Polar Regions,
there are comparatively fewer studies on the impacts in mountainous areas of the tropics. This
continues to be true even though recent research indicates that climate change may have large
impacts on high altitudes, which are likely to be greater than those on lowlands (Urrutia and
Vuille 2009). Reports based on projections made using the Earth Simulator (Vergara et al. 2007)
and ensemble results from various GCMs estimate that surface temperature in the Andes
mountains might increase as much as two times more than in the surrounding lowlands by the
end of the century. Field measurements in the central range of the Andes already indicate a
warming rate that exceeds the average registered for lowlands (Ruíz et al. 2010), and climate
change is projected to result in even more significant temperature variation for the Andes in the
future (Bradley et al 2006, see Figure 2.1). Besides the greater rate of overall warming
anticipated for the American Cordillera, which includes mountainous terrains such as the Andes,
the topography produces large climate variations along the mountain range. In particular, the
Andean Cordillera acts as a barrier that drives variations in precipitation along and surrounding
the range, which adds complexity to the analysis. These variations have important implications
for mountain hydrology.
Figure 2.1. Anticipated temperature anomalies in the American Cordillera
Source: Bradley et al, 2006
Note: Blue stars = Quito, Cusco and La Paz, cities that derive water from higher-elevation sources.
7
This chapter discusses some of the anticipated climate change impacts on glaciers and mountain
wetland ecosystems which might affect the runoff patterns of mountain water basins.
There is already clear evidence that the climate is changing in the Andes, particularly with
respect to changes in temperature and rainfall. Recent studies from Peru’s national
meteorological and hydrological institute (SENAMHI, 2007, 2009a and 2009b) have identified
new climate patterns in mountainous areas in Peru, including changes in minimum daily
temperatures, increases in maximum daily temperatures, reductions in relative humidity, changes
in precipitation patterns, and changes in expected total precipitation. For example, with respect to
temperatures, as summarized in Table 2.1, there are a decreasing number of cold nights and an
increasing trend for warmer nights. These trends are not equally distributed throughout the year.
Table 2.1. Number of stations in Peruvian Andes with significant (at 5% level) trends for
seasonal temperature indices during the 1960–2000 period
Visible impacts of the changes caused by these new climate patterns are already evident.
Warming temperatures have caused rapid retreat of glaciated areas (see Figure 2.2), and
variability and extremes in weather conditions have started affecting Andean ecosystems.
Warmer temperatures are affecting evaporation rates, water storage in natural and man-made
reservoirs, soil moisture and rates of evapo-transpiration of mountain vegetation. Perhaps more
importantly, change in weather conditions is expected to affect rainfall in the Andean.
These changes are expected to have repercussions on water regulation and water and power
supply, since rainfall is a major source of runoff that feed various power reservoirs, run-of-river
plants, urban water supply systems and agriculture. Also tropical glaciers, Andean lakes and
mountain wetlands contribute to the runoff. From 2006-2009, over 50 percent of the electric
power in Peru was produced by hydropower, and much of that is dependent on the mountain
water basins.
Specific impacts on rainfall, glaciers, mountain wetlands and water regulation are described in
the following sections.
8
2.1. CLIMATE IMPACTS ON WATER REGULATION
There is mounting information that climate is affecting surface hydrology. In reference to water
and water regulation the IPCC concludes:
There is high confidence that hydrological systems are being affected: increased runoff and
earlier spring peak discharge in many glacier- and snow-fed rivers, and warming of lakes and
rivers in many regions, with effects on thermal structure and water quality.
Some extreme weather events have changed in frequency and/or intensity over the last 50
years: It is likely that the frequency of heavy precipitation events (or proportion of total
rainfall from heavy falls) has increased over most areas.
Similarly, a wide array of satellite and field measurements document that climate is affecting
water stocks and flows in mountain systems. In practice this would most likely mean higher
fluctuations and loss of stream flows, which would have a direct impact on the available water
resources, power supply, and ecosystem integrity.
2.2. CLIMATE IMPACTS ON GLACIERS
In Peru, glaciers had an area of 2,041 km2 in 1970 but this number had declined nearly 22
percent to 1,595 km2 by 1997 (see Figure 2.3 and Bradley et al. 2006).
6 Major additional
reductions in surface area have been measured since. The largest of the studied glaciers in Peru’s
Cordillera Blanca lost 15 percent of its glacier surface area in 30 years. Figure 2.2 illustrates the
rapid decrease in surface area being measured for the Santa Isabel Glacier in the central range of
the Colombian Andes. The glacier is losing volume at a rate that would result in its
disappearance in a few decades7. Similarly, many of the smaller glaciers in Peru have been
heavily affected and some are likely to disappear within a generation (Francou et al. 2003).
Figure 2.2. Reduction in surface area of Glacier Santa Isabel in Colombia
IMAGEN DE SATELITE ALOS
(RGB:321)
Febrero 6 de 2009
0 1 Km
0 1 Km
SPOT 2001 ALOS 2009 Source: SPOT and ALOS images collected by the INAP project in Colombia and stored at IDEAM’s archive.
Note: Documented through satellite images (2001-09).
6 Glacier retreat is being monitored using a variety of techniques in the Andes. Some of these measurements are being facilitated
through the Colombia: Integrated National Project and the Regional Andes: Adaptation to Glacier Retreat project. 7 The second photograph is part of an archive being constructed through images taken every 48 days by the Advanced Land
Observation Satellite (ALOS) of Japan.
9
Anticipated and already observed climate change-related impacts caused by glacier retreat
include deterioration of river basins, depletion of water recharge capacities, and biotic changes in
ecosystem thresholds and composition, which affect the ecosystem’s ability to store water. The
effects and consequences may be different at the initial and final stages of glacier retreat and may
differ depending on location.
Figure 2.3. Cumulative loss in length for selected glaciers in the Andes since 1870
Source: IRD 2007
2.3. CLIMATE IMPACTS ON MOUNTAIN WETLANDS
High-mountain ecosystems, including páramos, are among the environments most sensitive to
climate change. These ecosystems have unique endemic flora and provide numerous and
valuable environmental goods and services. Recently published data from a study funded by the
World Bank (Ruíz et al. 2010) suggest that temperatures have indeed increased at a significant
rate at páramo altitudes (Figure 2.4). Also, according to the study, climate impacts have already
altered the circulation patterns of producing and moving water vapor within these ecosystems. It
is possible that these changes have contributed to the disappearance of high-altitude water
bodies, as well as to the increased occurrence of natural and man-induced mountain fires.
10
Figure 2.4. Observed pattern of temperature as it has changed at Paramo and other
altitudes during the period 1950-2007 in the Valley of Rio Claro in the Northern Andes
0
500
1000
1500
2000
2500
3000
3500
4000
4500
5 10 15 20 25 30 35
Maximum temperatures [°C] – warmest days
Alt
itu
de [
m]
LOWER LEVELS ON THE WESTERN FLANK
HIGHER LEVELS ON THE WESTERN FLANK
LOWER LEVELS ON THE EASTERN FLANK
X
+1.50
+0.60
+0.20
+0.30
+0.50
+0.50
G MTmax = - 5.9 K/km
Source: Ruiz et al, 2010.
Note: Paramo starts at around 3000 m at this location. Annual values (boxes, crosses and circles) and long-term
trends (arrows) observed in the spatial domain 04° 25’N-05° 15’N and 75° 00’W-76° 00’W. Grey Trends in
temperature are expressed in C/decade.
11
C H A P T E R 3
Climate Analysis
This chapter attempts to make a projection of the future climatic conditions in Peru, notably with
respect to precipitation levels through the use of three tools, with an emphasis on the
complementarities between these tools.
The first climate model tool presented in this chapter is a high-resolution climate
projection for Peru, generated by running the MRI (Meteorological Research Institute of
Japan) GCM in a super computer called the Earth Simulator-2. The projections resulting
from the model include data on rainfall, soil moisture and evaporation at a very high-
resolution (20-km), which is unique in global climate change studies and preferable since
it can capture the intensity and frequency of extreme weather events.8 Despite the use of
the super computer, the available computing power is insufficient to enable multiple
emission scenario runs. This limited the application of the high-resolution GCM to a
single run. To address this limitation and strengthen the robustness of the projections
made at 20-km resolution, ensemble simulations of the same GCM were made with the
Earth Simulator-2 at lower resolutions (60-km and lower). These results were compared
with the 20-km version.
Since GCMs used for projecting the changing climate are imprecise representations of the
Earth’s climate system, all of them are deemed to have ―model errors,‖ and the best
practice is to use the results of multiple GCMs in future climate projections. Therefore
the second tool presented is the use of combined output from multiple GCMs (an
ensemble projection). Under this approach, data outputs from 16 GCMs used in the
Fourth Assessment Report of the IPCC were used to project the potential range of
precipitation and temperature changes that might be anticipated at a basin level.
In order to further strengthen the robustness of the projections, the analysis also uses
trends from observed past meteorological data as a third tool to verify the projections of
local future projections.
8 The Fourth Assessment Report of the Intergovernmental Panel on Climate Change (AR4) uses a dataset of 24
global coupled atmosphere-ocean general circulation models (AOGCM, or GCM for short) to project future climate
under various scenarios. The use of numerous models is intended to reduce errors and uncertainty. However, most of
these models have a very coarse resolution (100–400km) and this has an undesirable impact on results, particularly
as it relates to extreme weather events. This is because global warming would result not only in changes in mean
climate conditions but also in increases in the amplitude and frequency of extreme events that would not be captured
in a meaningful way with coarse resolutions. Changes in extremes are more important for assessing adaptation
strategies to climate changes.
12
At the end of this chapter, the limitations of the various tools for modeling future climate are
discussed.
3.1. HIGH RESOLUTION CLIMATE PROJECTION FOR PERU FOR 21ST
CENTURY
The high resolution climate projection for Peru was generated primarily to assess temperature
and rainfall changes and extreme events. This projection used the atmospheric GCM of the
Meteorological Research Institute of the Japan Meteorological Agency (MRI), which is a high-
resolution atmospheric climate model9. The horizontal grid size is about 20 km (Mizuta et al.
2006), which offers unequaled high-resolution projection. The use of the Earth Simulator-210
,
which has a top speed of 130 teraflops (TFLOPS), made the high-resolution simulation possible.
A detailed description of the model and its performance in the 10-year present-day simulation
with sea surface temperature can be found in Mizuta et al. (2006).
The MRI-GCM was used to project Peru’s hydrologic response to climate change in the mid
(2035–49) and end of the 21st century (2075–99). The IPCC has designed standard reference
scenarios that are used in the GCM projections in its reports (SRES scenarios) and are widely
used in modeling studies elsewhere (see Annex 1 on the different IPCC-SRES emission
scenarios). The IPCC’s emission scenario used in this exercise was A1B,11
which is considered
a ―middle of the road‖ projection of greenhouse gas emissions, resulting in an average
temperature increase of between 1.3 and 3.5 degrees Celsius by the end of the century.
While the high resolution projection resulted in 95 outputs, only those that are of direct relevance
to the subsequent hydrology analysis are summarized here. These include annual mean rainfall,
extreme rainfall patterns (maximum 5-day precipitation and consecutive dry days), and top soil
layer moisture, which are all compared with historical data (1979–2003) to estimate changes
over time. These results are then used for an initial estimation of the changes in runoff and river
flows, prior to using the hydrology model.
3.1.1. Results for Average Precipitation
First, in order to develop an understanding of the soundness of the model in the region, historical
data was compared to historical simulations. The results of this exercise are presented in Figure
3.1, which summarizes the distribution of annual mean rainfall averaged for a 25-year period
(1979–2003). The vertical panels show the data at different resolutions, with the right panel
summarizing the results of the 20 km grid size simulation. The bottom row of plots in Figure 3.1
presents the actual observed mean annual precipitation data, while the top row summarizes the
results from a simulation analysis. In comparing these images, it is clear that the simulations
reproduce general observed patterns reasonably well. However, these images also reveal the
challenges of securing a good representation over the Andes especially at the lower resolutions
(60km and higher).
9 The atmospheric GCM is a global hydrostatic atmospheric general circulation model developed by MRI/JMA. This
model is an operational short-term numerical weather prediction model of JMA and is part of the next-generation
climate models for long-term climate simulation at MRI. 10
The data generated by the Earth Simulator was made available under the five-year Memorandum of
Understanding between MRI and the World Bank. 11
When initially developed, the A1B scenario was thought to be a description of the middle range of GHG emission
scenarios. Today’s emissions trajectory is already well above the A1B scenario. Therefore, this scenario may no
longer represent a plausible future.
13
Figure 3.1. Annual mean rainfall (mm d-1
) over Peru for 1979-2003
180 Km 120 Km 60 Km 20 Km
Note: Plots correspond to datasets of actual observations and different resolutions for projections.
Once the validity of the model was established for the region, the annual mean rainfall in the
near future (2035-49) and by the end of the 21st century (2075-99) were simulated to estimate
future changes in the annual mean rainfall. This data was then compared to the currently
available data on mean rainfall. The projected changes are presented in Figure 3.2. The overall
pattern of precipitation change simulated by the 20-km and 60-km models is similar. The largest
anomalies are seen over the Andes Cordillera while the smallest changes are seen in the eastern
lowlands and the southern coastal areas.
14
Figure 3.2. Annual mean precipitation changes (mm) for the a, b) near future (2035–49)
and c, d) end of the 21st century (2075–99) for 60-km and 20-km resolution
Note: Areas statistically significant at 95 percent level are hatched.
3.1.2. Extremes Precipitation Events Results
Global warming is not only expected to change the mean conditions but also to increase the
amplitude and frequency of extreme precipitation events. Understanding where these changes in
extremes will take place is important for recognizing where adaptation measures are needed.
Two distinct impacts can be identified: (i) changes in heavy precipitation (measured through the
maximum total in five days), and (ii) days without rain (measured through the number of
consecutive dry days). These two extreme indexes for precipitation were calculated to illustrate
changes in precipitation extremes over Peru. The change for heavy precipitation is represented
by the maximum 5-day total precipitation (in mm), termed RX5D. Figure 3.3 shows the
projected changes in the heavy precipitation in comparison to today, at both 60-km and 20-km
resolutions for the near future and end of the century. Throughout Peru, RX5D is projected to
increase, leading to rainfall intensification. The largest rainfall intensification is found over the
northwestern coast and the Andean Cordillera. At a higher resolution (20 km), the model projects
even greater increases in RX5D by the end of the century. This means that the country would
experience more concentrated and heavier downpours, increasing the likelihood of floods and
reducing the stability of stream-flows.
15
Figure 3.3. Changes in maximum 5-day precipitation (total in mm) for the a, b) near future
and c, d) end of the century, for 60 km and 20 km
Note: Hatched areas show consistency in ensemble results at 60-km resolution.
It is also possible that climate change will lead to an increased number of days without rain. To
illustrate the changes in dryness, the consecutive dry days (CDD) were estimated for the same
periods. A ―dry day‖ is defined as a day with precipitation of less than 1 mm d-1
. Figure 3.4
shows the changes in maximum number CDDs over Peru for the near future and end of the
century at 20 and 60 km. Major CDD change is projected over the entire country, with droughts
increasing over time.
16
Figure 3.4. Changes in maximum consecutive dry days for the a, b) near future and c, d)
end of the century, for 60 km and 20 km
Note: The scale is in days.
3.1.3. Surface Hydrology Results: Soil Wetness and Runoff
As a result of changes in precipitation, surface hydrology will be affected. A key parameter for
surface hydrology is the soil moisture content (wetness) of the upper layer of soil (WETSL1),
which is estimated on the basis of the modeling results. The calculations included an estimate of
the change in top soil moisture made on the basis of evaporation and changes in rainfall and dry
spells, the results of which are presented in Figure 3.5. The results indicate that soil will become
considerably drier in most parts of the country. The largest negative numbers, which indicate
changes in the water content in the soil, are shown in Peru’s Amazon region, as well as in the
southern coastal areas. The Andean Cordillera region is projected to experience mix changes in
soil moisture content. Dryness is projected to increase over time.
17
Figure 3.5. Change in wetness index of top layer of soil for the a, b) near future and c, d)
end of the century, for 60 km and 20 km
Note: The wetness index WSLN is defined as average change in moiture content of top 10 cm of surface soil.
Another statistic of interest for surface hydrology is runoff. Using the runoff data derived from
rainfall projections presented in Section 3.1.1 and 3.1.2 (Figures 2.1 - 2.5), the net flow of rivers
at a macro-basin level was calculated. The analysis used a ―GRiveT‖ river model.12
The results
are presented in the Figure 3.6. In the figure, the panel a) shows the current situation and panel b)
shows the expected change in the river flows. The end-of-century projection seems to indicate a
significant reduction in net discharges on the southern coast and in the northeastern part of the
country. The results also indicate that in the Andean region some areas will experience increased
runoff, while other areas will reduce the total volume of water in the hydrographic system. No
simple generalization is possible, nor it is desirable. Each area of interest needs studies and
analysis at resolutions adjusted to the size of the watershed and the details of the information
needed.
12 GRiveT: Global Discharge model using Total Runoff Integrating Pathways (TRIP), the 0.5 x 0.5 version with
global data for discharge channels (Nohara et al. 2006). The river runoff assessed in the land surface model is
horizontally interpolated as external input data into the TRIP grid so that the flow volume is saved. A similar
analysis made for the Magdalena river in Colombia has recently been published (Nakaegawa and Vergara, 2010).
18
Figure 3.6. Changes in river flows: a) current annual flows (mm) and b) change between
the present and the end of the century (%)
Note: The picture a) presents the absolute annual flow and therefore the scale is in mm. The picture b) presents the
change, and the scale is in percentage.
3.2. ENSEMBLES TO SIMULATE FUTURE CLIMATE AT A BASIN LEVEL
To complement the results obtained through the use of the high resolution projections of the
Earth Simulator-2, an ensemble output from 16 other GCMs with lower resolutions (grid sizes of
200 Km or more) was used to address uncertainty in the estimates potential future temperature
and precipitation changes. These 16 models were not run specifically for this study, but instead,
existing data was gathered from the IPCC Fourth Assessment Report.13
Data was used from two
scenarios, (i) the A1B scenario, used in the previous sub-chapter in the application of the high-
resolution GCM, and (ii) the B1 scenario, which is considered optimistic given the lower rate of
emissions and lower rate of warming.
The range of precipitation and temperature changes obtained from the IPCC data was used to
project changes at selected meteorological stations (the Collota, Huaraz, Paron and Caraz
stations) in the Santa River basin over the 21st century. The IPCC data was collected on a
monthly basis. For each month in the simulation period from 2000 to 2100, the mean
precipitation and temperature change, along with the mean plus/minus one standard deviation,
were calculated for each basin using all data points (16 models times two emission path
scenarios). These results estimate the upper and lower envelopes of precipitation and
13
Under the World Climate Research Programme (WCRP), the Working Group on Coupled Modelling (WGCM)
established the Coupled Model Intercomparison Project (CMIP) as a standard experimental protocol for studying the
output of coupled atmosphere-ocean general circulation models (AOGCMs). CMIP provides diagnosis, validation,
intercomparison, documentation and data access. Virtually the entire international climate modeling community has
participated in this project since its inception in 1995. The Program for Climate Model Diagnosis and
Intercomparison (PCMDI) archives much of the CMIP data and provides other support for CMIP.
http://cmip-pcmdi.llnl.gov/
19
temperature change around 2040, as shown in Figure 3.7. The output from the 16 GCMs was
generally consistent with the outputs from the Earth Simulator.
Figure 3.7. Potential changes in temperature and precipitation in the Santa River basin
(Collota, Huaraz, Paron and Caraz meteorological stations)
a) temperature
b) precipitation
Note: Outputs are from a set of 16 GCMs.
The information from the simulations was used to create two temperature and precipitation
scenarios for the Santa River basin for the decade of 2040 by estimating the range of two
extremes. The first consists of: marginally warmer conditions with a 0.5oC increase in
20
25
30
35
40
Sep-65 Sep-75 Sep-85 Sep-95 Sep-05 Sep-15 Sep-25 Sep-35
oC
Recuay1
Recuay1+0.5
+0.5
25
30
35
40
Sep-65 Sep-75 Sep-85 Sep-95 Sep-05 Sep-15 Sep-25 Sep-35
oC
Recuay1
Recuay1+2
+2
temperature and a 15 percent increase in precipitation (roughly corresponding to the outputs
from the Earth Simulator-2); and the second represents fast warming conditions with a 2oC
increase in temperature and a 10 percent decrease in precipitation.
These sets of conditions were used to project climate data for a 30-year period, 2010-2040,
building two climate paths for the immediate future, as shown in Figure 3.8.14
For the baseline
conditions (the average between 1979 and 2003), annual precipitation is around 870 mm/year.
Under fast warming conditions, precipitation is reduced to 780 mm/year by 2040 while under the
marginally warmer conditions, precipitation is projected at 1,000 mm/year. Precipitation is on
the left side while temperature in on the right side. This approach produces the possible range of
results to be expected from climate change. The results illustrate the uncertainty associated with
the use of the 16 independent GCMs deployed in the IPCC process.
Figure 3.8. Precipitation and temperature time-series associated with two sets of conditions
in the Santa River basin
Marginally warmer: +15% increase in precipitation, +0.5oC increase in temperature
0
50
100
150
200
250
300
350
Sep-65 Sep-75 Sep-85 Sep-95 Sep-05 Sep-15 Sep-25 Sep-35
mm
/mo
nth
Agua08
Agua08+15%
+15%
Faster warming: -10% increase in precipitation, +2oC increase in temperature
0
50
100
150
200
250
300
350
Sep-65 Sep-75 Sep-85 Sep-95 Sep-05 Sep-15 Sep-25 Sep-35
mm
/mo
nth
Agua08
Agua08-10
-10%
14
This approach has serious limitations, as described by Stone and Knutti (2010):
a) All climate projections are necessarily uncertain (Knutti et al. 2002). The largest contribution to this
uncertainty is the limited understanding of many of the interactions and feedbacks in climate change. The
ensemble hypothesis comes from the observation that combined information from many models performs
better than a single model (for example: Yun et al. 2002; Thomson et al. 2006).
b) The GCMs are interdependent: many models are based on the same theoretical and sometimes empirical
assumptions; all models have similar resolutions (too coarse to solve small-scale processes); model
development is not independent: models are frequently compared and successful concepts are copied.
21
3.3. RAINFALL TREND ANALYSIS
The projections from the Earth Simulator-2 high resolution model and ensembles from coarser
projections were also complemented with local rainfall trend analysis based on observed data.
The hydrologic trend analysis provides statistics by which to judge the adequacy of GCMs in
simulating local conditions.
This part of the analysis builds on the work conducted by SENAMHI during the preparation of
Peru’s Second National Communication to the United Nations Framework Convention on
Climate Change (UNFCCC). It looks at existing climate records from the past to identify trends
that are mutually consistent. The application of this analysis is summarized in Annex 4.
This analysis was completed for each of the three case study river basins. As an example of the
results produced, the Mantaro River basin is explained in this report. The map in Figure 3.9
shows the Mantaro River basin as well as the location of meteorological stations upon which the
analysis is based.
Figure 3.9. Trend analysis: Spatial distribution of areas exhibiting similar trends
The trend analysis was used to identify different precipitation trends within the climate. Table
3.2 shows the observed annual precipitation trends in the Mantaro River basin. A positive
precipitation trend was identified in the southern valley region (Region III). In contrast, for the
northern valley region (Region I) and the eastern region (Region VI), the trends indicate no, or
hardly any, changes in annual precipitation. The remaining identified regions exhibit a
decreasing trend, particularly significant in the high mountains surrounding the northern valley
where substantial reductions in precipitation are anticipated.
II
I
II
I
I
I
I
V
V
I
V
22
Table 3.2. Observed annual precipitation trends in the Mantaro River basin (mm/period)
Area Location Annual DJF MAM JJA SON
I
Northern
Valley 0 0 0 0 0
II North -11 -6 -2 -1 -2
III
Southern
Valley 7 3 1 1 2
IV Central -4 -2 -2 1 -1
V South -3 -1.5 -1.5 0 0
VI East -1 -1 0 0 0
These trends were used to make linear projections into the future. The precipitation patterns
resulting from this analysis are summarized in Figure 3.10. The figure facilitates the
identification of areas with similar behavior. Trend analysis only makes reference to the general
tendencies, that is, to changes in the mean value with time.15
The results generated are consistent
with those found when utilizing GCMs. That is, the magnitude of the trends fall within the range
previously identified from the CGM results.
15
This is a very serious limitation. Thus, the analysis presented needs to be complemented in order to include
hypotheses on the expected trends in the other relevant statistics. The results of the downscaling exercise described
in Annex 3 were used to assess the changes in variability. The analysis indicates that the standard deviation of future
scenarios exhibits increased variability. The increase in variability also remained very similar among the ten areas,
measured as percent increase over the variability of the 1990–1999 climate simulations. The project adopted these
increases as a reasonable basis to assess the expected trends in variability as global warming continues.
From a statistical standpoint, the key statistics have been analyzed and projection-type methodologies have been
used. But many uncertainties remain, among others those associated with the many, albeit reasonable, hypotheses
made. A linear trend for the mean is assumed; variability is thought to follow the variability increases exhibited in
GCM results; and temporal and spatial covariance are assumed to be time invariant. Although it is not possible to
assess the uncertainty associated with the climate scenarios developed, it is clear that they represent good
approximations for likely future climates. It is therefore suggested that the climate scenarios built through the
approach developed here be called ―climate planning scenarios,‖ indicating that they closely resemble climate
projections but retain enough uncertainty as to be called ―scenarios.‖ The use of these planning scenarios should be
accompanied by rough sensibility analyses to reduce the inherent uncertainties, and the need to search for water
resources solutions that are resilient over a range of plausible ―climate planning scenarios‖ to be used as inputs to
the hydrologic model.
23
Figure 3.10. Precipitation patterns in the Mantaro River basin (mm/year), average for a)
1990–99 and b) 2030–39
a) b)
3.4. LIMITATIONS IN CLIMATE PROJECTIONS
Although there are different ways to develop future climate scenarios, there is no consensus on
which ways are more applicable or useful in planning activities. Both GCM and trend analysis
have serious limitations.
While GCM models are powerful for representing global processes, they do not have the detailed
resolution required for water resources planning at the basin level. The output of GCMs should
be understood as the average climate condition found in the cell, which encompasses large areas
that are thought to be homogeneous. This hypothesis is not valid in mountainous terrain.
Interpretation of the models is therefore difficult. When using results from GCMs, the following
general limitations should be considered:
a. Model uncertainty. Global oceanographic and atmospheric dynamic circulation models
are representations, based on our scientific understanding, of very complex phenomena
that span many levels of resolution. By definition, models are simplifications based on
our present understanding. As more research becomes available and observations provide
new insights into the many physical and chemical processes in the ocean and the
atmosphere, models are expected to improve. The models used have considerable model
uncertainty, as indicated by the dissimilar estimates they offer of future climates.
b. Emissions paths. Future GHG emissions depend on economic activity and political
decisions, both subject to great fluctuations and uncertainty.
c. Model resolution. For hydrologic application in complex terrain, such as that in Peru,
data input to hydrologic models is needed at very detailed resolution. Such resolution is
only available in one model among the 16 used in the Forth Assessment Report, with the
24
additional drawback that it has a limited number of independent simulations of future
climate scenarios for sensitivity analysis. Most GCMs are run at resolutions of several
hundreds of kilometers (parcel units are several hundreds of kilometers by side), which
imply considerable theoretical and methodological difficulties.
d. Model independence. The use of model ensembles has become a common practice to
assess future climate results. The underlying hypothesis is that the average of
independent outcomes provides better mean estimates than single model experiments.
Recent research suggests that this hypothesis is difficult to prove because models are not
independent. They share similar routines; frequently one is the basis of improved
versions—although they remain on the list of available models—and sharing of
information and research among model developers prevents the use of the independent
hypothesis. Some authors have proposed weighing the likelihood of model results based
on a model’s agreement with observation. Despite the potential of this option, many
details must be worked out before this approach can become operational.
e. Inability to accommodate the complex atmospheric process in mountain terrains. A
well-known limitation of GCMs is their inability to capture the many local atmospheric
processes present in mountain landscapes. Several procedures have been developed to
cope with these limitations, but they lack scientific and observational footing.
Hydrologic trend analysis is a complementary approach that is based on observational data.
However, data availability is a limitation. It would be desirable to analyze good-quality weather
observation spanning 30 to 40 years, however, this amount of data is often not available. Even if
such information is available, the basic hypothesis is that the identified trends are linear and will
continue independent of the emission path. As a result, trend analysis can provide results by
which to judge the accuracy of GCMs in simulating local conditions, as done in this study. Thus,
the selective use of those GCMs that reproduce (simulate) the observed trends should have more
weight in the construction of future climate scenarios, as suggested in another recent World Bank
analytical study (Assessment of the Risk of Amazon Dieback, 2010-submitted).
25
C H A P T E R 4
Development of a Tool for the Hydrology
Analysis
Before the outputs from Peru’s climate projections could be fed into the hydrologic model to
gain a more detailed assessment of the impact of climate change on the hydrology in the
Northern Andes, the hydrologic model had to be adapted to handle the specific conditions of the
region. This chapter discusses the modifications made to the Water Evaluation and Planning
Model (WEAP); specifically, the addition of a glacier module and páramo module. The glacier
module allows the model to reflect the dynamic behavior of glaciers and estimate their net
contribution to runoff. The páramo module does the same with mountain wetlands.
After the initial creation of these modules, a calibration and verification analysis was needed to
assess the applicability of the modified WEAP model. The calibration was done using historic
climate data and observed runoffs at key points.
4.1. INTRODUCTION OF THE WATER EVALUATION AND PLANNING (WEAP) MODEL
The WEAP modeling software, developed by the Stockholm Environment Institute, was used as
a base for building the hydrology components. It is a generic, object-oriented water resources
modeling system that includes options to simulate both the natural rainfall runoff processes and
the management of installed water infrastructure. It was selected for this analysis based on a
comparison exercise that was done for a Bank-financed, project called the Regional Adaptation
to Glacier Retreat Project (PRAA). The comparison concluded that the water management
component of WEAP was an important addition to the model’s adaptability to different basin
configurations and to the flexibility in data requirements. The WEAP model was deemed suitable
for use in the Andean countries where data availability represents a major barrier. To enhance the
modeling system appropriateness to the Andean region, glacier and páramo modules were
developed and added to this model.
4.2. DEVELOPMENT OF A GLACIER MODULE
A comparative analysis of different modeling approaches to glacier behavior was conducted as a
starting point in the development of the glacier module. There are multiple approaches available
to represent tropical glaciers and high-elevation wetlands, including statistical models,
conceptual models, quasi-physical models and process-based models. There were a number of
factors considered in selecting the glacier dynamic representation included simplicity,
correspondence with published information the demand for data and data availability of the basin
under study. Annex 5 (Technical Report on glacier and high-elevation wetland model selection
and parameterization) presents in detail several options for glacier modeling, which are
summarized in Table 4.1 below.
26
As indicated in the table, data availability was ultimately a limiting factor for models, as only
one of them could model all three basins. As a result the simple degree-day model was selected
and its application was adjusted to each of several elevation bands describing the glaciers16
.
Table 4.1. Principal selection criteria of glacier modeling
The modeling of glacier dynamics follows the standard approach in WEAP, which involves
simulating rainfall-runoff processes after first dividing a basin into sub-basins. These basins are
each upstream from a ―pour point,‖ which is a point where stream-flow is measured or where the
river is actively managed. Following, the sub-basin area above a ―pour point‖ is divided into i
elevation bands. Each sub-basin/elevation band is defined as a unique WEAP catchment object,
which represents an area with similar hydrologic behavior. Within each catchment object
temporally variable land cover and climate conditions can be added on a time step by time step
basis. Each elevation band, i, is divided into either a glaciated (j=1) or nonglaciated (j=2)
portion. In Figure 4.1 below the blue portion represents the glacier, while the green area
represents the sub-watershed. Catchment object elevation bands are divided by the red lines.
16
Represents an area with similar hydrologic behavior.
27
Figure 4.1. Illustrating sub-basins and elevation bands
CATCHMENT Object in WEAP
Streamflow Measurement orManagement Point
i=1
j=1 j=1 j=1
j=2
j=2
j=2
j=2
i=m
i=n
Ei
EmEm+1
i=m+1
En
Ai, j=2Ai, j=1
RUNOFF in WEAP Elevation BandDelimiter
Subwatershed
Glacier CATCHMENT Object in WEAP
Streamflow Measurement orManagement Point
i=1
j=1 j=1 j=1
j=2
j=2
j=2
j=2
i=m
i=n
Ei
EmEm+1
i=m+1
En
Ai, j=2Ai, j=1
RUNOFF in WEAP Elevation BandDelimiter
Subwatershed
Glacier
Once the areas are split, runoff from the evolving glaciers and rainfall was assessed and added.
The calculations were made on two time scales: a monthly time step, t, and an annual time step,
T. The procedure to assess runoff is summarized as follows:
Step 0 – Initial Conditions: The initial step is to define the base hydrologic conditions within
each sub-basin. Satellite data was used to determine the initial allocation of glaciated versus non
glaciated land. This area within each catchment is represented by Ai and is defined in units of
km2. Based on a published empirical relationship, the glacier ice volume (V) is estimated in km
3
(Bahr et al. 1997).
Step 1 – Estimated Runoff from Melting Snow and Ice: The contribution to surface runoff from
the glaciated portion of the catchment area is extrapolated from estimates of melting glacial snow
and ice. This is done for each month, t, within a hydrologic year, T.
Step 2 – Surface Runoff at the Sub-basin Level: The volume of surface runoff within a sub-
basin is calculated for each monthly time step as the sum of the contribution of a) melting snow
and ice for the glaciated portion of the sub-basin and b) the runoff coming from the simulation of
rainfall-runoff processes in non-glaciated portions of the sub-basin.
Step 3 – Annual Mass Balance: A mass balance is used to assess changes in the area covered by
glaciers within a sub-basin at the end of the 12 monthly time steps of a hydrologic year.
Step 4 – Annual Glacier Geometry Evolution: The overall volume and extent of the glacial ice
within a sub-basin is calculated prior to moving on to the subsequent hydrologic year.
Step 5 – Calibration: The key criterion for calibration is the adjustment of parameters. This is
needed to obtain a good fit with the surface area of the glaciers and the measured stream-flow at
selected pour points where gauging stations are available.
The glacier module was tested using available data on the evolution of glaciated areas over time
(as estimated from satellite images), and the observed stream-flows at key gauging stations
28
below glaciers. The algorithm, which describes snow and ice accumulation and melt along with
glacier area and volume evolution, was implemented using the WEAP software’s user-defined
variable functionality. This process is described in the calibration-verification section of the next
chapter. Details about the glacier module are presented in Annex 6.
Initial calibration and verification of the glacier module were conducted for the Artesón glacier
in the Santa River basin. The Artesón sub-basin in the Santa River basin is a 6.2 km2 area with
80 percent glacier coverage. This site was selected because IRD had, over number of years
(2000–2007), collected extensive data on glacier evolution and glacier melt outflow for this sub-
basin. The calibration statistics summarized in Table 4.2 indicate a good fit and low discrepancy
between calculated and observed glacier extension (bias).17
Table 4.2. Statistics of glacier model results for the Artesón sub-basin in the Santa River
basin
Calibracion
Periodo n RMSE BIAS NASH
Arteson 2000-2007 79 45% 5% 10%
2001-2007 69 33% -4% 51%
Module parameterization for the Artesón system was used in the estimation of parameter values
in the wider Santa River basin. In a similar manner, work with the glacier module in the Santa
River system guided implementation of algorithms in the Mantaro/Rímac River system (see
Annex 6). The main outputs of the glacier routine include changes in glacier area runoffs.
4.3. THE PÁRAMO MODULE
Two alternatives were used in the development of a páramo module. The first represented the
hydrologic role of páramos by linking the behavior of a reservoir (the Paramo as a storage of
water, akin to a dam).
The second attempted to parameterize the existing rainfall-runoff routine in WEAP to capture the
unique nature of hydrologic processes in river basins dominated by páramos. The conclusion of
this exercise was that the first approach would require further refinement before it could be used
to represent páramo hydrology. Based on the analysis conducted as part of the second approach,
the use of the existing WEAP rainfall-runoff routines is likely sufficient to represent and model
páramo hydrology.
Although extensive páramo landscapes are not present in the three pilot river basins in Peru, and
therefore the module was not used in this study, attempts were made to calibrate existing rainfall-
runoff models in a basin dominated by páramos. Data from a small river basin near Quito in
Ecuador, with 90 percent of páramo landscape, were used. Both hydrologic approximations were
implemented and the details can be found in Annex 7. The results showed that the paramo
module sufficiently modeled the dynamic behavior of the paramo and its impacts on runoff.
17
Root Mean Square Error (RMSE) is a frequently-used measure of the differences between values predicted by a
model and the values actually observed. It is a good measure of precision or repeatability, which is the degree to
which repeated measurements under unchanged conditions show the same results. Bias: perceptual bias; and Nash-
Sutcliffe efficiency coefficient is used to assess the predictive power of hydrologic models. It is a criterion that
measures the fraction of the variance of observed values explained by the model. It ranges from minus infinity to
1.0. An efficiency of 1 corresponds to a perfect match of modeled discharge to the observed data. Therefore, large
positive values, close to 1.0, are sought.
29
C H A P T E R 5
Testing the Hydrology Tool at a Basin Level
Once the hydrologic model was adjusted to represent mountain systems typical of the Peruvian
Andes (Chapter4), the climate projections from the ensemble results and the trend analysis were
entered into the WEAP model to assess how the various river basins in Peru might be impacted
by climate change. Before applying the WEAP hydrologic model in the studied three basins, the
model first was calibrated and validated in sub-basins with no glacier coverage to obtain a
preliminary set of parameters to be applied in all glaciated sub-basins. After that, this chapter
discusses the results from the calibration and verification of the glacier module in the three sub-
basins under focus of this study.
Before applying the WEAP hydrologic model to the three selected basins, the glacier module
was calibrated and validated to ensure that it satisfactorily represents glacier dynamics. The
model was first calibrated in sub-basins that do not have glacier coverage to check the existing
rainfall-runoff routines. Based on the initial findings of the calibration in non-glaciated river
basins, a preliminary set of parameters was applied when re-calibrating the adjusted model in the
studied glaciated basins. The modeling period for calibration was 1970–1984, and 1985–1998 for
validation.
5.1. CALIBRATION IN NON-GLACIATED SUB-BASINS
The modified WEAP model was initially calibrated in the nonglaciated Corongo and Tablachaca
sub-basins. Corongo has an area of 561 km2. Discharge is recorded at the Manta gauging station.
Tablachaca has an area of 3,179 km2 and stream flow is measured at the Condorcerro gauging
station. The time-series of observed and simulated stream flows for Tablachaca and Corongo are
presented graphically in Figure 5.1.
Rainfall-runoff parameters related to subsurface conductivity, soil water capacity, runoff
resistance and flow direction were adjusted to obtain the peak flows in the winter and the base
flows in the summer that approximated the observed conditions. The calibration was rated
satisfactory.
30
Figure 5.1. Calibration and validation of model in the Corongo basin for two historical
periods: 1967–83 and 1984–99 Periodo Calibracion 1967-1983
SJN
KNG
KAW
TUL
KRN
Periodo Validacion 1984-1999
Caudal observado vs Caudal simulado en Corongo
05
1015202530354045505560
Jan-6
8
Jul-6
8
Jan-6
9
Jul-6
9
Jan-7
0
Jul-7
0
Jan-7
1
Jul-7
1
Jan-7
2
Jul-7
2
Jan-7
3
Jul-7
3
Jan-7
4
Jul-7
4
Jan-7
5
Jul-7
5
Jan-7
6
Jul-7
6
Jan-7
7
Jul-7
7
Jan-7
8
Jul-7
8
Jan-7
9
Jul-7
9
Jan-8
0
Jul-8
0
Jan-8
1
Jul-8
1
Jan-8
2
Jul-8
2
Jan-8
3
Jul-8
3
m3/
s
13 \ Q_Manta
14 \ Reach
Caudal observado vs Caudal simulado Tablachaca
0
20
40
60
80
100
120
140
160
180
200
Jan-
69
Jul-6
9
Jan-
70
Jul-7
0
Jan-
71
Jul-7
1
Jan-
72
Jul-7
2
Jan-
73
Jul-7
3
Jan-
74
Jul-7
4
Jan-
75
Jul-7
5
Jan-
76
Jul-7
6
Jan-
77
Jul-7
7
Jan-
78
Jul-7
8
Jan-
79
Jul-7
9
Jan-
80
Jul-8
0
Jan-
81
Jul-8
1
Jan-
82
Jul-8
2
Jan-
83
Jul-8
3
m3/
s
25 \ Q_Condorcerro
26 \ Reach
Caudal observado vs Caudal simulado Tablachaca
0
20
40
60
80
100
120
140
160
180
200
Jan-
68
Jul-6
8
Jan-
69
Jul-6
9
Jan-
70
Jul-7
0
Jan-
71
Jul-7
1
Jan-
72
Jul-7
2
Jan-
73
Jul-7
3
Jan-
74
Jul-7
4
Jan-
75
Jul-7
5
Jan-
76
Jul-7
6
Jan-
77
Jul-7
7
Jan-
78
Jul-7
8
Jan-
79
Jul-7
9
Jan-
80
Jul-8
0
Jan-
81
Jul-8
1
m3/
s
25 \ Q_Condorcerro
26 \ Reach
Caudal observado vs Caudal simulado en Corongo
05
1015202530354045
505560
Jan-
68
Jul-6
8
Jan-
69
Jul-6
9
Jan-
70
Jul-7
0
Jan-
71
Jul-7
1
Jan-
72
Jul-7
2
Jan-
73
Jul-7
3
Jan-
74
Jul-7
4
Jan-
75
Jul-7
5
Jan-
76
Jul-7
6
Jan-
77
Jul-7
7
Jan-
78
Jul-7
8
Jan-
79
Jul-7
9
Jan-
80
Jul-8
0
Jan-
81
Jul-8
1
m3/
s
13 \ Q_Manta
14 \ Reach
The calibration parameters obtained for the non-glaciated sub-basin were applied uniformly in all
the three studied basins. However, a modified set of parameters was employed in the La Balsa
sub-basin in the Santa River basin, as explained next, to implicitly account for agricultural water
use in the region. See Annex 7 for further details.
observed
dd
simulated
31
5.2. CALIBRATION AND VALIDATION: THE SANTA RIVER BASIN
Once the modified WEAP model had been tested in non-glaciated sub-basins, it was calibrated
and verified in the glaciated sub-basins of the Santa River, particularly focusing on a comparison
of observed vs. simulated stream flow and observed vs. simulated glacier area. When calibrating
the model in the Santa River, particular attention was paid to assessing the performance of the
glacier module in a subset of the sub-basins the Llanganuco, Paron and La Balsa.
Figure 5.2 presents the comparison of observed and simulated average annual stream-flows for
the entire calibration and verification period (1970-1998) for the selected sub-basins in the Santa
basin. The discontinuous line represents the observed data while the simulated data is the solid
line.
Figure 5.2. Observed and simulated stream-flow in Llanganuco, Paron and La Balsa in the
Santa River basin
La Balsa
Chuquicara
Colcas
Quillcay
Chancos
Quitaracsa
Olleros
Recreta
Pachacoto
Corongo
CollotaLos Cedros
Llanganuco
Paron
Querococha
Aguascocha
Rajucolta
Tablachaca
Cullicococha
Caudal observado vs Caudal simulado en Paron
0
2
4
6
Sep
-69
Sep
-70
Sep
-71
Sep
-72
Sep
-73
Sep
-74
Sep
-75
Sep
-76
Sep
-77
Sep
-78
Sep
-79
Sep
-80
Sep
-81
Sep
-82
Sep
-83
Sep
-84
Sep
-85
Sep
-86
Sep
-87
Sep
-88
Sep
-89
Sep
-90
Sep
-91
Sep
-92
Sep
-93
m3
/s
31 \ Q_Paron
32 \ Reach
Caudal observado vs Caudal simulado en La Balsa
0
50
100
150
200
250
300
350
400
450
500
Sep
-69
Sep
-70
Sep
-71
Sep
-72
Sep
-73
Sep
-74
Sep
-75
Sep
-76
Sep
-77
Sep
-78
Sep
-79
Sep
-80
Sep
-81
Sep
-82
Sep
-83
Sep
-84
Sep
-85
Sep
-86
Sep
-87
Sep
-88
Sep
-89
Sep
-90
Sep
-91
Sep
-92
Sep
-93
Sep
-94
Sep
-95
Sep
-96
Sep
-97
Sep
-98
m3
/s
67 \ Q_La Balsa
68 \ Reach
Caudal observado vs Caudal simulado en Llanganuco
0
5
10
Sep
-69
Sep
-70
Sep
-71
Sep
-72
Sep
-73
Sep
-74
Sep
-75
Sep
-76
Sep
-77
Sep
-78
Sep
-79
Sep
-80
Sep
-81
Sep
-82
Sep
-83
Sep
-84
Sep
-85
Sep
-86
Sep
-87
Sep
-88
Sep
-89
Sep
-90
Sep
-91
Sep
-92
Sep
-93
Sep
-94
Sep
-95
m3
/s
21 \ Q_Llanganuco
22 \ Reach
La Balsa
Chuquicara
Colcas
Quillcay
Chancos
Quitaracsa
Olleros
Recreta
Pachacoto
Corongo
CollotaLos Cedros
Llanganuco
Paron
Querococha
Aguascocha
Rajucolta
Tablachaca
Cullicococha
Caudal observado vs Caudal simulado en Paron
0
2
4
6
Sep
-69
Sep
-70
Sep
-71
Sep
-72
Sep
-73
Sep
-74
Sep
-75
Sep
-76
Sep
-77
Sep
-78
Sep
-79
Sep
-80
Sep
-81
Sep
-82
Sep
-83
Sep
-84
Sep
-85
Sep
-86
Sep
-87
Sep
-88
Sep
-89
Sep
-90
Sep
-91
Sep
-92
Sep
-93
m3
/s
31 \ Q_Paron
32 \ Reach
Caudal observado vs Caudal simulado en La Balsa
0
50
100
150
200
250
300
350
400
450
500
Sep
-69
Sep
-70
Sep
-71
Sep
-72
Sep
-73
Sep
-74
Sep
-75
Sep
-76
Sep
-77
Sep
-78
Sep
-79
Sep
-80
Sep
-81
Sep
-82
Sep
-83
Sep
-84
Sep
-85
Sep
-86
Sep
-87
Sep
-88
Sep
-89
Sep
-90
Sep
-91
Sep
-92
Sep
-93
Sep
-94
Sep
-95
Sep
-96
Sep
-97
Sep
-98
m3
/s
67 \ Q_La Balsa
68 \ Reach
Caudal observado vs Caudal simulado en Llanganuco
0
5
10
Sep
-69
Sep
-70
Sep
-71
Sep
-72
Sep
-73
Sep
-74
Sep
-75
Sep
-76
Sep
-77
Sep
-78
Sep
-79
Sep
-80
Sep
-81
Sep
-82
Sep
-83
Sep
-84
Sep
-85
Sep
-86
Sep
-87
Sep
-88
Sep
-89
Sep
-90
Sep
-91
Sep
-92
Sep
-93
Sep
-94
Sep
-95
m3
/s
21 \ Q_Llanganuco
22 \ Reach
Caudal observado vs Caudal simulado en Llanganuco
0
5
10
Sep
-69
Sep
-70
Sep
-71
Sep
-72
Sep
-73
Sep
-74
Sep
-75
Sep
-76
Sep
-77
Sep
-78
Sep
-79
Sep
-80
Sep
-81
Sep
-82
Sep
-83
Sep
-84
Sep
-85
Sep
-86
Sep
-87
Sep
-88
Sep
-89
Sep
-90
Sep
-91
Sep
-92
Sep
-93
Sep
-94
Sep
-95
m3
/s
21 \ Q_Llanganuco
22 \ Reach
32
The simulation results show a basic agreement in the pattern of low and high discharge.
However, there are important differences between simulated and observed values. Most notably,
some of the peak flows are missed, which is not surprising given the use of a monthly rainfall-
runoff model applied at a fairly coarse spatial resolution. This indicates that more fine-tuning is
desirable. However, the model captures well the conditions during the low-flow period, which is
of critical importance to hydropower system operators, particularly at the tail end of the
modeling period.
To simplify these results, Figure 5.3 depicts the aggregated observed and simulated stream flow
for the period of 1970 to 1999 with the glacier module.
Figure 5.3. Comparison of the simulated and observed Q for 1970–1999 (Q = average flow in selected glaciated sub-basins)
-
1
2
3
4
5
6
7
8
9
10
Chancos Colcas Cedros Llanganuco Paron
Q (m
3/s
)
Q av sim
Q av obs
Table 5.1 below shows the full calibration and validation statistics for the Santa River basin. The
calibration of larger glaciers is better than that of smaller glaciers, most likely due to the fact that
small glaciers are more likely to be dominated by unique conditions that are not well captured by
either the glacier module itself or the regional parameterization that was developed for the Santa
River basin. The final results of the Santa River’s monthly discharges at La Balsa, which is the
point of diversion to the Cañón del Pato hydropower project, and other points directly below
glaciers provide reasonable calibration and validation statistics.
33
Table 5.1. Calibration and validation statistics for the Santa River18
Calibration Validation
Sub-watershed Period n RMSE BIAS Ef Period n RMSE BIAS Ef
1 - La Recreta 1969-1979 120 0.43 39% 0.63 1979-1989 120 0.43 44% 0.50
2 - Pachacoto 1969-1979 120 0.51 9% 0.64 1979-1989 120 0.53 13% 0.55
3 - Querococha 1969-1979 120 1.37 1% 0.72 1979-1989 120 1.50 20% 0.73
4 - Olleros 1969-1979 120 0.54 7% 0.73 1979-1989 120 0.55 -4% 0.63
5 - Quillcay 1969-1979 120 0.43 9% 0.65 1979-1989 120 0.45 -2% 0.63
6 - Chancos 1969-1979 120 0.26 20% 0.30 1979-1989 120 0.36 -4% 0.63
7 - Llanganuco 1969-1979 120 0.73 35% -0.60 1979-1989 120 0.92 -15% 0.13
8 - Paron 1969-1979 120 1.70 6% 0.25 1979-1989 120 0.74 -44% -1.60
9 - Artesoncocha * * * * * * * * * *
10 - Colcas 1969-1979 120 0.34 24% 0.34 1979-1989 120 0.38 4% 0.24
11 - Los Cedros 1969-1979 120 0.98 3% 0.34 1979-1989 120 0.79 -17% 0.20
12 - Quitaracsa 1969-1979 120 0.29 -7% 0.64 1979-1989 120 0.23 -23% 0.20
13 - La Balsa 1969-1979 120 0.03 3% 0.70 1979-1989 120 0.03 1% 0.72
14 - Corongo (Manta) 1969-1979 120 0.19 -12% 0.54 1979-1989 120 0.20 -9% 0.40
15 Chuquicara 1969-1979 120 0.01 4% 0.69 1979-1989 120 0.02 1% 0.77
16 - Tablachaca (Condorcerro) 1969-1979 120 0.05 17% 0.67 1979-1989 120 0.05 16% 0.61
17 - Puente Carretera 1969-1979 120 0.01 290% -165.00 1979-1989 120 0.01 360% -190.00
Table 5.1 summarizes the results of the calibration and validation statistics for the Santa River
basin. The period comprising 1969 to 1979 was used for calibration; the observed data is used to
estimate the model parameters. The period 1979 to 1989 is used for verification. A period not
used in the calibration that seeks to confirm that the model is applicable to conditions different
from those used in the calibration. Three measures of goodness of fit are used19
, these include the
Root Mean Square Error (RMSE), Bias and the Efficiency coefficient, which are standard
measurements used in this type of analysis.
The calibration of larger glaciers is better than that of smaller glaciers, most likely due to the fact
that small glaciers are more likely to be dominated by unique conditions that are not well
captured by either the glacier module itself or the regional parameterization that was developed
for the Santa River basin20
.
The final results of the Santa River’s monthly discharges at La Balsa, which is the point of
diversion to the Cañón del Pato hydropower project, and other points directly below glaciers
18
Root Mean Square Error (RMSE) is a frequently-used measure of the differences between values predicted by a
model and the values actually observed. It is a good measure of precision or repeatability, which is the degree to
which repeated measurements under unchanged conditions show the same results. Bias: perceptual bias; and Ef:
Nash-Sutcliffe efficiency coefficient is used to assess the predictive power of hydrologic models. It is a criterion that
measures the fraction of the variance of observed values explained by the model. It ranges from minus infinity to
1.0. An efficiency of 1 corresponds to a perfect match of modeled discharge to the observed data. Therefore, large
positive values, close to 1.0, are sought. 19
Root Mean Square Error (RMSE) is a frequently-used measure of the differences between values predicted by a
model and the values actually observed. It is a good measure of precision or repeatability, which is the degree to
which repeated measurements under unchanged conditions show the same results. Bias: difference between mean
values (observed and simulated) expressed in percentage; and Ef: Nash-Sutcliffe efficiency coefficient is used to
assess the predictive power of hydrologic models. It is a criterion that measures the fraction of the variance of
observed values explained by the model. It ranges from minus infinity to 1.0. An efficiency of 1 corresponds to a
perfect match of modeled discharge to the observed data. Therefore, large positive values, close to 1.0, are sought. 20
The data at the Llaganuco station exemplifies the calibration and verification results in small sub-basins. The bias
estimator is high and its efficiency coefficient is low.
34
indicate a good model fit, as exemplified by the calibration and validation statistics; small biases
and high efficiency coefficient.
Figure 5.4 illustrates the results at La Balsa with and without glacier retreat and Figure 5.5
represents the aggregated values shown in Figure 5.3 (earlier). These results show that the
average simulated discharge of about 95 m3/sec with glaciers, are roughly 58 percent higher than
the average discharge of 60 m3/sec simulated in the absence of glaciers. These results thus
confirm both that glaciers are important in the production of water in the Santa River basin
(Vergara et al. 2007), and that the glacier module is important in accurately modeling the runoff
in the glaciated basins.
Figure 5.4. Comparison of stream-flow at La Balsa with and without glaciers
Observed vs Simulated Streamflow in La Balsa
0
50
100
150
200
250
300
350
400
450
500
Sep
-69
Sep
-70
Sep
-71
Sep
-72
Sep
-73
Sep
-74
Sep
-75
Sep
-76
Sep
-77
Sep
-78
Sep
-79
Sep
-80
Sep
-81
Sep
-82
Sep
-83
Sep
-84
Sep
-85
Sep
-86
Sep
-87
Sep
-88
Sep
-89
Sep
-90
Sep
-91
Sep
-92
Sep
-93
Sep
-94
Sep
-95
Sep
-96
Sep
-97
Sep
-98
m3
/s
With glacier
Without glacier
Figure 5.5. Another comparison of stream-flow at La Balsa with and without glaciers
-
20
40
60
80
100
Q ave obs Q ave sim with glacier Q ave sim without glacier
Q (m
3/s
)
In conclusion, the calibrated hydrologic model results, in the selected discharge points of
interest, provide a reasonable representation of the hydrologic behavior of the basin under
current climate conditions. These results represent the baseline for analyzing possible changes in
hydrologic behavior. Results from the final calibration-validation of the Santa River model are
presented in detail in Annex 8.
35
5.3. CALIBRATION AND VALIDATION: RÍMAC-MANTARO RIVER BASINS
A similar calibration-validation effort was undertaken for the combined Rímac-Mantaro River
basins. These basins are more complex than the Santa River basin, with many more sub-basins
and a higher level of man-made infrastructure to regulate water flows like reservoir storage and
release. As a consequence, the final calibration-validation of the application of the hydrologic
model in the Mantaro/Rímac River is less precise; it presents higher uncertainty and statistics
denote medium fit.
The Rimac-Mantaro River basins (Figure 4.3) includes 38 reservoir objects, 22 demand sites
representing the urban and rural water demands in individual provinces, along with 276 (102 for
Rímac and 174 for Mantaro) sub-catchment objects. Five diversion points and nine run-of-river
hydropower objects are used to represent the hydropower production system. There were 28
stream-flow gauges available for calibration-validation of the hydrologic routines.
The calibration-validation focused on the analysis of several gauge stations (19 stations for
Mantaro, 6 stations for Rímac), with the 1970–81 calibration period; the validation 1981–96
period. For presentation purposes the analysis is focused on sites with large mean water
discharges. For the Mantaro River, stations include Pongor and Mejorada; for the Rímac River,
the two main stations upstream from Lima: Chosica and Surco. While modeling in the Mantaro
River basin extends to the location of the projected hydropower facility at La Guitarra,
information for calibration only reaches the hydrometric station of Pongor. For practical
purposes, input to the La Guitarra hydropower development is defined at the Pongor site.
Similarly, while the Rímac River modeling extends to the point of water diversion to the city of
Lima, the closest gauge in the system is Chosica, located upstream. In analyzing the results it is
necessary to make adjustments, which is a recurring theme in hydrologic modeling.
5.3.1. Parameter setting
The calibration first focused on obtaining a set of parameters applicable to the entire river basin
to reasonably represent the hydrology of the main course of the Mantaro and Rímac Rivers.
Because both rivers are located in different types of basins (the Mantaro is in the Amazon Basin,
the Rímac drains to the Pacific), parameter values for each basin were adjusted separately to
represent the different physical processes of each basin. Each basin arrived at an internally
uniform set of parameters, as indicated in Table 5.2. The table shows that the runoff resistance
factor and the root zone conductivity parameters were defined in terms of land cover (crops,
underbrush [matorral], coastal plain, tundra).
36
Table 5.2. Land use parameter values for the glacier module
Temperature and humidity data was also needed. The only good-quality set of long and
continuous time-series data that exists is collected from the Cercapuquio station in Mantaro basin
(12.422°S, 75.417°W). Continuous temperature data for each catchment was obtained by adding
a temperature gradient of 0.6°C/100m to the temperature observed at Cercapuquio. For humidity
and wind speed, the long-term monthly time-series at Cercapuquio was assumed to apply to all
catchments.
5.3.2. The Mantaro River basin
The Mantaro River basin was calibrated by using monthly average stream flows for the
calibration period. Figure 5.6 presents the seasonal fluctuations of observed and simulated stream
flows for the large downstream stations of Mantaro basin, showing the relationship between the
observed and simulated discharge values. The calibration indicates a good fit in reproducing the
monthly distribution of discharges. An exception is the case of the site at Pongor, where the
simulated runoff fails to capture the pronounced high flows characteristics of the February and
March periods.
37
Figure 5.6. Observed (Qo) and simulated (Qs) monthly average stream flows for the
Mantaro River basin during the 1970–81 calibration period
For validation in the Mantaro basin, the simulated monthly flows for both the calibration and
validation period (1966-1996) were plotted against the observed values, as shown in Figure 5.7.
Good correlations appear to exist for these stations during this period, with the exception of the
Sheque station21
.
21 However, this error can be explained by the fact that the station has problems in correctly estimating flows during
low-water periods, and the simulation shows 50 percent of the amount observed.
38
Figure 5.7. Observed (blue line) and simulated (red line) flow rates for the 1966–96
reference period at four gauge stations in the basin
To represent the results graphically, Figure 5.8 plots the simulated vs. observed data during the
reference period. In the case of these stations, the correlation coefficients are appropriate. Based
on the results the selected stations can be used in an analysis of the evolution of flow in the
future.
39
Figure 5.8. Correlation between observed and simulated flow rates for the reference period
In summary for the Mantaro basin, validation results for the selected stations indicate a
reasonable (good) fit, although bias and efficiency indicators are not as good. For detailed
applications it is recommended that calibration be improved by incorporating the spatial
distribution of soil characteristics.
40
Table 5.3. Criteria for the calibration and validation periods for the Mantaro sub-basins
5.3.3. The Rimac River basin
Next, the calibration-verification focused on the Rimac River. Figure 5.9 presents seasonal
fluctuations during the calibration period of observed and simulated stream flows at the Chosica
and Surco stations. In Chosica the simulation seems to correlate the observed trend rather well,
only showing clearly higher values during the high peak period (February-March).
Figure 5.9. Observed and simulated monthly average stream flows for the Rímac River
basin during the 1970–81 calibration period
41
Similarly to the data for Mantaro, earlier presented, the monthly observed and simulated stream
flows for the validation period are plotted for the Chosica and Surco stations in Figures 5.10 and
5.11. Good correlations appear to exist for these stations.
Figure 5.10. Observed (Qo)and simulated (Qs) stream flows at the Chosica station in the
Rímac River basin
Figure 5.11. Observed (Qo) and simulated (Qs) streamflows at the Surco station in the
Rímac River basin
42
5.3.4. Conclusions of the calibration-validation in the Rimac-Mantaro River system
A double validation of the model was conducted by comparing first, observed and simulated
stream flows at 25 control points in the Rímac-Mantaro system; and second, the glacier area
observed with Landsat images for two years (1988 and 1996) with that calculated by the model.
This validation yielded reasonable results. However, for the Rímac-Mantaro system, future
simulations would need to reproduce the reservoirs’ operation and management: 16 reservoirs
were operational from 1995 to 2000, and 28 reservoirs are planned for the future.
The flows in the Mantaro/Rímac River system are more hydraulically altered by the operation of
reservoirs and diversions than are those in the Santa River. As a result, the model does not
perform as well in some stations. However, it does perform better during the critical low-flow
period. It is likely that the overall performance of the model could be improved if the
parameterization of the WEAP reservoir and diversion infrastructure is improved.
The results of the glacier area evolution simulation in the Mantaro/Rímac River system were
satisfactory. The observed initial glacier area in these basins in 1970 was 113 km2, which
decreased to roughly 40 km2 in 1997. This trend was well captured by the model when simulated
and observed glaciated areas were compared at discrete times during the calibration-validation
period (see Annex 9 for more information). Compared to the Santa River, the glaciers in the
Mantaro and Rímac River basins are much smaller, and thus their runoff contribution is not
significant.
43
Chapter 6
Results from the Hydrology Analysis
This chapter presents the results of the model once calibrated. The analysis includes an
estimate of increases in temperature, changes in precipitation, rate of glacier retreat and
drying of mountain wetlands. Although the purpose of this study is to develop a useful
methodology, and not necessarily to produce an assessment of the impacts of climate change in
the selected basins, the chapter nonetheless also discusses how hydrology might be impacted by
the projected consequences of climate change.
6.1. VISUALIZATION OF CLIMATE CHANGE IN THE SANTA BASIN
The two future climate conditions developed from the 16 GCM ensemble (marginal warming
conditions or 0.5oC increase in temperature and a 15 percent increase in precipitation; and fast
warming conditions, 2oC increase in temperature and a 10 percent decrease in precipitation),
along with results from the trend analysis, were used as the bases for the analysis. This section
describes the possible hydrologic response of the Santa River basin to these climate change
scenarios.
6.1.1. Simulation of future glacier dynamics
The glaciers of the Santa River basin are likely to experience significant size reduction under
both conditions. Using 2006 glacier coverage as a baseline, the simulated reduction in glaciated
area in the Santa River basin over the next 30 years would be 25 percent for the marginal
warming scenario and 47 percent for the fast warming scenario22
, as shown in Table 6.1.
Table 6.1. Simulated reductions in glaciated area between 2006 and 2036 under two climate
projections Total Area 2006 2021 2036
(km2) 06-21 21-36 06-36
Dry 347 257 182 26% 29% 47%
Wet 347 300 260 14% 13% 25%
% Change
When assessing the spatial distribution of the glacier retreat under fast warming conditions in the
Santa River basin by 2036, it can be noted that the impacts will be distributed unevenly (Figure
6.1). Under these conditions, it is likely that the some smaller glaciers in the Santa River basin
would dramatically shrink in the future. Although the impact of these reductions may not be
22
These conditions are developed in Chapter 2.2 ―Ensembles to simulate future climate at a basin level‖.
44
acutely felt everywhere in the basin, locally the effects of the loss of glacier ice could have a
profound impact on stream flow conditions.
Figure 6.1. Simulated remaining glacier ice coverage in 2036 under the fast warming
climate projection
Based on these glacier cover reduction projections, a preliminary assessment was conducted on
how changes in the seasonal patterns of stream-flow may constrain the operations of the Cañón
del Pato project. This assessment was done under both climate conditions, slow warming and fast
warming. Over the entire period, the contribution of glacier melt to the total runoff at the Cañón
del Pato diversion was estimated at 9 percent under slower warming conditions and 16 percent
for the fast warming conditions. Based on these estimates under slower warming (and wetter)
conditions, rainfall-runoff would create relatively more stream flow in non-glaciated regions of
the Santa River basin, reducing the relative contribution of glacier melt. At the same time, glacier
retreat would be accelerated and more of the stream flow would come from glacier melt.
The assessment indicates that both scenarios would provide more water during the summer and
fall high-flow period (rainy season) when compared to historical conditions. At the same time,
the flow would be reduced during the low-flow period in both scenarios (Figure 6.2)23
. Although
23
As the climate warms, less of the precipitation falls as snow and more falls as rain on snow- and ice-covered
peaks. This produces reductions in the accumulation of snow during the wet season and in the melting of snow and
ice during the dry season. Since snowmelt is an important water resource during the dry season, there is concern that
45
these results are by no means definitive, the repercussions of the increased variability and
reduced minimum flows might require additional water regulation infrastructure to cope with
increased variability and reduced minimum flows.24
Figure 6.2. Flow through Cañón del Pato for historical conditions, marginally warmer and
fast warming scenarios Streamflow Through Canon del Pato Diversion
-
50
100
September December March June
Mo
nth
ly A
ve
rag
e (
m3
/s)
Max Diversion
Historic
Wet
Dry
|---Spring----|---Summer---|----Fall------|---Winter---|
Streamflow Through Canon del Pato Diversion
-
50
100
September December March June
Mo
nth
ly A
ve
rag
e (
m3
/s)
Max Diversion
Historic
Wet
Dry
|---Spring----|---Summer---|----Fall------|---Winter---|
6.1.2 Expected overall hydrologic response to observed trends
Data from the trend analysis (described in Section 3.3) was also used as an input to the WEAP
hydrologic model in the Santa River basin in order to strengthen the analysis. The results are
summarized by analyzing the expected hydrologic changes at two sites: La Balsa and
Condorcerro. To understand the future changes, it is important to have a historical record for
comparison. Figure 6.3 presents the flows for the baseline period, based on data provided by
SENAMHI and the results from the WEAP simulations.
global warming will reduce the supply of water during the dry season in mountainous regions in many parts of the
world (Vergara et al. 2009; Leung and Ghan 1999; Ghan and Shippert 2006). 24
Although the ensemble approach and the use of extreme climate scenarios reduce uncertainty, they do not provide
guidance on the relative probabilities of such scenarios or their consistency with observed trends. Risk analysis
based on these scenarios is thus uncertain.
46
Figure 6.3. Observed (blue line) and simulated (red line) discharges using trend analysis at
selected sites on the Santa River, 1966–96
0
50
100
150
200
250
300
350
400
450
Se
p 1
96
6
Se
p 1
96
7
Se
p 1
96
8
Se
p 1
96
9
Se
p 1
97
0
Se
p 1
97
1
Se
p 1
97
2
Se
p 1
97
3
Se
p 1
97
4
Se
p 1
97
5
Se
p 1
97
6
Se
p 1
97
7
Se
p 1
97
8
Se
p 1
97
9
Se
p 1
98
0
Se
p 1
98
1
Se
p 1
98
2
Se
p 1
98
3
Se
p 1
98
4
Se
p 1
98
5
Se
p 1
98
6
Se
p 1
98
7
Se
p 1
98
8
Se
p 1
98
9
Se
p 1
99
0
Se
p 1
99
1
Se
p 1
99
2
Se
p 1
99
3
Se
p 1
99
4
Se
p 1
99
5
M3
/se
g
La Balsa
Observado Simulado
0100200300400500600700800900
1000
Se
p 1
96
6
Se
p 1
96
7
Se
p 1
96
8
Se
p 1
96
9
Se
p 1
97
0
Se
p 1
97
1
Se
p 1
97
2
Se
p 1
97
3
Se
p 1
97
4
Se
p 1
97
5
Se
p 1
97
6
Se
p 1
97
7
Se
p 1
97
8
Se
p 1
97
9
Se
p 1
98
0
Se
p 1
98
1
Se
p 1
98
2
Se
p 1
98
3
Se
p 1
98
4
Se
p 1
98
5
Se
p 1
98
6
Se
p 1
98
7
Se
p 1
98
8
Se
p 1
98
9
Se
p 1
99
0
Se
p 1
99
1
Se
p 1
99
2
Se
p 1
99
3
Se
p 1
99
4
Se
p 1
99
5
M3
/se
g
Condorcerro
Observado Simulado
In terms of projected impact, the expected discharges by mid-century are lower than the
historical record (Figure 6.4 displays the results of the hydrologic response in the 2050s). which
summarizes the expected changes in the average monthly discharges at the selected sites.
Monthly discharges by 2050 at La Balsa (Fig. 6.5) are expected to be lower all year-round in
comparison to historical values. The mean annual discharge in the observed period is 85.8 m3/sec
and would be reduced by 21 percent by mid-century. Minimum values in August are also
reduced by 28 percent. Similarly, average flow at Condorcerro is expected to decrease by more
than 6 percent and its minimum flow by 18 percent.
Figure 6.4. Results of the trend analysis for mid-century in the Santa River
0
50
100
150
200
250
Jan
20
50
Jul 2
05
0
Jan
20
51
Jul 2
05
1
Jan
20
52
Jul 2
05
2
Jan
20
53
Jul 2
05
3
Jan
20
54
Jul 2
05
4
Jan
20
55
Jul 2
05
5
Jan
20
56
Jul 2
05
6
Jan
20
57
Jul 2
05
7
Jan
20
58
Jul 2
05
8
Jan
20
59
Jul 2
05
9
M3
/se
g
La Balsa (2050-59)
0
50
100
150
200
250
300
350
400
450
500
Jan
20
50
Jul 2
05
0
Jan
20
51
Jul 2
05
1
Jan
20
52
Jul 2
05
2
Jan
20
53
Jul 2
05
3
Jan
20
54
Jul 2
05
4
Jan
20
55
Jul 2
05
5
Jan
20
56
Jul 2
05
6
Jan
20
57
Jul 2
05
7
Jan
20
58
Jul 2
05
8
Jan
20
59
Jul 2
05
9
M3
/se
g
Condorcerro(2050-59)
47
Figure 6.5. Comparison of average monthly discharges in the Santa River between
historical (blue line) projected by mid-century (red line) values (m3/s), based on trend
analysis
As previously mentioned, the analysis assumes that the linear trends identified will continue into
the future. This hypothesis may be considered a lower limit with respect to future climates, since
it assumes that the warming will not intensify and that the trend is linear throughout the
projection period. Also, no sensitivity analysis was conducted, since the main objective of this
study was developing methodologies rather than providing results for a particular application.
However, a proper sensitivity analysis is required for any formal application on the possible
range of climate impacts expected in Peru.
6.2. VISUALIZATION OF CLIMATE CHANGE IN THE RÍMAC AND MANTARO BASINS
The glaciers in Mantaro and Rimac River basins are rather small and their runoff contribution is
not significant. Therefore instead of simulating future glacier dynamics, the study focuses only
on analyzing the expected hydrologic response to observed trends in these two basins.
The same procedure of applying data from the rainfall trend analysis was used for the Rímac and
Mantaro basins. The results from the analysis are presented for four sites: Chosica and Surco in
the Rímac River basin, and Pongor and Mejorada in the Mantaro basin. Once again, using the
observed linear trend of the past 30 years as a scenario of how quickly climate will continue to
change in the future is a conservative assumption.
Again, the first step was to establish a baseline for comparison. Figure 6.6 shows the
comparison between the simulated and observed discharges in the selected control sites from
1966 through 1996. For the four gauge stations used in the analysis, the data shows a good fit. In
the Rímac basin and especially in the Chosica station, simulated runoff gives higher values than
the observed data during the peak periods.
48
Figure 6.6. Observed (blue line) and simulated (red line) discharges using trend analysis at
selected sites on the Rímac and Mantaro Rivers, 1966–96
The historical trends were compared with the hydrologic response in the Rímac-Mantaro basin in
the 2050s ( Figure 6.7). The decade of analysis includes a dry period similar to the observed
during 1992-93. The climate scenario is able to capture this extreme event, and the hydrologic
model describes the expected flows caused by such extreme event. Notice also that for the
stations located upstream, Chosica and Surco, the first half of the decade is dry and the second
half is wetter. But this behavior has only limited influence the flows downstream. The stations at
Pongor and Mejarada present a dry spell only in the middle of the decade. Only a sustained dry
period upstream produces drought conditions in the lower Mantaro basin.
49
Figure 6.7. Results of the trend analysis for the mid-century in the Rímac and Mantaro
Rivers
Summarizing the results for the Mantaro River basin, Figure 6.8 provides the average monthly
results for the historical data, for 2050-2059, and for 2090-2099, when the trend-line scenario is
used as an input in the hydrologic model. For the Mantaro River basin, as indicated in the figure
for the sites at Pongor and Mejorada, overall water availability decreases. However, a shift in the
distribution of runoff is expected, with reduction during the wet months. In Mejorada, the low-
flow period is considerably reduced, with runoff increasing in November, December and January
and decreasing in the high-peak periods of February and March.
50
Figure 6.8. Comparison of average monthly discharges in the Mantaro River for observed,
mid-century and end of century values (m3/s), based on trend analysis
For the same climate scenario, the expected hydrologic changes along the Rímac River have
different trends. As illustrated in Figure 6.9 for the Chosica and Surco stations, water availability
is expected to grow, particularly with increasing discharges in February and March, while there
will be little or no reduction in the dry months from June to November. More water is welcome,
especially for hydropower production, agriculture and water supply to Lima. On the other hand,
excessive runoff might lead to flooding, which could affect the most vulnerable inhabitants of
Lima.
51
Figure 6.9. Comparison of average monthly discharges in the Rímac River for observed
(blue), mid-century (red) and end of century (green) values (m3/s), based on trend analysis
6.3. RESULTS
The application of the methodology in Santa, Rimac and Mantaro River basins suggests that the
hydrology model is useful for a wide range of conditions. The results indicate that climate
change brings new considerations into rainfall-runoff modeling. The combined expertise of
glaciologists, hydrologists, researchers and practitioners was necessary to produce a module that
provides good results in the sites studied.
Although the purpose of this study is not to produce an assessment of climate change in the
selected basins, the analysis of the data provides some insights. The results do highlight the
inherent difficulties of assessing the hydrologic response to climate change in complex
orographic settings such as those found in Peru. Each basin needs to be studied in detail. It is
recommended that detailed sensitivity analyses on the hydrologic and climate components be
conducted to gain reliability in the results and possible recommendations.
In summary, to illustrate the developed approach and the usefulness of the data generated under
the selected climate scenarios, the following results are presented:
The simulated flows for the 2050–59 period show a reduction in their peaks for all the
monitoring stations in the Mantaro River. Net reductions in water availability are also
projected in the Rímac River. Monthly discharges at the La Balsa station of the Santa
River are also expected to decrease year-round in comparison to observed historical
values. The mean annual discharge could be reduced by 21 percent by mid century.
Glacier loss is expected to be significant during the period.
52
Simulated low water flows are expected to increase in the Mantaro River. There are no
significant changes expected in low water flows in the Rímac River, with the exception
of Sheque during September, October and November when the flows in this station are
expected to be lower. In the Santa River low flows are estimated to decrease by around
28 percent, which raises immediate concerns due to its impact on energy and agricultural
production.
Based on trend analysis, it was not possible to estimate when the loss of glacier runoffs
will be severe enough to reach a tipping point beyond which the ability to generate
hydropower at the Cañón del Pato facility will be substantially reduced on an annual
basis.
.
53
C H A P T E R 7
Conclusions
This report describes a combination of climate and hydrology analysis used to simulate current
and projected future conditions in mountain regions at a regional and at a basin-level. The tools
used were applied to Peru but could be similarly deployed for other mountain basins in the
Andes to simulate current and predict future impacts of climate on hydrology.
7.1. KEY TOOLS
In order to assess the expected future climate conditions, three tools were used.
The Earth Simulator running the MRI-GCM on the A1B emission trajectory provided a
high resolution simulation of the future climate in Peru. The simulation included a
detailed map of annual main rainfall for end of century and the estimate of precipitation
anomalies. The Earth Simulator results indicate an increased variability in hydrology
conditions in the future, including intensification of rainfall and lengthening of drought
periods. These conditions would lead to increases in incidents of flooding, losses of
topsoil and erosion as well as a lengthening of dry periods. Soil is expected to become
considerably drier in most parts of the country, with the exceptions of Peru’s Amazon
region and southern coastal areas. Drier soil would place additional pressure on ground
vegetation but would also reduce the net runoffs from mountain ecosystems to
downstream basins. An analysis of anomalies in runoffs at ground level, using the results
of the Earth Simulator, also points to significant changes in stream-flows in the northeast
and east of the country and in southern coastal areas.
As a complement, and in order to assess the range of precipitation and temperature
change that might be anticipated at selected meteorological stations over the 21st century,
the report uses the outputs from a set of 16 GCM. Based on this information, two 2040
climate change conditions were selected for use at a basin level: a slower warming
condition with a 15 percent increase in precipitation and a 0.5oC increase in temperature;
and a faster warming condition with a 10 percent decrease in precipitation and a 2oC
increase in temperature. These represent a range of conditions projected by the ensemble
and were used to estimate a hydrology response at a basin level.
To strengthen the robustness of those projections, the study also used a trend analysis and
compared the results to the GCMs.
Most importantly, the study develops a hydrologic planning tool suited to be used in
mountainous regions. The WEAP model was used as a starting point for building the hydrologic
modeling tool, and it was modified to incorporate glacier and páramo modules. This resulted in a
unique tool that incorporates mountain systems such as glaciers and neotropical alpine wetland
ecosystems locally known as páramos into the hydrology analysis in mountain regions, for the
54
first time. Although the model was adapted to include páramos, the páramo module was not used
in the scope of this study because there are no extensive páramo landscapes present in the three
selected river basins in Peru.
7.2. RESULTS
The estimated hydrologic response to climate change by mid-century, using the projected climate
conditions at a basin level can be characterized as follows:
The mean annual discharge in the Santa River basin could decrease even by 21 percent
by the mid century. Low flows on the Santa River are estimated to decrease by around 28
percent, a figure that calls for further analysis due to its impact on energy and agricultural
production. Monthly discharges at the La Balsa station on the Santa River are expected to
decrease year-round, in comparison to observed historical values. Glacier loss is expected
to be significant during the period.
At the Mantaro River the simulated flows for the 2050–59 period show a reduction in
peaks for all the monitoring stations. Net reductions in water availability are also
projected in the Rímac River. At the same time, low water flows are expected to increase
in the Mantaro River. In the Rímac River no significant low flow changes are expected.
However, these should be considered preliminary results obtained in the process of developing a
methodology to assess climate impacts in mountain hydrology.
7.3. LIMITATIONS
The conducted work presents limitations that should be kept in mind when the results are
assessed, or the methodology is used elsewhere. These limitations include:
a. Deficiencies in data collection and analysis in defining the inputs to the hydrologic
model: Data related to hydrologic soil characteristics was not collected, reducing the
model’s ability to account for variations in soil storage capabilities, and constraining
better model performance statistics.
b. Calibration: More work is recommended before preliminary results are used in the
decision-making process. Greater flexibility should be allowed in the spatial variation of
the hydrologic parameters used.
c. Climate planning scenarios: More research is needed to assess variability and its trends,
the validity of the linear extrapolation of observed trends, and improved statistical tools
to assess the reliability of trends individually and in clusters.
d. Need for sensitivity analysis: A thorough sensitivity analysis is strongly recommended in
order to analyze the expected variability of the results. This sensitivity should span from
climate planning scenarios to land use changes driven by changes in precipitation and
temperature. To support decision-making processes, the sensitivity analysis should also
include socioeconomic considerations.
Next steps could be to analyze the social and economic consequences associated with hydrologic
changes, including those for the energy, agriculture and water supply sectors.
55
7.4. PRODUCTS
The study achieved its goal of developing an approach to assess the hydrologic response to
global climate change in tropical mountain basins. Below is a list of products that were produced
as a result of the analysis:
A high-resolution projection of the water cycle in Peru, including net impact on stream
flows.
A database of climate scenarios built in the National Meteorological and Hydrological
Service of Peru (Servicio Nacional de Meteorología e Hidrología del Perú SENAMHI)
with public access, covering the entire country. The database contains the results from the
Earth Simulator as well as the ensemble from the Community Climate System Model
(Version 3) of NCAR, downscaled at a 2.5-minute spatial resolution.25
The application and calibration of a basin-level hydrologic model for river basins in the
Peruvian Andes.
A hydrologic simulation module of the dynamic behavior of glaciers under the influence
of global warming.
Integration of a glacier and a páramo module into a hydrologic planning model (WEAP).
Analysis of the impact of climate change on the hydrologic response in three emblematic
basins in Peru.
25 The ensemble covers two periods; five simulations will be conducted for each period: a decade within the recorded climate
history in Peru (1990–1999), and a mid-century scenario (2050–2059).
56
References
Bahr, D. B., Meier, M. F., and Peckham, S. D. 1997. ―The Physical Basis of Glacier Volume-
area Scaling‖, J. Geophys. Res., 102 (B9), 20 355–20 362.
Bradley, R. S., Vuille, M., Diaz, H. F, and Vergara, W. 2006. ―Climate Change: Threats to Water
Supplies in the Tropical Andes.‖ Science, 312, 1755–1756, doi:10.1126/science.1128087.
Comité de Operación Económica del Sistema Interconectado Nacional (COES). 2009.
―Estadística Diaria de Operaciones.‖ [www.coes.org.pe/coes/Estadistica/Diario.asp]
Francou, B., Vuille, M., Wagnon, P., Mendoza, J., and Sicart, J.-E. 2003. ―Tropical Climate
Change Recorded by a Glacier in the Central Andes during the Last Decades of the Twentieth
Century.‖ Chacaltaya, Bolivia, 16_ S, J. Geophys. Res., 108(D5), 4154,
doi:10.1029/2002JD002959, 2003.
Ghan, S. J., and T. Shippert. 2006. ―Physically-based Global Downscaling Climate Change
Projections for a Full Century‖, J. Clim., 19, 1589–1604.
Stone, D. A., and R. Knutti. 2010. Weather and Climate. In: Climate Change and Water
Resources Modelling, Eds: C. F. Fung and A. Lopez. In press.
Knutti, R., T. F. Stocker, F. Joos, and G. K. Plattner. 2002. ―Constraints on Radiative Forcing
and Future Climate Change from Observations and Climate Model Ensembles.‖ Nature, 416,
719– 723.
Leung, L., and S. Ghan. 1999. ―Pacific Northwest Climate Sensitivity by a Regional Climate
Model Driven by a GCM. Part II: 2 x CO2 Simulations.‖ J. Climate 12:2031–2053.
Mizuta R, K. Oouchi, H. Yoshimura, A. Noda, K. Katayama, S. Yukimoto, M. Hosaka, S.
Kusunoki, H. Kawai, and M. Nakagawa. 2006. ―20-km-mesh Global Climate Simulations Using
JMA-GSM model ‒Mean climate states‒‖. Journal of the Meteorological Society of Japan 84:
165‒185.
Nakaegawa, T. and W. Vergara, 2006. First Projection of Climatological Mean River Discharges
in the Magdalena River Basin, Colombia, in a Changing Climate during the 21st Century.
Hydrological Research Letters, 4, 50–54 (2010). Published online in J-STAGE
(www.jstage.jst.go.jp/browse/HRL). DOI: 10.3178/HRL.4.50.
Nohara, D., M. Hosaka, A. Kitoh and T. Oki, 2006. Impact of climate change on river runoff
using multi-model ensemble. J. Hydrometerol., 7, 1076-1089.
57
SENAMHI (Servicio Nacional de Meteorología e Hidrología). 2007. ―Escenarios de Cambio
Climático en la Cuenca del Río Mantaro para el Año 2100.‖ Centro de Predicción Numérica –
CPN.
SENAMHI. 2009a. ―Climate Scenarios for the Santa River Basin to 2030.‖ In Second National
Communication on Climate Change.
SENAMHI. 2009b. ―Climate Scenarios for Peru to 2030.‖ In Second National Communication
on Climate Change.
Thomson M.C, F. J. Doblas-Reyes, S. J. Mason, R. Hagedorn, S. J. Connor, T. Phindela, A. P.
Morse, T. N. Palmer. 2006. ―Malaria Early Warnings Based on Seasonal Climate Forecasts from
Multi-model Ensembles.‖ Nature 439:576–579.
Urrutia, R. and M. Vuille. 2009. ―Climate Change Projections for the Tropical Andes Using a
Regional Climate Model: Temperature and precipitation simulations for the end of the 21st
century‖, J. Geophys. Res., 114, D02108, doi:10.1029/2008JD011021. 1823, 1824, 1825, 1828,
1832, 1837
Vergara, W., N. Toba, D. Mira-Salama and A. Deeb. 2009. ―The Consequences of Climate-
induced Coral Loss in the Caribbean by 2050-80‖. In Environment Matters at the World Bank.
Valuing Coastal and Marine Ecosystem Services. The World Bank, Washington D.C.
Vergara, W., A. Deeb, A. Valencia, R. S. Bradley, B. Francou, S. Hauessling, A. Grunwaldt and
A. Zarzar. 2007. ―Economic Consequences of Rapid Glacier Retreat in the Andes.‖ Journal of
the American Geophysical Union, EOS. Vol. 88, No. 25. June.
―Assessment of the Risk of Amazon Dieback.‖ 2010-submitted. Ed by. W. Vergara and S.
Scholz. World Bank. February 4, Washington D.C.
Yun, X., B. Y. Liu, and M. A. Nearing. 2002. ―A Practical Threshold for Separating Erosive and
Non-erosive Storms.‖ Transactions of the American Society of Agricultural Engineers. 45(6):
1843-1847.