gruppe für hydrologie - framework to assess the effects of · 2015. 11. 17. · a physically based...
TRANSCRIPT
Microsoft Word - whatever
Universität Bern
Universität Bern
Von der Philosophischnaturwissenschaftlichen Fakultät angenommen.
iii
Summary This study presents a robust modelling framework developed for assessing the available water
resources in a mountainous environment today and in the future. The semidiscrete, physicallybased
Penn State Integrated Hydrologic Model (PIHM) has been identified as a suitable hydrological model
for this challenging task. Here we present the customization, the enhancement and the application of
this on a catchment with sparse data in the past and in the present, we evaluate extensively and
thoroughly its performance, its potential and its limitations. The melt modules for snow and ice were
upgraded from a simple temperatureindex model to a model including the influence of global
radiation. Stationary catchment attributes such as soil and land cover data were used to distribute
parameters, and the latters were mainly estimated basing on literature or experiments on site. Besides
the examination of the internal consistency with a multicriteria validation, evaluating snowpack,
icemelt and different components of the water balance, model results have been further validated
with discharge measurements. Results show a very good performance of the model over different
spatial and temporal scales.
Sound hydrological modeling frameworks are needed to serve and support decision making in local to
regional water management, in particular in mountainous regions, as main suppliers of natural water
resources. In view of climate change, increasingly complex physically based models are applied in
order to guarantee the predictability for new climate or environmental boundary conditions. However,
while the hydrological predictability might increase with complex models, data demands of course
increases too.
In the present study we investigate the challenges, the gains and the problems of an increased
complexity in describing snow and precipitation patterns in data sparse regions at higher altitudes and
in an environment with a highly complex topography. Particular efforts were put in improving the
modelling of snow and glacier dynamics and melt including a timevarying albedo and gravitational
redistribution of snow in a hydrological model. Furthermore, an extensive meteorological measuring
network was set up and exploited generating high resolution input data.
We provide evidence of the limits of an enhanced description of the albedo of snow depending on the
surrounding conditions, when other processes we cannot describe emerge. We also show the
importance of the maintenance and extension of shortterm monitoring networks in mountain
regions, demonstrating that an increased sampling of hydrometeorological and cryospheric data –
even on the basis of one to three years – retrieves the best model performance, and therefore these
are crucial for the improvement of the hydrological knowledge of the local system.
With the developed modelling framework we estimated available water resources for the headwaters
of the study area, and made an overall plausibility assessment using all the available data, with very
good results. Overall the headwaters of the catchment provide 110 mio m3. Generally the western
part of the catchment is richer in water resources than the eastern part, corresponding respectively to
80% and 20% of the total water resources. Tseuzier, the subbasin at the western boundary, represents
the “water tower” of the area, supplying 70% of the total estimated water amount
In this work we integrated many relevant aspects of hydrological modelling, especially pertinent to the
context of climate change in alpine regions, like transferability in time and space, as well as flexibility,
and proved our modelling framework to be robust over different scales. As such, it provides a solid
Summary
iv
basis for hydrological analysis of the system under changing forcing conditions, like in the case of
climate change.
We performed an exploratory climate sensitivity analysis, basing on the data of ten climate model
chains driven by the single emission scenario A1B, postprocessed and interpolated to the MeteoSwiss
weather station Montana (as provided by the CH2011 initiative). Two scenario periods in the 21st
century were assessed relative to the short reference period 20072012: the near future 20482053
and the far future 20972102. The most pronounced changes are expected in the snow cover
dynamics in the headwaters of the study area, and evaporation rates in the lower parts of the study
area, with changes in temperature as the main drivers. Basing on the applied scenarios, for all models,
a reshaped annual cycle, with earlier rise of spring runoff, significantly reduced summer runoff, and a
tendency for increased winter runoff in the first half of the 21st century, becoming increasingly
pronounced towards the end of the century, and resulting in a significant change of regime: first
transforming from a bglacionival regime in a nival alpin regime as a transition before becoming,
characterized by a nival de transition regime in the far future.
All in all, the overall available water resources are going to be affected only to a minor extent, with
mean changes in the order of magnitude of 510%, however the major changes estimated for the
summer runoff, reduced by about 20% in the near future, and almost 50% in the far future are
expected to have major consequences for the water management strategies of the region. In
particular in dry years conflicts might arise between different usages, which here are mainly irrigation,
hydropower production, tourism and artificial snow making.
Ackowledgements
“I am neither the first commentator, nor the most luminous. Therefore I will
make no particular claim of originality for the remarks presented here, but can
only hope that they are framed in useful ways”
J.W. Kirchner
Ackowledgements This study is embedded in the framework of the MontanAqua project. MontanAqua is financed by the
NRP 61 (National Research Project 61, Sustainable Water Management). I would like to thank the
Swiss National Science Foundation for funding the MontanAqua research project (4061401259646 1)
within the National Research Program “Sustainable Water Management” (NRP 61).
I would like to thank Gopal Bhatt and Xuan Yu at the Penn State University for their patience, their
help for the customization of the model (developed by their research group), and their support
throughout the first simulations. I also thank Christopher Duffy for his support and openness, as well
as for evaluating this work.
I gratefully acknowledge M. Huss for providing the stake measurements from the Plaine Morte glacier,
as well as the detailed mass balance data. Furthermore, other data were thankfully provided by the
FOEN (discharge data), Maurice Perraudin from Lienne SA (discharge data), Meteoswiss
(meteorological data) and SLF (meteorological data). All people involved in the tracer experiment
organization, sampling, etc… are also deeply acknowledged.
The CH2011 data were obtained from the Center for Climate Systems Modeling (C2SM).
I am particularly grateful to my mentor Bruno Schädler, who always followed me through the wide
meanders of the thesis, providing me the right instruments for this long journey, good advices and a
lot of optimism.
He and my supervisor Rolf Weingartner always supported and trusted me, even in the most uncertain
and in the darkest moments, and this is something I will never forget.
A big and fat thank you goes to my colleagues and friends, for their help in every field of life, their
love, their patience and for sharing their lifes and their interests with me, you made my days. In
particular I want to thank my sister, Martìn, Giovanna, Nina, Jan, Ole, Anne Catherine, Raffi, Yuri,
Judith, Michl, my parents, Ioan and Alex, who each in its own way helped me to find my way.
Table of contents
I 2 MontanAqua ...................................................................................................................................................... 6
I 4 Structure of the thesis ....................................................................................................................................... 9
II CONTEXT AND DATA ....................................................................................................... 11 II 1 Study region .................................................................................................................................................... 12
II 1.1 Glacier, karst and tracer experiments .................................................................................................... 14
II 1.2 Case study: the Tseuzier catchment ....................................................................................................... 19
II 1.3 Data ......................................................................................................................................................... 22
II 1.3.2 Available meteorological and hydrological data .............................................................................. 24
II 2 Temporary measurement network ................................................................................................................ 28
II 2.2 Data ......................................................................................................................................................... 28
II 3 Climatic data – the basics ............................................................................................................................... 36
II 3.1 Projections for climate change – CH2011 .............................................................................................. 36
II 3.2 Signals for different time periods ........................................................................................................... 38
II 3.2.1 Main climatological variables: precipitation and temperature ....................................................... 38
II 3.2.2 Discharge .......................................................................................................................................... 44
III METHODS ......................................................................................................... 47 III 1 Physically based hydrological framework ..................................................................................................... 48
III 1.1 Model setup .......................................................................................................................................... 49
III 1.2 Input data preprocessing ....................................................................................................................... 54
III 1.2.1 Parameter setting ............................................................................................................................ 54
III 1.2.2 Incorporate topographic effects on the energy input .................................................................... 57
III 1.2.3 Precipitation: necessary adjustments ............................................................................................. 59
III 2 Cryosphere and high resolution precipitation .............................................................................................. 63
Table of contents
III 2.2 Increasing complexity of the modelling framework .............................................................................. 63
III 2.2.1 Cryosphere: varying albedo and snow redistribution ..................................................................... 64
III 2.2.2 Precipitation: data preprocessing and interpolation ...................................................................... 66
III 3 Sensitivity and climate change ...................................................................................................................... 69
III 3.1 Sensitivity ............................................................................................................................................... 70
III 3.2 Climate change ....................................................................................................................................... 71
IV RESULTS ............................................................................................................ 77 IV 1 Application of the developed physically based hydrological framework: the Tseuzier case study and
plausibility of the results ...................................................................................................................................... 78
IV 1.1 Plausibility: a multicriterial validation ................................................................................................... 78
IV 1.1.1 Snow and ice .................................................................................................................................... 78
IV 1.1.2 Evapotranspiration .......................................................................................................................... 82
IV 1.2 Results: Tseuzier case study .................................................................................................................. 84
IV 1.2.1 Simulation results ............................................................................................................................ 84
IV 1.2.1.1 External contribution of karstified area added separately .................................................... 84
IV 1.2.1.2 External contribution of karstified area added as a source at the Loquesse spring .............. 88
IV 1.2.2 Performance .................................................................................................................................... 89
IV 2 Adressing key elements of mountain hydrology in a data sparse alpine environment: the assets and
drawbacks of an increased complexity ................................................................................................................ 91
IV 2.1 Plausibility: a multicriterial validation ................................................................................................... 91
IV 2.1.1 Snow and ice .................................................................................................................................... 91
IV 2.1.2 Water balance ................................................................................................................................. 94
IV 2.2 Results ............................................................................................................................................ 94
IV 2.2.2 Performance .................................................................................................................................... 96
IV 3.1 Spatial distribution of available water resources .................................................................................. 98
IV 3.1.1 Tièche .............................................................................................................................................. 98
IV 3.1.2 Ertentse ......................................................................................................................................... 103
IV 3.1.3 Vatseret ......................................................................................................................................... 104
IV 3.2.4 Boverèche ...................................................................................................................................... 106
IV 3.1.3 Headwaters ................................................................................................................................... 108
IV 4 Sensitivity and climate change .................................................................................................................... 109
IV 4.1 Sensitivity analysis: effects of different settings and different time frames ...................................... 109
IV 4.2 Climate change impact assessment for the headwaters of the study region .................................... 113
IV 4.2.1 Effects of climate change on snow cover ..................................................................................... 113
IV 4.2.2 Effects of climate change on the hydrological cycle and on the water availability ..................... 117
IV 4.2.3 Effects of climate change on the water balance ........................................................................... 119
IV 5 Discussion .................................................................................................................................................... 121
A.1 Sources ..................................................................................................................................................... 146
A.2 Evaporation .............................................................................................................................................. 146
B.2 The effect of inclination and exposition .................................................................................................. 150
C Snow and icemelt parameters ....................................................................................................................... 152
D Land cover, soil and geology parameters ....................................................................................................... 153
E Performance: Indexes ...................................................................................................................................... 155
F Discharge measurements ................................................................................................................................ 157
Figures
ix
Figures Fig. 1: Schema of Working package 1 (WP1): the availability of natural water resources. ....................................... 7
Fig. 2: Study area with catchments’ and subcatchments’ boundaries in black, main rivers and springs ................ 13
Fig. 3: Uranine colouring the water coming from the Loquesse spring on 4 September 2012 ............................... 18
Fig. 4: Study area with measuring stations, catchments’ and subcatchments’ boundaries. ................................... 20
Fig. 5: Mean monthly cycle of springs’ discharge (Lourantse, Loquesse and the sum of these two) for the period
between October 1976 and September 1981 compared with the discharge recorded at the Lie110 gauging
station .............................................................................................................................................................. 26
Fig. 6: Mean annual cycle of the relationship between discharge recorded at the Lie110 gauging station and the
discharge estimated by Lienne SA at the Tseuzier lake dam .......................................................................... 28
Fig. 7: Map of the study area showing the boundaries of the case study catchment Tseuzier, land use, the river
network and the position of the measuring stations ...................................................................................... 29
Fig. 8: Overview of installed gauging stations, together with the main springs present in the area, and when
available the corresponding mean specific discharge Qspec [mm/d], as estimated by Crestin (2001). ........ 33
Fig. 9: Annual cycle of Delta T, and Delta P for the scenario period 20212050 (left) and 20702099 (right) at the
station MVE, as provided by CH2011. ............................................................................................................. 38
Fig. 10: Monthly cycle of precipitation P at MVE for the period 20072012, represented as the median of the
time series together with the interquartile range (between 25 and 75% quartile); envelope of the minimum
and maximum monthly quartiles as well as the interquartile range computed using all possible consecutive
6 years blocks between 1980 and 2009 .......................................................................................................... 39
Fig. 11: (above) Monthly cycle of precipitation P at MVE for the two periods 19802009 and 20072012, with the
medians of the time series, their interquartile range (between 25 and 75% quartile), and their envelope
representing the minimum and maximum monthly precipitation; (below) the same as above but for
temperature T. ................................................................................................................................................. 41
Fig. 12: Seasonal anomalies of temperature against precipitation for winter – between November and April –and
for summer – between May and October – .................................................................................................... 42
Fig. 13: Annual (hydrological year between October of the previous year and September of the year of interest)
anomalies of temperature against precipitation. ............................................................................................ 44
Fig. 14: (above) Monthly cycle of measured discharge Q of the Tseuzier catchment for the two periods 1980
2009 and 19742012, with the medians of the time series, their interquartile range (between 25 and 75%
quartile), and their envelope representing the minimum and maximum monthly discharge; (below) the
same as above but the second time series covering the period 20072012. ................................................. 45
Fig. 15: Overview of input data required to run a simulation of PIHM. ................................................................... 55
Fig. 16: Mean daily incoming clear sky solar radiation in the main Tseuzier subcatchment for the four seasons:
winter (DJF), spring (MAM), summer (JJA) and autumn (SON). ...................................................................... 58
Fig. 17: Transverse profile in the Tseu_Lie subcatchment of the shading factor, defined as the ratio between
potential incoming solar radiation on a flat surface at the same location and the effectively incoming solar
radiation, on 4 different dates during the year. .............................................................................................. 59
Fig. 18: Plot of annual precipitation against elevation for precipitation used in the case study between 1975 and
1982 as well as between 2007 and 2012 with 5 to 95% range of the Meteoswiss corrected data; for the
longterm precipitation climatology by Kirchhofer and Sevruk (1992) 19511980 as well as for the longterm
precipitation climatology by Schwarb et al.(2011) 19711990; and data recorded at the totalizer WEH ..... 62
Fig. 19: Daily interpolated precipitation on the 17th Ocober 2010 and on the 10th October 2011,as well as their
standardized values ......................................................................................................................................... 68
Fig. 20: (above) Monthly cycle of precipitation P at MVE for the two periods present 20072012 and near future
20482053, with the medians of the time series, their interquartile range (between 25 and 75% quartile),
and their envelope representing the minimum and maximum monthly precipitation; (below) the same as
above but for temperature T ........................................................................................................................... 73
Figures
x
Fig. 21: (above) Monthly cycle of precipitation P at MVE for the two periods present 20072012 and far future
20972102, with the medians of the time series, their interquartile range (between 25 and 75% quartile),
and their envelope representing the minimum and maximum monthly precipitation; (below) the same as
above but for temperature T ........................................................................................................................... 74
Fig. 22: Tseuzier catchment, the extension of it across the hydrographic boundaries due to the Karst drainage
system, Plaine Morte glacier in 3D and projected in 2D. ................................................................................ 78
Fig. 23: Seasonal snow cover evolution between 2007 and 2012: at the VDS2 snow station as measured with a
ultrasonic sensor, as modelled by using precipitation recorded at the station, as modelled by using the
modified grid precipitation of Meteoswiss, and on element 452 ................................................................... 79
Fig. 24: Calculated melt water runoff from Glacier de la Plaine Morte for the hydrological years 2009/2010–
2011/2012 and reconstruction by Huss et al.(2013) based on stake measurements .................................... 81
Fig. 25: Daily observed and simulated runoff between 1975 and 1982, modelled daily ablation with the fractions
of ice melt and snowmelt and precipitation .................................................................................................... 85
Fig. 26: Monthly observed and simulated runoff between 1975 and 1982, modelled monthly ablation with the
fractions of ice melt and snowmelt, monthly observed and simulated discharge from the Loquesse source
and precipitation .............................................................................................................................................. 86
Fig. 27: Measured daily discharge at Lie110, simulated in the river element at the location of Lie110, simulated in
the river element at the location of Lie110 and adding the simulated contribution from the external karstic
area until September, simulated in the river element at the location of Lie110 and adding the simulated
contribution from the external karstic area until October in 2011 (first panel); the same but in 2012 (second
panel); the same but between 12 July 2012 and 25 October 2012 and with additionally daily discharge
simulated in the river element downstream of the Loquesse spring, measured just downstream of the
Loquesse spring, simulated in the river element at the location of Lou, measured at Lou and simulated
contribution from the external karstic area until September (third and fourth panel) .................................. 87
Fig. 28: Daily observed and simulated runoff between 1975 and 1982 in the Tseuzier catchment, with three
different simulations: once simply adding the monthly measured contribution from the two springs
Lourantse and Loquesse as an external source when simulating Tseu_Lie, once adding the simulated
external contribution from the karstic area as an external source when simulating Tseu_Lie and once
summing separately – or subsequentely – the simulated external contribution from the karstic area with
the streamflow simulated at the outlet of Tseu_Lie. ...................................................................................... 89
Fig. 29: Flow duration curves of observed and simulated streamflow for the period 19751982 and for the period
20072012. ....................................................................................................................................................... 90
Fig. 30: (above) Seasonal snow cover evolution between 2007 and 2012: at the VDS2 snow station as measured,
and as simulated with a fix albedo, as well as with a varying albedousing different precipitation data;
(below) calculated melt water runoff from Glacier de la Plaine Morte for the hydrological years 2009/2010–
2011/2012 applying a fix albedo as well as applying a varying albedo, and reconstruction by Huss et
al.(2013) based on stake measurements ......................................................................................................... 92
Fig. 31: Daily observed and simulated runoff, modelled daily ablation with the fractions of ice melt and snowmelt
and precipitation in the Tseuzier catchment between 2010 and 2012, using different model settings and
different meteorological data.. ........................................................................................................................ 95
Fig. 32: Flow duration curves of observed and simulated streamflow for the period 1.10.200731.12.2012 using
different model settings and different meteorological data ........................................................................... 97
Fig. 33: Runoff simulated at the outlet of Tièche between 2007 and 2012, and discharge measured at Tie100 in
the same period ............................................................................................................................................... 99
Fig. 34: Runoff simulated at the outlet of Tièche in 2008 and 2011, and discharge measured at Tie100 in the
same period ................................................................................................................................................... 100
Fig. 35: Runoff simulated at the outlet of Tièche in 2012 using set ups ; and discharge measured at Tie100 in the
same period ................................................................................................................................................... 101
Fig. 36: Daily simulated runoff at the outlet of Ertentse between 1975 and 1980, and measured discharge
between 1956 and 1961. ............................................................................................................................... 103
Fig. 37: Monthly discharge between October 1975 and October 1981: measured on the Ertentse, simulated at
the outlet of the Ertentse subcatchment, measured at Lie110 and the sum of the measured outflow at the
Lourantse and Loquesse springs. ................................................................................................................... 104
Fig. 38: Monthly discharge in the Vatseret subcatchment between October 1975 and October 1979; and
between January 2007 and December 2012. ................................................................................................ 105
Fig. 39: Mean monthly annual cycle between October and September (hydrological year) of outflow from: the
sum of the sources MOL38 and MOL9, the RAN 1 source and the RAN 28 source, and with additionally
simulated streamflow at the outlet of the Boverèche subcatchment .......................................................... 107
Fig. 40: Hourly discharge of the Boverèche measured at Colombire Oct.2010Sept.2011. .................................. 108
Fig. 41: Daily simulated discharge of the Tseu_Lie subcatchment between 2007 and 2012 using different
settings: applying a varying soil depth, adding the gravitational redistribution of snow and changing the
description of albedo from a fix albedo to a varying albedo ......................................................................... 110
Fig. 42: (above) Annual cycle of simulated discharge of the Tseu_Lie subcatchment between 2007 and 2012
using different settings: applying a varying soil depth, adding the gravitational redistribution of snow,
changing the description of albedo from a fix albedo to a varying albedo, and additionally with the
projections for the near future; (below) annual cycle of simulated discharge in the Tseuzier catchment in
different years and as the mean over different periods: for the period 19751982, for the corresponding
projection in the near future 20162023, for the corresponding projection in the far future 20652072 and
for the same period but assuming clogging on the glacier, in 1976, in 1980 and in 2011 ........................... 112
Fig. 43: Simulated snow cover in the Tseu_Lie subcatchment for the three periods present (20072012), near
future (20482053) and far future (20972102) ............................................................................................ 115
Fig. 44: Spread of the simulated snow cover in the Tseu_Lie subcatchment for the two periods near future
(20482053) and far future (20972102) ....................................................................................................... 116
Fig. 45: Monthly cycle of discharge Q of the headwaters of the study area, as well as of each of its
subcatchments for the three periods present (20072012), near future (20482053) and far future (2097
2102), with the medians of the time series, their interquartile range (between 25 and 75% quartile), and
their envelope representing the minimum and maximum monthly precipitation runoff ............................ 118
Fig. 46: Monthly cycle of the components of the water balance for the headwaters of the study area in the three
periods present (20072012), near future (20482053) and far future (20972102). .................................. 120
Fig. 47: Monthly coefficient of determination for different extrapolations of solar radiation. ............................ 157
Fig. 48: Conversion factor for direct radiation for a southwest exposed surface with 25° slope, and for a north
west exposed surface with 25° slope at the same location of the Montana station.................................... 157
Fig. 49: Global solar radiation measured at the Montana station: measured and computed with GRASS .......... 157
Fig. 50: View of the cross section of the Lie110 gauging station ........................................................................... 157
Fig. 51: Water level measured at Loquesse against water level measured at Lie110 ........................................... 159
Fig. 52: Stagedischarge relationship at the Lie110 gauging station ...................................................................... 160
Fig. 53: Picture of the Loquesse spring the day of the installation of the gauging station (24 July 2012), as seen
from the station’s site; and picture of the installed water pressure sensor ................................................. 161
Fig. 54: Difference in discharge between Lie110 and Lou as a function of water level recorded at Loq .............. 162
Fig. 55: Stagedischarge relationship at the Lourantse gauging station for the period between 12.7.2012 and
25.10.2012. .................................................................................................................................................... 163
Fig. 56: Picture of the gauging station Erte_2011 on 11 October 2011 ................................................................ 164
Fig. 57: Stagedischarge relationship at the Erte_2011 gauging station for the period between 27.7.2011 and
25.9.2011. ...................................................................................................................................................... 164
Fig. 58: Picture of the installed water pressure sensor for Erte_2012 and overview of the location of the
Erte_2012 gauging station. ............................................................................................................................ 165
Fig. 59: Stagedischarge relationship at the Erte_2012 gauging station for the period between 25.7.2012 and
31.12.2012. .................................................................................................................................................... 165
Fig. 60: View of the cross section of the Tie100 gauging stations ......................................................................... 166
Fig. 61: Picture of the Tie100 gauging station. ....................................................................................................... 166
Fig. 62: Stagedischarge relationships at the Tie100 gauging station .................................................................... 169
Fig. 63: (above) Monthly cycle of precipitation P at MVE for the two periods 19312012 and 19802009, with the
medians of the time series, their interquartile range (between 25 and 75% quartile), and their envelope
representing the minimum and maximum monthly precipitation; (below) the same as above but for
temperature T. ............................................................................................................................................... 170
Tables
xiii
Tables Tab. 1: Headwaters’subcatchments description ...................................................................................................... 14
Tab. 2: Tseuzier’s subcatchments description .......................................................................................................... 21
Tab. 3: List of meteorological and gauging stations already present in the study area .......................................... 25
Tab. 4: List of meteorological, snow and rain gauging stations in the study area available 20072012 ................. 30
Tab. 5: List of gauging stations installed in the study area in the period 20112012. ............................................. 34
Tab. 6: Parameters values defined for the simulation of snow and icemelt with a varying albedo ...................... 65
Tab. 7: Measured seasonal mass balance of Glacier de la Plaine Morte for the hydrological years 2009/2010–
2011/2012 [m w.e.] (Huss et al. 2013) compared with the simulated melt. .................................................. 82
Tab. 8: Mean annual simulated water balance components in the Tseuzier catchment for the two simulation
periods 19751982 and 20072012 in [mm].................................................................................................... 83
Tab. 9: Performance indexes for the two simulation periods 19751982 and 20072012...................................... 90
Tab. 10: Measured seasonal mass balance of Glacier de la Plaine Morte for the hydrological years 2009/2010–
2011/2012 (Huss et al. 2013) compared to the seasonal mass balance simulated applying a fix albedo
(SIMalbfix) as well as a varying albedo (SIMalbvar) in [m w.e.]. ..................................................................... 93
Tab. 11: Mean annual simulated water balance components in the Tseuzier catchment between 2010 and 2012:
using different model settings and different meteorological data. ................................................................ 94
Tab. 12: Performance indexes for different model set ups and input data ............................................................. 97
Tab. 13: Mean annual simulated water balance components between 2007 and 2012 of the Tseuzier basin and
of the headwaters of the study area. ............................................................................................................ 109
Tab. 14: Overview of the water availability in the headwaters of the study area for different periods ............... 119
Tab. 15: Mean annual simulated water balance components of the headwaters of the study area for the three
periods present (20072012), near future (20482053) and far future (209720102). ................................ 121
Tab. 16: Monthly regression factors for the period 1.1.1975 31.12.1980 for computing global radiation at the
MVE station .................................................................................................................................................... 148
Tab. 17: Monthly regression factors for the gap period between 28.2.197731.5.1977 for computing global
radiation at the MVE station using data at the SIO station ........................................................................... 149
Tab. 18: Parameters values defined for the simulation of snow and icemelt with a fix albedo .......................... 152
Tab. 19: Landcover and topsoil parameters. .......................................................................................................... 153
GHG Greenhouse gas
FOEN Swiss Federal Office of the Environment (BAFU in german: Bundesamt für Umwelt)
Lie SA Lienne SA hydropower production company
Loq Loquesse spring
Lou Lourantse spring
NRP61 National Research Programme "Sustainable Water Management"
PIHM Penn State Integrated Hydrologic Model
RCM Regional climate model
2
I 1 Object of research In the last decades with the awareness of a changing and evolving environment, the number of studies
on how the climate and human activities affect the natural systems and cycles has constantly risen,
with the spillover effect to stimulate the development of more sophisticated models able to describe
the system processes, and improving predictions (Silberstein 2006; Liu and Gupta 2007). In addition,
efforts have been made in order to enhance the spatial and temporal scale at which predictions are
made. However the support of these developments by the increasing power of computers was neither
accompanied by the same significant increase in the availability of data, nor in the quality of the
measured data (Drécourt 2004a; Silberstein 2006).
Particularly in conjuction with the changes expected to happen to the locally available water resources
resulting from climate change, an increasing number of regional climate change impact assessment
studies have been launched the last decade, with the increasing awareness that global sustainability is
made of local/regional sustainability, and that resources management and natural variability are
tightly coupled and interact. As matter of fact stakeholders, managers and politicians need to be
informed and included in such studies, as we need them to be able to undertake measures and make
decisions on adaptation and mitigation strategies for the future (Reynard et al. 2014; Schneider et al.
2014). For this purpose, the establishment of a robust and reliable modelling framework is required.
Hydrological or watershed models are crucial, as they serve here as exploratory and predictive tools.
Usually to be able to adequately address questions about the past, present and future status of an
environment it would be appropriate to focus efforts to monitor and anticipate changes and have the
means to provide a historical context for the measurements. Yet, climate and hydrological monitoring
in mountain areas are known to be difficult and challenging tasks, as besides the tough environment
conditions to which measuring instruments are exposed, these remote areas require major efforts to
visit, maintain and keep the measurements ongoing (Diaz 2005). An other fundamental problem is
that often many of the equations used to represent processes occurring in the hydrological cycle
require calibration, thus the parameters involved cannot be directly measured, or they are invariably
applied at a scale different to that at which they were derived (Grayson and Blöschl 2000), and this is
even more true in alpine areas. Hence, in such regions since usually available observations are
discontinuous in space and time, and furthermore do not provide sufficient information about the
detailed processes that are represented by the model, it is often of practical impossibility to calibrate
it properly for any time and spatial scales.
Generally it could be said that the inclusion of more processes and/or controlling variables in the
system can only be justified on the basis that the inclusion of additional controlling mechanisms
should both improve predictive skill and facilitate the estimation of parameter values on the basis of
physiological characteristics or measurements (Montaldo et al. 2007). On the other hand, especially
for impact studies it is quintessential to keep the physical basis in the description of the dynamics, i.e.
more complex and detailed, as it assures a consistent reproduction of the behaviour of the system.
The higher the degree of conceptualization, the higher is the danger this would lead to a model that
mimics the system without understanding it.
At this point it is clear that the choice of an appropriate model is a demanding task, requiring good
diplomatic skills: the tradeoffs between parsimony, complexity and robustness should be tackled
identifying the optimum between data availability, model complexity and predictive performance. It
appears that in this sense an implicit requirement is the model to be flexible, i.e. extensible and
I INTRODUCTION I 1 Object of research
adaptable to the given circumstances.
For most of the countries around the world basic digital geospatial data such as a digital elevation
model (DEM), soil, geology and landuse maps are actually available, with varying resolution and
precision. They allow a topographic as well as a physiographic characterization of the environment,
whose features can be described with attributes. If sufficiently accurate these attributes have a great
potential, and regardless of being quantitative or qualitative, are viewed as relevant and
discriminatory indicators for processes (Pflaunder 2001). These data, as well as any other source of
information like studies carried out within or close to the study area or literature should be combined
and exploited in order to allow the implementation of a physically based model, despite the possible
scarcity of data and observations on site. Maybe one or some of the processes might need some
degree of simplification, in which case adjustments of the parameters will be needed, allowing
tailoring the model to the specific behaviour of the studied system. Automatic methods for parameter
adjustment seek to take advantage of the speed and power of digital computers, while being objective
and relatively easy to implement. In contrast, the trial and error method (manual approach), which has
been developed and refined over the years to result in excellent model calibration, is complicated and
highly laborintensive, and the expertise acquired by the modeller is not easily transferred (Boyle et al.
2000). However here this limitation is not considered decisive in carrying out the modelling task, as
this configuration is still considered representative of the best process understanding achievable from
available data and catchment knowledge (Konz et al. 2010).
All in all, the use of a procedure including manual calibration and commonly available data appears
particularly promising, as it offers the possibility to rely almost entirely on the available data, exploit
the hydrological knowledge of experts and transfer the model settings established in subbasins with
relatively good data to other ungauged basins. Such applications suggest that the model can be
regarded as a very powerful tool for monitoring water resources: it serves as an interpolator at
locations where it is practically impossible to observe the necessary information (Drécourt 2004a).
Of course, all of this envisages the availability of (at least) one subbasin where the model settings can
be verified either through direct measurements, or indirectly through some kind of plausibility checks.
In order to judge model’s predictive performance meaningful criteria need to be chosen. Spatio
temporal dynamics as well as spatial fields of instantaneous and timeintegrated hydrological
variables, such as evapotranspiration, soil moisture, channel discharge, or more typical and
characteristic for an alpine environment such as snowpack and snow melt, are adequate variables to
make such an evaluation. The quality and confidence of these different intermediate results,
respectively measurements, must be carefully appraised, because of course data can be corrupted by
different types of error. Uncertainties might be present in the forcing terms, in the measurements
themselves as well as in the spatial (or eventually temporal) extrapolation of these, in the model
structure and parametrization. Moreover scaling uncertainties arise from differences in the
discretization of the model, in the description of the physics behind this and finally from the
observations, which are usually carried out at a precise point location (Melching et al. 1990). Still,
usually the accuracy of at least some of these data is good enough to represent a precious source of
information, enabling to evaluate reasonably well the outcomes of the applied modelling chain.
Montanari and Di Baldassarre showed that if measurements are made following stateoftheart
techniques, observation uncertainty has a limited impact, with respect to model structural
uncertainty, on the results of hydrological models (Montanari and Di Baldassarre 2013). Further they
I INTRODUCTION I 1 Object of research
4
showed that particular care should be taken in discarding measurements, as in hydrological modelling
any information is important and the presence of data errors does not necessarily limit the usefulness
of observed records, from what it follows that an appropriate selection of hydrological complexity and
calibration strategy can increase the robustness of hydrological applications against data errors
(Montanari and Di Baldassarre 2013).
The hydrological cycle in alpine environments is to a large extent controlled by snow accumulation,
storage, redistribution, and melting (Parajka et al. 2012; Warscher et al. 2013). High altitudinal
gradients, a strong variability of meteorological variables in time and space, usually only locally
quantified snow cover dynamics, complex and often unknown hydrogeological settings, and
heterogeneous land use and soil formations result in high uncertainties in the quantification of the
water balance and the prediction of discharge rates (Warscher et al. 2013). However, despite these
difficulties, hydrological modeling systems are needed and applied to serve and support decision
making in water management. This is particularly the case in mountainous regions, which play a crucial
role as the “water towers” feeding downstream areas (Viviroli et al. 2007). The more the processes
occurring at these high elevations are simplified and conceptualized within a model, the more they
suffer from a lack of physical relevance and physical parameter interpretability (Drécourt 2004b; Clark
and Vrugt 2006). This implies that their predictability for new climate or environmental boundary
conditions might be restricted and not representative. Therefore increasingly complex physically
based models are applied. This may enable a more comprehensive and enhanced perspective of the
sensitivity and the effects of climate change on the water balance, including the consideration of
feedback processes on the different components of the hydrological cycle (for example the effects of
snow albedo on snow cover pattern, and ultimately on runoff generation (Jost et al. 2012; Pellicciotti
et al. 2012)). Furthermore, internal inconsistencies, such as an underestimation of precipitation input
that can be compensated for by an overestimation of meltwater (Konz and Seibert 2010; Pellicciotti et
al. 2012), might be reduced or avoided. However, while the hydrological predictability might increase
with complex models, data demands of course increase as well. A parallel evaluation of these two
issues, increased complexity and increased data availability, should help us to evidence, wheter we are
getting the right answers for the right reasons. In a time of local and global change in the water cycle,
when practical hydrological applications are increasingly used for impact studies and risk analysis this
is crucial.
During the past decades the Alpine climate has been subject to pronounced decadalscale variability,
but also to distinctive longterm trends consistent with the global climate response to increasing
greenhouse gas (GHG) concentrations (Gobiet et al. 2014). In the last 100 years the average annual
temperature in Switzerland has risen by more than 1.5° C (FOEN 2012). A trend analysis of 1959–2008
gridded Swiss temperatures showed that the seasonal trends are all positive and highly significant,
with an average annual warming rate of 0.35°C/decade (Ceppi et al. 2012). Spatial and temporal
variability are pronounced on a seasonal scale, however they clearly identified an anomalouslystrong
warming at low elevations in autumn and early winter and aboveaverage spring temperature trends
at elevations close to the snowline (Ceppi et al. 2012). Warming in Switzerland , particularly
pronounced from 1980 onwards, appears to be about twice that of the global average (FOEN 2012;
Gobiet et al. 2014), which may be explained in part by the differences in physical characteristics of
land and sea surfaces and is mainly caused by water vapour enhanced greenhouse warming (FOEN
2012; Philippona 2013). Furthermore, large areas in the northern hemisphere, and in particular the
Alps, are permanently or during prolonged periods covered with ice and snow. These areas are getting
I INTRODUCTION I 1 Object of research
5
smaller, meaning there is a larger dark surface area and
Universität Bern
Universität Bern
Von der Philosophischnaturwissenschaftlichen Fakultät angenommen.
iii
Summary This study presents a robust modelling framework developed for assessing the available water
resources in a mountainous environment today and in the future. The semidiscrete, physicallybased
Penn State Integrated Hydrologic Model (PIHM) has been identified as a suitable hydrological model
for this challenging task. Here we present the customization, the enhancement and the application of
this on a catchment with sparse data in the past and in the present, we evaluate extensively and
thoroughly its performance, its potential and its limitations. The melt modules for snow and ice were
upgraded from a simple temperatureindex model to a model including the influence of global
radiation. Stationary catchment attributes such as soil and land cover data were used to distribute
parameters, and the latters were mainly estimated basing on literature or experiments on site. Besides
the examination of the internal consistency with a multicriteria validation, evaluating snowpack,
icemelt and different components of the water balance, model results have been further validated
with discharge measurements. Results show a very good performance of the model over different
spatial and temporal scales.
Sound hydrological modeling frameworks are needed to serve and support decision making in local to
regional water management, in particular in mountainous regions, as main suppliers of natural water
resources. In view of climate change, increasingly complex physically based models are applied in
order to guarantee the predictability for new climate or environmental boundary conditions. However,
while the hydrological predictability might increase with complex models, data demands of course
increases too.
In the present study we investigate the challenges, the gains and the problems of an increased
complexity in describing snow and precipitation patterns in data sparse regions at higher altitudes and
in an environment with a highly complex topography. Particular efforts were put in improving the
modelling of snow and glacier dynamics and melt including a timevarying albedo and gravitational
redistribution of snow in a hydrological model. Furthermore, an extensive meteorological measuring
network was set up and exploited generating high resolution input data.
We provide evidence of the limits of an enhanced description of the albedo of snow depending on the
surrounding conditions, when other processes we cannot describe emerge. We also show the
importance of the maintenance and extension of shortterm monitoring networks in mountain
regions, demonstrating that an increased sampling of hydrometeorological and cryospheric data –
even on the basis of one to three years – retrieves the best model performance, and therefore these
are crucial for the improvement of the hydrological knowledge of the local system.
With the developed modelling framework we estimated available water resources for the headwaters
of the study area, and made an overall plausibility assessment using all the available data, with very
good results. Overall the headwaters of the catchment provide 110 mio m3. Generally the western
part of the catchment is richer in water resources than the eastern part, corresponding respectively to
80% and 20% of the total water resources. Tseuzier, the subbasin at the western boundary, represents
the “water tower” of the area, supplying 70% of the total estimated water amount
In this work we integrated many relevant aspects of hydrological modelling, especially pertinent to the
context of climate change in alpine regions, like transferability in time and space, as well as flexibility,
and proved our modelling framework to be robust over different scales. As such, it provides a solid
Summary
iv
basis for hydrological analysis of the system under changing forcing conditions, like in the case of
climate change.
We performed an exploratory climate sensitivity analysis, basing on the data of ten climate model
chains driven by the single emission scenario A1B, postprocessed and interpolated to the MeteoSwiss
weather station Montana (as provided by the CH2011 initiative). Two scenario periods in the 21st
century were assessed relative to the short reference period 20072012: the near future 20482053
and the far future 20972102. The most pronounced changes are expected in the snow cover
dynamics in the headwaters of the study area, and evaporation rates in the lower parts of the study
area, with changes in temperature as the main drivers. Basing on the applied scenarios, for all models,
a reshaped annual cycle, with earlier rise of spring runoff, significantly reduced summer runoff, and a
tendency for increased winter runoff in the first half of the 21st century, becoming increasingly
pronounced towards the end of the century, and resulting in a significant change of regime: first
transforming from a bglacionival regime in a nival alpin regime as a transition before becoming,
characterized by a nival de transition regime in the far future.
All in all, the overall available water resources are going to be affected only to a minor extent, with
mean changes in the order of magnitude of 510%, however the major changes estimated for the
summer runoff, reduced by about 20% in the near future, and almost 50% in the far future are
expected to have major consequences for the water management strategies of the region. In
particular in dry years conflicts might arise between different usages, which here are mainly irrigation,
hydropower production, tourism and artificial snow making.
Ackowledgements
“I am neither the first commentator, nor the most luminous. Therefore I will
make no particular claim of originality for the remarks presented here, but can
only hope that they are framed in useful ways”
J.W. Kirchner
Ackowledgements This study is embedded in the framework of the MontanAqua project. MontanAqua is financed by the
NRP 61 (National Research Project 61, Sustainable Water Management). I would like to thank the
Swiss National Science Foundation for funding the MontanAqua research project (4061401259646 1)
within the National Research Program “Sustainable Water Management” (NRP 61).
I would like to thank Gopal Bhatt and Xuan Yu at the Penn State University for their patience, their
help for the customization of the model (developed by their research group), and their support
throughout the first simulations. I also thank Christopher Duffy for his support and openness, as well
as for evaluating this work.
I gratefully acknowledge M. Huss for providing the stake measurements from the Plaine Morte glacier,
as well as the detailed mass balance data. Furthermore, other data were thankfully provided by the
FOEN (discharge data), Maurice Perraudin from Lienne SA (discharge data), Meteoswiss
(meteorological data) and SLF (meteorological data). All people involved in the tracer experiment
organization, sampling, etc… are also deeply acknowledged.
The CH2011 data were obtained from the Center for Climate Systems Modeling (C2SM).
I am particularly grateful to my mentor Bruno Schädler, who always followed me through the wide
meanders of the thesis, providing me the right instruments for this long journey, good advices and a
lot of optimism.
He and my supervisor Rolf Weingartner always supported and trusted me, even in the most uncertain
and in the darkest moments, and this is something I will never forget.
A big and fat thank you goes to my colleagues and friends, for their help in every field of life, their
love, their patience and for sharing their lifes and their interests with me, you made my days. In
particular I want to thank my sister, Martìn, Giovanna, Nina, Jan, Ole, Anne Catherine, Raffi, Yuri,
Judith, Michl, my parents, Ioan and Alex, who each in its own way helped me to find my way.
Table of contents
I 2 MontanAqua ...................................................................................................................................................... 6
I 4 Structure of the thesis ....................................................................................................................................... 9
II CONTEXT AND DATA ....................................................................................................... 11 II 1 Study region .................................................................................................................................................... 12
II 1.1 Glacier, karst and tracer experiments .................................................................................................... 14
II 1.2 Case study: the Tseuzier catchment ....................................................................................................... 19
II 1.3 Data ......................................................................................................................................................... 22
II 1.3.2 Available meteorological and hydrological data .............................................................................. 24
II 2 Temporary measurement network ................................................................................................................ 28
II 2.2 Data ......................................................................................................................................................... 28
II 3 Climatic data – the basics ............................................................................................................................... 36
II 3.1 Projections for climate change – CH2011 .............................................................................................. 36
II 3.2 Signals for different time periods ........................................................................................................... 38
II 3.2.1 Main climatological variables: precipitation and temperature ....................................................... 38
II 3.2.2 Discharge .......................................................................................................................................... 44
III METHODS ......................................................................................................... 47 III 1 Physically based hydrological framework ..................................................................................................... 48
III 1.1 Model setup .......................................................................................................................................... 49
III 1.2 Input data preprocessing ....................................................................................................................... 54
III 1.2.1 Parameter setting ............................................................................................................................ 54
III 1.2.2 Incorporate topographic effects on the energy input .................................................................... 57
III 1.2.3 Precipitation: necessary adjustments ............................................................................................. 59
III 2 Cryosphere and high resolution precipitation .............................................................................................. 63
Table of contents
III 2.2 Increasing complexity of the modelling framework .............................................................................. 63
III 2.2.1 Cryosphere: varying albedo and snow redistribution ..................................................................... 64
III 2.2.2 Precipitation: data preprocessing and interpolation ...................................................................... 66
III 3 Sensitivity and climate change ...................................................................................................................... 69
III 3.1 Sensitivity ............................................................................................................................................... 70
III 3.2 Climate change ....................................................................................................................................... 71
IV RESULTS ............................................................................................................ 77 IV 1 Application of the developed physically based hydrological framework: the Tseuzier case study and
plausibility of the results ...................................................................................................................................... 78
IV 1.1 Plausibility: a multicriterial validation ................................................................................................... 78
IV 1.1.1 Snow and ice .................................................................................................................................... 78
IV 1.1.2 Evapotranspiration .......................................................................................................................... 82
IV 1.2 Results: Tseuzier case study .................................................................................................................. 84
IV 1.2.1 Simulation results ............................................................................................................................ 84
IV 1.2.1.1 External contribution of karstified area added separately .................................................... 84
IV 1.2.1.2 External contribution of karstified area added as a source at the Loquesse spring .............. 88
IV 1.2.2 Performance .................................................................................................................................... 89
IV 2 Adressing key elements of mountain hydrology in a data sparse alpine environment: the assets and
drawbacks of an increased complexity ................................................................................................................ 91
IV 2.1 Plausibility: a multicriterial validation ................................................................................................... 91
IV 2.1.1 Snow and ice .................................................................................................................................... 91
IV 2.1.2 Water balance ................................................................................................................................. 94
IV 2.2 Results ............................................................................................................................................ 94
IV 2.2.2 Performance .................................................................................................................................... 96
IV 3.1 Spatial distribution of available water resources .................................................................................. 98
IV 3.1.1 Tièche .............................................................................................................................................. 98
IV 3.1.2 Ertentse ......................................................................................................................................... 103
IV 3.1.3 Vatseret ......................................................................................................................................... 104
IV 3.2.4 Boverèche ...................................................................................................................................... 106
IV 3.1.3 Headwaters ................................................................................................................................... 108
IV 4 Sensitivity and climate change .................................................................................................................... 109
IV 4.1 Sensitivity analysis: effects of different settings and different time frames ...................................... 109
IV 4.2 Climate change impact assessment for the headwaters of the study region .................................... 113
IV 4.2.1 Effects of climate change on snow cover ..................................................................................... 113
IV 4.2.2 Effects of climate change on the hydrological cycle and on the water availability ..................... 117
IV 4.2.3 Effects of climate change on the water balance ........................................................................... 119
IV 5 Discussion .................................................................................................................................................... 121
A.1 Sources ..................................................................................................................................................... 146
A.2 Evaporation .............................................................................................................................................. 146
B.2 The effect of inclination and exposition .................................................................................................. 150
C Snow and icemelt parameters ....................................................................................................................... 152
D Land cover, soil and geology parameters ....................................................................................................... 153
E Performance: Indexes ...................................................................................................................................... 155
F Discharge measurements ................................................................................................................................ 157
Figures
ix
Figures Fig. 1: Schema of Working package 1 (WP1): the availability of natural water resources. ....................................... 7
Fig. 2: Study area with catchments’ and subcatchments’ boundaries in black, main rivers and springs ................ 13
Fig. 3: Uranine colouring the water coming from the Loquesse spring on 4 September 2012 ............................... 18
Fig. 4: Study area with measuring stations, catchments’ and subcatchments’ boundaries. ................................... 20
Fig. 5: Mean monthly cycle of springs’ discharge (Lourantse, Loquesse and the sum of these two) for the period
between October 1976 and September 1981 compared with the discharge recorded at the Lie110 gauging
station .............................................................................................................................................................. 26
Fig. 6: Mean annual cycle of the relationship between discharge recorded at the Lie110 gauging station and the
discharge estimated by Lienne SA at the Tseuzier lake dam .......................................................................... 28
Fig. 7: Map of the study area showing the boundaries of the case study catchment Tseuzier, land use, the river
network and the position of the measuring stations ...................................................................................... 29
Fig. 8: Overview of installed gauging stations, together with the main springs present in the area, and when
available the corresponding mean specific discharge Qspec [mm/d], as estimated by Crestin (2001). ........ 33
Fig. 9: Annual cycle of Delta T, and Delta P for the scenario period 20212050 (left) and 20702099 (right) at the
station MVE, as provided by CH2011. ............................................................................................................. 38
Fig. 10: Monthly cycle of precipitation P at MVE for the period 20072012, represented as the median of the
time series together with the interquartile range (between 25 and 75% quartile); envelope of the minimum
and maximum monthly quartiles as well as the interquartile range computed using all possible consecutive
6 years blocks between 1980 and 2009 .......................................................................................................... 39
Fig. 11: (above) Monthly cycle of precipitation P at MVE for the two periods 19802009 and 20072012, with the
medians of the time series, their interquartile range (between 25 and 75% quartile), and their envelope
representing the minimum and maximum monthly precipitation; (below) the same as above but for
temperature T. ................................................................................................................................................. 41
Fig. 12: Seasonal anomalies of temperature against precipitation for winter – between November and April –and
for summer – between May and October – .................................................................................................... 42
Fig. 13: Annual (hydrological year between October of the previous year and September of the year of interest)
anomalies of temperature against precipitation. ............................................................................................ 44
Fig. 14: (above) Monthly cycle of measured discharge Q of the Tseuzier catchment for the two periods 1980
2009 and 19742012, with the medians of the time series, their interquartile range (between 25 and 75%
quartile), and their envelope representing the minimum and maximum monthly discharge; (below) the
same as above but the second time series covering the period 20072012. ................................................. 45
Fig. 15: Overview of input data required to run a simulation of PIHM. ................................................................... 55
Fig. 16: Mean daily incoming clear sky solar radiation in the main Tseuzier subcatchment for the four seasons:
winter (DJF), spring (MAM), summer (JJA) and autumn (SON). ...................................................................... 58
Fig. 17: Transverse profile in the Tseu_Lie subcatchment of the shading factor, defined as the ratio between
potential incoming solar radiation on a flat surface at the same location and the effectively incoming solar
radiation, on 4 different dates during the year. .............................................................................................. 59
Fig. 18: Plot of annual precipitation against elevation for precipitation used in the case study between 1975 and
1982 as well as between 2007 and 2012 with 5 to 95% range of the Meteoswiss corrected data; for the
longterm precipitation climatology by Kirchhofer and Sevruk (1992) 19511980 as well as for the longterm
precipitation climatology by Schwarb et al.(2011) 19711990; and data recorded at the totalizer WEH ..... 62
Fig. 19: Daily interpolated precipitation on the 17th Ocober 2010 and on the 10th October 2011,as well as their
standardized values ......................................................................................................................................... 68
Fig. 20: (above) Monthly cycle of precipitation P at MVE for the two periods present 20072012 and near future
20482053, with the medians of the time series, their interquartile range (between 25 and 75% quartile),
and their envelope representing the minimum and maximum monthly precipitation; (below) the same as
above but for temperature T ........................................................................................................................... 73
Figures
x
Fig. 21: (above) Monthly cycle of precipitation P at MVE for the two periods present 20072012 and far future
20972102, with the medians of the time series, their interquartile range (between 25 and 75% quartile),
and their envelope representing the minimum and maximum monthly precipitation; (below) the same as
above but for temperature T ........................................................................................................................... 74
Fig. 22: Tseuzier catchment, the extension of it across the hydrographic boundaries due to the Karst drainage
system, Plaine Morte glacier in 3D and projected in 2D. ................................................................................ 78
Fig. 23: Seasonal snow cover evolution between 2007 and 2012: at the VDS2 snow station as measured with a
ultrasonic sensor, as modelled by using precipitation recorded at the station, as modelled by using the
modified grid precipitation of Meteoswiss, and on element 452 ................................................................... 79
Fig. 24: Calculated melt water runoff from Glacier de la Plaine Morte for the hydrological years 2009/2010–
2011/2012 and reconstruction by Huss et al.(2013) based on stake measurements .................................... 81
Fig. 25: Daily observed and simulated runoff between 1975 and 1982, modelled daily ablation with the fractions
of ice melt and snowmelt and precipitation .................................................................................................... 85
Fig. 26: Monthly observed and simulated runoff between 1975 and 1982, modelled monthly ablation with the
fractions of ice melt and snowmelt, monthly observed and simulated discharge from the Loquesse source
and precipitation .............................................................................................................................................. 86
Fig. 27: Measured daily discharge at Lie110, simulated in the river element at the location of Lie110, simulated in
the river element at the location of Lie110 and adding the simulated contribution from the external karstic
area until September, simulated in the river element at the location of Lie110 and adding the simulated
contribution from the external karstic area until October in 2011 (first panel); the same but in 2012 (second
panel); the same but between 12 July 2012 and 25 October 2012 and with additionally daily discharge
simulated in the river element downstream of the Loquesse spring, measured just downstream of the
Loquesse spring, simulated in the river element at the location of Lou, measured at Lou and simulated
contribution from the external karstic area until September (third and fourth panel) .................................. 87
Fig. 28: Daily observed and simulated runoff between 1975 and 1982 in the Tseuzier catchment, with three
different simulations: once simply adding the monthly measured contribution from the two springs
Lourantse and Loquesse as an external source when simulating Tseu_Lie, once adding the simulated
external contribution from the karstic area as an external source when simulating Tseu_Lie and once
summing separately – or subsequentely – the simulated external contribution from the karstic area with
the streamflow simulated at the outlet of Tseu_Lie. ...................................................................................... 89
Fig. 29: Flow duration curves of observed and simulated streamflow for the period 19751982 and for the period
20072012. ....................................................................................................................................................... 90
Fig. 30: (above) Seasonal snow cover evolution between 2007 and 2012: at the VDS2 snow station as measured,
and as simulated with a fix albedo, as well as with a varying albedousing different precipitation data;
(below) calculated melt water runoff from Glacier de la Plaine Morte for the hydrological years 2009/2010–
2011/2012 applying a fix albedo as well as applying a varying albedo, and reconstruction by Huss et
al.(2013) based on stake measurements ......................................................................................................... 92
Fig. 31: Daily observed and simulated runoff, modelled daily ablation with the fractions of ice melt and snowmelt
and precipitation in the Tseuzier catchment between 2010 and 2012, using different model settings and
different meteorological data.. ........................................................................................................................ 95
Fig. 32: Flow duration curves of observed and simulated streamflow for the period 1.10.200731.12.2012 using
different model settings and different meteorological data ........................................................................... 97
Fig. 33: Runoff simulated at the outlet of Tièche between 2007 and 2012, and discharge measured at Tie100 in
the same period ............................................................................................................................................... 99
Fig. 34: Runoff simulated at the outlet of Tièche in 2008 and 2011, and discharge measured at Tie100 in the
same period ................................................................................................................................................... 100
Fig. 35: Runoff simulated at the outlet of Tièche in 2012 using set ups ; and discharge measured at Tie100 in the
same period ................................................................................................................................................... 101
Fig. 36: Daily simulated runoff at the outlet of Ertentse between 1975 and 1980, and measured discharge
between 1956 and 1961. ............................................................................................................................... 103
Fig. 37: Monthly discharge between October 1975 and October 1981: measured on the Ertentse, simulated at
the outlet of the Ertentse subcatchment, measured at Lie110 and the sum of the measured outflow at the
Lourantse and Loquesse springs. ................................................................................................................... 104
Fig. 38: Monthly discharge in the Vatseret subcatchment between October 1975 and October 1979; and
between January 2007 and December 2012. ................................................................................................ 105
Fig. 39: Mean monthly annual cycle between October and September (hydrological year) of outflow from: the
sum of the sources MOL38 and MOL9, the RAN 1 source and the RAN 28 source, and with additionally
simulated streamflow at the outlet of the Boverèche subcatchment .......................................................... 107
Fig. 40: Hourly discharge of the Boverèche measured at Colombire Oct.2010Sept.2011. .................................. 108
Fig. 41: Daily simulated discharge of the Tseu_Lie subcatchment between 2007 and 2012 using different
settings: applying a varying soil depth, adding the gravitational redistribution of snow and changing the
description of albedo from a fix albedo to a varying albedo ......................................................................... 110
Fig. 42: (above) Annual cycle of simulated discharge of the Tseu_Lie subcatchment between 2007 and 2012
using different settings: applying a varying soil depth, adding the gravitational redistribution of snow,
changing the description of albedo from a fix albedo to a varying albedo, and additionally with the
projections for the near future; (below) annual cycle of simulated discharge in the Tseuzier catchment in
different years and as the mean over different periods: for the period 19751982, for the corresponding
projection in the near future 20162023, for the corresponding projection in the far future 20652072 and
for the same period but assuming clogging on the glacier, in 1976, in 1980 and in 2011 ........................... 112
Fig. 43: Simulated snow cover in the Tseu_Lie subcatchment for the three periods present (20072012), near
future (20482053) and far future (20972102) ............................................................................................ 115
Fig. 44: Spread of the simulated snow cover in the Tseu_Lie subcatchment for the two periods near future
(20482053) and far future (20972102) ....................................................................................................... 116
Fig. 45: Monthly cycle of discharge Q of the headwaters of the study area, as well as of each of its
subcatchments for the three periods present (20072012), near future (20482053) and far future (2097
2102), with the medians of the time series, their interquartile range (between 25 and 75% quartile), and
their envelope representing the minimum and maximum monthly precipitation runoff ............................ 118
Fig. 46: Monthly cycle of the components of the water balance for the headwaters of the study area in the three
periods present (20072012), near future (20482053) and far future (20972102). .................................. 120
Fig. 47: Monthly coefficient of determination for different extrapolations of solar radiation. ............................ 157
Fig. 48: Conversion factor for direct radiation for a southwest exposed surface with 25° slope, and for a north
west exposed surface with 25° slope at the same location of the Montana station.................................... 157
Fig. 49: Global solar radiation measured at the Montana station: measured and computed with GRASS .......... 157
Fig. 50: View of the cross section of the Lie110 gauging station ........................................................................... 157
Fig. 51: Water level measured at Loquesse against water level measured at Lie110 ........................................... 159
Fig. 52: Stagedischarge relationship at the Lie110 gauging station ...................................................................... 160
Fig. 53: Picture of the Loquesse spring the day of the installation of the gauging station (24 July 2012), as seen
from the station’s site; and picture of the installed water pressure sensor ................................................. 161
Fig. 54: Difference in discharge between Lie110 and Lou as a function of water level recorded at Loq .............. 162
Fig. 55: Stagedischarge relationship at the Lourantse gauging station for the period between 12.7.2012 and
25.10.2012. .................................................................................................................................................... 163
Fig. 56: Picture of the gauging station Erte_2011 on 11 October 2011 ................................................................ 164
Fig. 57: Stagedischarge relationship at the Erte_2011 gauging station for the period between 27.7.2011 and
25.9.2011. ...................................................................................................................................................... 164
Fig. 58: Picture of the installed water pressure sensor for Erte_2012 and overview of the location of the
Erte_2012 gauging station. ............................................................................................................................ 165
Fig. 59: Stagedischarge relationship at the Erte_2012 gauging station for the period between 25.7.2012 and
31.12.2012. .................................................................................................................................................... 165
Fig. 60: View of the cross section of the Tie100 gauging stations ......................................................................... 166
Fig. 61: Picture of the Tie100 gauging station. ....................................................................................................... 166
Fig. 62: Stagedischarge relationships at the Tie100 gauging station .................................................................... 169
Fig. 63: (above) Monthly cycle of precipitation P at MVE for the two periods 19312012 and 19802009, with the
medians of the time series, their interquartile range (between 25 and 75% quartile), and their envelope
representing the minimum and maximum monthly precipitation; (below) the same as above but for
temperature T. ............................................................................................................................................... 170
Tables
xiii
Tables Tab. 1: Headwaters’subcatchments description ...................................................................................................... 14
Tab. 2: Tseuzier’s subcatchments description .......................................................................................................... 21
Tab. 3: List of meteorological and gauging stations already present in the study area .......................................... 25
Tab. 4: List of meteorological, snow and rain gauging stations in the study area available 20072012 ................. 30
Tab. 5: List of gauging stations installed in the study area in the period 20112012. ............................................. 34
Tab. 6: Parameters values defined for the simulation of snow and icemelt with a varying albedo ...................... 65
Tab. 7: Measured seasonal mass balance of Glacier de la Plaine Morte for the hydrological years 2009/2010–
2011/2012 [m w.e.] (Huss et al. 2013) compared with the simulated melt. .................................................. 82
Tab. 8: Mean annual simulated water balance components in the Tseuzier catchment for the two simulation
periods 19751982 and 20072012 in [mm].................................................................................................... 83
Tab. 9: Performance indexes for the two simulation periods 19751982 and 20072012...................................... 90
Tab. 10: Measured seasonal mass balance of Glacier de la Plaine Morte for the hydrological years 2009/2010–
2011/2012 (Huss et al. 2013) compared to the seasonal mass balance simulated applying a fix albedo
(SIMalbfix) as well as a varying albedo (SIMalbvar) in [m w.e.]. ..................................................................... 93
Tab. 11: Mean annual simulated water balance components in the Tseuzier catchment between 2010 and 2012:
using different model settings and different meteorological data. ................................................................ 94
Tab. 12: Performance indexes for different model set ups and input data ............................................................. 97
Tab. 13: Mean annual simulated water balance components between 2007 and 2012 of the Tseuzier basin and
of the headwaters of the study area. ............................................................................................................ 109
Tab. 14: Overview of the water availability in the headwaters of the study area for different periods ............... 119
Tab. 15: Mean annual simulated water balance components of the headwaters of the study area for the three
periods present (20072012), near future (20482053) and far future (209720102). ................................ 121
Tab. 16: Monthly regression factors for the period 1.1.1975 31.12.1980 for computing global radiation at the
MVE station .................................................................................................................................................... 148
Tab. 17: Monthly regression factors for the gap period between 28.2.197731.5.1977 for computing global
radiation at the MVE station using data at the SIO station ........................................................................... 149
Tab. 18: Parameters values defined for the simulation of snow and icemelt with a fix albedo .......................... 152
Tab. 19: Landcover and topsoil parameters. .......................................................................................................... 153
GHG Greenhouse gas
FOEN Swiss Federal Office of the Environment (BAFU in german: Bundesamt für Umwelt)
Lie SA Lienne SA hydropower production company
Loq Loquesse spring
Lou Lourantse spring
NRP61 National Research Programme "Sustainable Water Management"
PIHM Penn State Integrated Hydrologic Model
RCM Regional climate model
2
I 1 Object of research In the last decades with the awareness of a changing and evolving environment, the number of studies
on how the climate and human activities affect the natural systems and cycles has constantly risen,
with the spillover effect to stimulate the development of more sophisticated models able to describe
the system processes, and improving predictions (Silberstein 2006; Liu and Gupta 2007). In addition,
efforts have been made in order to enhance the spatial and temporal scale at which predictions are
made. However the support of these developments by the increasing power of computers was neither
accompanied by the same significant increase in the availability of data, nor in the quality of the
measured data (Drécourt 2004a; Silberstein 2006).
Particularly in conjuction with the changes expected to happen to the locally available water resources
resulting from climate change, an increasing number of regional climate change impact assessment
studies have been launched the last decade, with the increasing awareness that global sustainability is
made of local/regional sustainability, and that resources management and natural variability are
tightly coupled and interact. As matter of fact stakeholders, managers and politicians need to be
informed and included in such studies, as we need them to be able to undertake measures and make
decisions on adaptation and mitigation strategies for the future (Reynard et al. 2014; Schneider et al.
2014). For this purpose, the establishment of a robust and reliable modelling framework is required.
Hydrological or watershed models are crucial, as they serve here as exploratory and predictive tools.
Usually to be able to adequately address questions about the past, present and future status of an
environment it would be appropriate to focus efforts to monitor and anticipate changes and have the
means to provide a historical context for the measurements. Yet, climate and hydrological monitoring
in mountain areas are known to be difficult and challenging tasks, as besides the tough environment
conditions to which measuring instruments are exposed, these remote areas require major efforts to
visit, maintain and keep the measurements ongoing (Diaz 2005). An other fundamental problem is
that often many of the equations used to represent processes occurring in the hydrological cycle
require calibration, thus the parameters involved cannot be directly measured, or they are invariably
applied at a scale different to that at which they were derived (Grayson and Blöschl 2000), and this is
even more true in alpine areas. Hence, in such regions since usually available observations are
discontinuous in space and time, and furthermore do not provide sufficient information about the
detailed processes that are represented by the model, it is often of practical impossibility to calibrate
it properly for any time and spatial scales.
Generally it could be said that the inclusion of more processes and/or controlling variables in the
system can only be justified on the basis that the inclusion of additional controlling mechanisms
should both improve predictive skill and facilitate the estimation of parameter values on the basis of
physiological characteristics or measurements (Montaldo et al. 2007). On the other hand, especially
for impact studies it is quintessential to keep the physical basis in the description of the dynamics, i.e.
more complex and detailed, as it assures a consistent reproduction of the behaviour of the system.
The higher the degree of conceptualization, the higher is the danger this would lead to a model that
mimics the system without understanding it.
At this point it is clear that the choice of an appropriate model is a demanding task, requiring good
diplomatic skills: the tradeoffs between parsimony, complexity and robustness should be tackled
identifying the optimum between data availability, model complexity and predictive performance. It
appears that in this sense an implicit requirement is the model to be flexible, i.e. extensible and
I INTRODUCTION I 1 Object of research
adaptable to the given circumstances.
For most of the countries around the world basic digital geospatial data such as a digital elevation
model (DEM), soil, geology and landuse maps are actually available, with varying resolution and
precision. They allow a topographic as well as a physiographic characterization of the environment,
whose features can be described with attributes. If sufficiently accurate these attributes have a great
potential, and regardless of being quantitative or qualitative, are viewed as relevant and
discriminatory indicators for processes (Pflaunder 2001). These data, as well as any other source of
information like studies carried out within or close to the study area or literature should be combined
and exploited in order to allow the implementation of a physically based model, despite the possible
scarcity of data and observations on site. Maybe one or some of the processes might need some
degree of simplification, in which case adjustments of the parameters will be needed, allowing
tailoring the model to the specific behaviour of the studied system. Automatic methods for parameter
adjustment seek to take advantage of the speed and power of digital computers, while being objective
and relatively easy to implement. In contrast, the trial and error method (manual approach), which has
been developed and refined over the years to result in excellent model calibration, is complicated and
highly laborintensive, and the expertise acquired by the modeller is not easily transferred (Boyle et al.
2000). However here this limitation is not considered decisive in carrying out the modelling task, as
this configuration is still considered representative of the best process understanding achievable from
available data and catchment knowledge (Konz et al. 2010).
All in all, the use of a procedure including manual calibration and commonly available data appears
particularly promising, as it offers the possibility to rely almost entirely on the available data, exploit
the hydrological knowledge of experts and transfer the model settings established in subbasins with
relatively good data to other ungauged basins. Such applications suggest that the model can be
regarded as a very powerful tool for monitoring water resources: it serves as an interpolator at
locations where it is practically impossible to observe the necessary information (Drécourt 2004a).
Of course, all of this envisages the availability of (at least) one subbasin where the model settings can
be verified either through direct measurements, or indirectly through some kind of plausibility checks.
In order to judge model’s predictive performance meaningful criteria need to be chosen. Spatio
temporal dynamics as well as spatial fields of instantaneous and timeintegrated hydrological
variables, such as evapotranspiration, soil moisture, channel discharge, or more typical and
characteristic for an alpine environment such as snowpack and snow melt, are adequate variables to
make such an evaluation. The quality and confidence of these different intermediate results,
respectively measurements, must be carefully appraised, because of course data can be corrupted by
different types of error. Uncertainties might be present in the forcing terms, in the measurements
themselves as well as in the spatial (or eventually temporal) extrapolation of these, in the model
structure and parametrization. Moreover scaling uncertainties arise from differences in the
discretization of the model, in the description of the physics behind this and finally from the
observations, which are usually carried out at a precise point location (Melching et al. 1990). Still,
usually the accuracy of at least some of these data is good enough to represent a precious source of
information, enabling to evaluate reasonably well the outcomes of the applied modelling chain.
Montanari and Di Baldassarre showed that if measurements are made following stateoftheart
techniques, observation uncertainty has a limited impact, with respect to model structural
uncertainty, on the results of hydrological models (Montanari and Di Baldassarre 2013). Further they
I INTRODUCTION I 1 Object of research
4
showed that particular care should be taken in discarding measurements, as in hydrological modelling
any information is important and the presence of data errors does not necessarily limit the usefulness
of observed records, from what it follows that an appropriate selection of hydrological complexity and
calibration strategy can increase the robustness of hydrological applications against data errors
(Montanari and Di Baldassarre 2013).
The hydrological cycle in alpine environments is to a large extent controlled by snow accumulation,
storage, redistribution, and melting (Parajka et al. 2012; Warscher et al. 2013). High altitudinal
gradients, a strong variability of meteorological variables in time and space, usually only locally
quantified snow cover dynamics, complex and often unknown hydrogeological settings, and
heterogeneous land use and soil formations result in high uncertainties in the quantification of the
water balance and the prediction of discharge rates (Warscher et al. 2013). However, despite these
difficulties, hydrological modeling systems are needed and applied to serve and support decision
making in water management. This is particularly the case in mountainous regions, which play a crucial
role as the “water towers” feeding downstream areas (Viviroli et al. 2007). The more the processes
occurring at these high elevations are simplified and conceptualized within a model, the more they
suffer from a lack of physical relevance and physical parameter interpretability (Drécourt 2004b; Clark
and Vrugt 2006). This implies that their predictability for new climate or environmental boundary
conditions might be restricted and not representative. Therefore increasingly complex physically
based models are applied. This may enable a more comprehensive and enhanced perspective of the
sensitivity and the effects of climate change on the water balance, including the consideration of
feedback processes on the different components of the hydrological cycle (for example the effects of
snow albedo on snow cover pattern, and ultimately on runoff generation (Jost et al. 2012; Pellicciotti
et al. 2012)). Furthermore, internal inconsistencies, such as an underestimation of precipitation input
that can be compensated for by an overestimation of meltwater (Konz and Seibert 2010; Pellicciotti et
al. 2012), might be reduced or avoided. However, while the hydrological predictability might increase
with complex models, data demands of course increase as well. A parallel evaluation of these two
issues, increased complexity and increased data availability, should help us to evidence, wheter we are
getting the right answers for the right reasons. In a time of local and global change in the water cycle,
when practical hydrological applications are increasingly used for impact studies and risk analysis this
is crucial.
During the past decades the Alpine climate has been subject to pronounced decadalscale variability,
but also to distinctive longterm trends consistent with the global climate response to increasing
greenhouse gas (GHG) concentrations (Gobiet et al. 2014). In the last 100 years the average annual
temperature in Switzerland has risen by more than 1.5° C (FOEN 2012). A trend analysis of 1959–2008
gridded Swiss temperatures showed that the seasonal trends are all positive and highly significant,
with an average annual warming rate of 0.35°C/decade (Ceppi et al. 2012). Spatial and temporal
variability are pronounced on a seasonal scale, however they clearly identified an anomalouslystrong
warming at low elevations in autumn and early winter and aboveaverage spring temperature trends
at elevations close to the snowline (Ceppi et al. 2012). Warming in Switzerland , particularly
pronounced from 1980 onwards, appears to be about twice that of the global average (FOEN 2012;
Gobiet et al. 2014), which may be explained in part by the differences in physical characteristics of
land and sea surfaces and is mainly caused by water vapour enhanced greenhouse warming (FOEN
2012; Philippona 2013). Furthermore, large areas in the northern hemisphere, and in particular the
Alps, are permanently or during prolonged periods covered with ice and snow. These areas are getting
I INTRODUCTION I 1 Object of research
5
smaller, meaning there is a larger dark surface area and