206 / 287
[
] 207
R
isk
G
overnance
and
M
anagement
casting and design studies for several decades. It has
the necessary components to account for snow accu-
mulation and melt, evapotranspiration, soil moisture
and runoff generation.
Swedish design guidelines in a changing climate
The simulation scheme for design flood determinations
in Sweden is rather unique. It was developed in the
1980s when it became obvious that current criteria were
obsolete and could lead to dangerous consequences.
The new guidelines were adopted in 1990, and a nation-
wide re-evaluation programme of all major dams in the
country started and is still ongoing. Recently the mining
industry has joined the hydropower industry in these
efforts.
The Swedish design flood calculation guidelines are
based on a classification of the dams into two main
categories, depending on the potential consequences
of a failure during flood conditions. Flood Design
Category I should be applied to dams for which a failure
could cause loss of life or personal injury, considerable
damage to infrastructure, property or to the environ-
ment, or other large economic damage. Flood Design
Category II should be applied to dams for which failure
could only cause damage to infrastructure, property or
the environment.
The PMF/PMP concept was found inadequate when
developing the Swedish guidelines because the river
systems are very complex, with many reservoirs and
floods, which are generated by an interaction between
snowmelt and heavy rainfall. Design flood determination
in Flood Design Category I is therefore based on a hydro-
logical simulation technique that describes the effects
of extreme precipitation under particularly unfavour-
able hydrological conditions. Reservoir operation is also
considered. Dams in Flood Design Category II should be
able to pass a flood with a return period of at minimum
100 years at full supply level.
Of particular interest is that in the new edition of
the Swedish guidelines for the determination of design
floods for dams, it is prescribed that climate change shall
be considered in all design studies. This has initiated a
research project with the aim of analysing possible impacts
of climate change on the design floods, and to find means
to account for climate change in future design studies.
A number of drainage basins and dams, relevant for
the power industry and the mining industry, have been
selected for the studies of climate change impacts on dams
according to both Design Flood Category I and Category
II. In these basins design floods are calculated to allow
for present day climate conditions and with available
regional climate scenarios for the future. Focus for the
design studies in a changing climate is on the first half of
the 21st century, but simulations will also be made up to
the year 2100. Results so far show that global warming
may have great significance for dam safety, flood risks
and the production of hydroelectric power in Sweden.
The milder and more unstable winters expected in the
future also mean that water will be released more often
especially with regards to extreme precipitation within relatively
small catchment areas – in other words, precisely that which is of
greatest interest from a dam safety point of view. It is also recognized
that new climate calculations are likely to be developed as science
itself advances. A new attitude must be developed by dam owners
to deal with this moving target. Truly – a new dimension in the
philosophy of dam safety has arrived.
Fundamental questions
The question of climate impact on dam safety is one of the subjects
addressed within the Nordic Climate and Energy Project – a collabo-
rative effort involving the Nordic and the Baltic countries. It is also a
matter addressed in several national research projects in the different
Nordic countries. The fundamental questions asked in these studies are:
• What will be the combined effects on dam safety of more
irregular winters including altered snow conditions, altered
precipitation and altered evaporation?
• How is the best use to be made of scenarios from meteorological
climate models in order to calculate the effects on design floods?
• What is the extent of uncertainty in the matter?
Methods
It is not easy to predict how global warming will affect flood risks in
the Nordic climate, where the most extreme floods are generated by
a combination of snowmelt and rainfall. What is most evident is that
a warmer climate will result in shorter and less stable winters, with
smaller spring floods and more frequent and less predictable floods in
winter. At the same time, climate scenarios indicate that there is a risk
of an increase in the most extreme rainfalls. But a higher temperature is
also expected to cause increased evaporation. The total picture is there-
fore a complex one. In regions in the north, with long winters and a lot
of snow, a decrease in the problems in conjunction with spring floods
may be expected unless there is an appreciable rise in winter precipita-
tion. In the south, where rain floods dominate, flooding problems may
increase unless the increase in evaporation is sufficient to counterbal-
ance the rise in precipitation.
Studies of the impact of climate change on water resources in
the Nordic area have to be based on a model strategy that properly
accounts for the hydrological complexity in the area. This strategy
is normally based on the combined use of different greenhouse gases
emission scenarios, output from global and regional climate models
and a hydrological runoff model. Usually several combinations of
modelling components are used, in order to explore uncertainties
in the estimates. In our earlier studies the number of combinations
used was normally only four: two emission scenarios combined with
two climate models. Our recent studies are based on an ensemble
of some 15 regional climate scenarios. It is a complex process to go
from a climate model to an off-line hydrological simulation without
losing statistical information about how temperatures, precipitation
and evapotranspiration are changing in the new climate.
Most often the work relies upon regional climate scenarios from
national research centres like the Rossby Centre regional climate
modelling team in Sweden. These will soon be complemented with
results from the European ENSEMBLES project, leading to a rather
large internationally co-ordinated ensemble of regional climate scenar-
ios (about 22 in total), which will facilitate uncertainty analyses and
other statistical processing.
The hydrological model used in the Nordic countries is usually
the HBV hydrological model, which has been widely used for fore-