Transcript statkraft in europe

Effects of climate scenarios on the
hydropower sector – needs and
challenges in development projects and
long time forecasting
COST VALUE-2012 End User Needs for Regional Climate Change Scenarios
Kiel, 7-9 march 2012
Eli Alfnes, Statkraft Energi AS
STATKRAFT IN EUROPE
Head Office
Offices
Trondheim
Subsidiari companies
Power production
hydro, gas, wind and bio
Baltic Cable
Oslo
Småkraft
Fjordkraft
Stockholm
Skagerak
Energi
London
Amsterdam
Düsseldorf
Brussel
Lyon
Podgorica
Tirana
Istanbul
Ankara
Marbella
BEYOND EUROPE - SN POWER
Nepal
Laos *
India
Peru
Sri Lanka
The Philippines
Singapore
Chile
* 20% i THPC eies av Statkraft SF
HYDROPOWER PRODUCTION
GWh/week
NOK/MWh
7000
250
6000
200
5000
Price
150
4000
3000
Demand
Runoff
100
2000
50
1000
0
Typical profiles for Norway
4
Statkraft presentation
0
Week no
CLIMATE PROJECTIONS - WHEN
New power plant projects
Economy, design and dimension
Hydro and wind
Rehabilitation and revision of power plants
Dam constructions
Dam safety
Power turbine installations
Financial market operations on long time bases
5
Statkraft presentation
CLIMATE INFORMATION - WHAT
Hydropower generation
Runoff / inflow
Precipitation and temperature
Evaporation, snow and glacier balance
Annual water volume
Seasonal profile / variation
Occurrence of succeeding wet or dry years
Dams
As Hydro
Extreme floods
Financial market operations
As Hydro
Changes in the demand and supply pattern
6
Statkraft presentation
WHERE ARE STATKRAFT TODAY
Historical data as ensembles
Hydrological models (HBV)
Present state
Power plant and catchment scale
Norway + Sweden
Delta change for climate projections
-> Runoff time series
Climate change (%) from gridded
runoff maps from NVE and SMHI
Time horizon: ~ 50 years
Time horizon: ~ 25 years
Changes in water volume and
seasonal profile
Changes in water volume
Models for production planning
Changes in water availability
Investment decisions
7
Statkraft presentation
Financial market operations
Global climate model
Regional climate model
Emission
scenario
Interface
Hydrological model
Snow
Evaporation
Soil moisture
Which method to achieve the best possible description of the
uncertainty / risk
Are the climate model realizations equally god / equally likely
Guidance to select a handful of models / realizations
Runoff
Want the expectation value for runoff and a measure on the risk
which the uncertainty brings
8
Statkraft presentation
CHALLENGES
Simulated runoff from hydrological models using downscaled
GCM/RCM results as input must have the same statistical
properties and seasonal profile as when driven by observed
meteorological data
Not sufficient for hydrological modelling that mean values and standard
deviations of monthly meteorological data are reproduced (Beldring et al.
2008)
Downscaling methods – what do they do with the climate signal
How is the climate signal preserved throughout downscaling ?
How does the bias correction alter the dependency between climate
elements obtained from the RCM (or GCM for stat. emp. downscaling) ?
Transferring the climate signal to the hydrological models
Delta change – losing information of changes in frequency
Transient scenarios – statistics and seasonal profiles must agree with the
observations in the reference period
9
Statkraft presentation
CHALLENGES
More and more realizations of climate models
How do they differ – strengths and weaknesses
How to select representative scenarios
Uncertainty – how to deal with that
Methods for quantifying the uncertainty
Evapotranspiration - the joker?
Neither atmospheric models nor hydrological models simulate observed evaporation
correctly under present climate (Engeland et al. 2004).
10
Statkraft presentation
SUMMARY
Parameters of interest
Runoff
Precipitation and temperature
Evaporation for ‘validation’
Time frame
25 – 50 years (+ 15 years before and after)
Daily values
Maintain dependencies between parameters
Consistence between downscaled climate scenarios and
observations
Expectation ‘value’ and measure of uncertainties and risk
Guidance to which models
where, for what and at which time frames
11
Statkraft presentation
Thank you for your attention !
Statkraft
presentation
12
EXAMPLE – HBV SIMULATIONS ON CATCHMENTS
HBV-simulated runoff
HBV-models calibrated against observed runoff
80 years of historical data homogenized to today's climate
Simulated runoff for 1961-1990 and 1981-2010 as reference
1 RCM
Delta change
Predict runoff for 2050
•
13
Production models
Cost-benefit analysis
for power plant rehab
Directly comparable with today’s
situation
•
Achieve changes in annual runoff level
and seasonal profile
•
Miss out changes in frequency
• inter- and intra annual patterns,
extremes etc.
•
Uncertainty in climate change not
considered
•
HBV-model performance uncertain
• snow and glacier
• evaporation
Statkraft presentation
Projected runoff 2050 for Hølen (western Norway),
ECHAM/OPYC3 IPCC SRES B2 – HIRHAM met.no
EXAMPLE – RUNOFF MAPS NORWAY
Runoff change maps in mm and percentage
6 bias corrected GCM/RCM realizations
HBV-simulations - 1x1 km grid
30-year time periods representing 2035 and 2085, and reference
periods 61-90 and 81-10.
Aggregating area values of runoff change - typically
inflow catchments to hydropower stations.
Used in Long time price models and Cost-benefit
analysis
Svelgen I - klimascenarier tilsig
30
25
Tilsigsendring (%)
20
dmiecha1b
hada1b
15
smhibcma1b
hada2
10
hadb2
mpib225
5
0
1961-1990
14
-5
Statkraft
presentation
1981-2010
2021-2050
2071-2100
EXAMPLE – RUNOFF MAPS SWEDEN
Percentile maps of change in annual runoff (%)
16 (12) bias corrected GCM/RCM realizations
HBV-simulations on catchment level
Raster 1x1 km
Median, max, min, 25- and 75-percentile and mean
30-year time periods representing 2030, 2040, … , 2080
Extracting the change for selected areas
typically inflow catchment to hydropower stations
Used in
Long time price models
Cost-benefit analysis
Source: SMHI report No 2011-1
15
Statkraft presentation
FROM TODAY TO TOMORROW
Today’s method only gives us part of the picture
Need to address the uncertainty and the change in
frequency
but at which scale and to which level
Not realistic to run all possible scenarios and
downscalings through the entire line of models.
How best to capture the changes of interest
16
Statkraft presentation