1. promotor: Prof. Dr. Nicole van Lipzig co-promotor: Dr. Matthias Demuzere
Wind energy in Europe under
future climate conditions
The statistical downscaling of
a CMIP5 model ensemble
Annemarie DEVIS
Sept 2014
8. Main Objective Estimation of the change in wind power (Pout) in
Europe under future climate conditions
GCM
Downscaling
Wind power (Pout)
Wind speed (U) at rotorheight
(at climate time scales for past & future)
GCM1
GCM2
GCM4
GCM5
GCM6
PDF
81. Introduction 4. Downsc. Development3. GCM evaluation 5. Downsc. Application2. Research Objectives 6. Conclusions
9. GCM
Downscaling
Wind power (Pout)
GCM1
GCM2
GCM4
GCM5
GCM6
• Evaluate all GCMs
- past
• Apply the downscaling on
GCM ensemble
- past
-future
• Develop downscaling to go
from one GCM to
rotorheight wind climate
- past
Sub Objectives
PDF
Main Objective Estimation of the change in wind power (Pout) in
Europe under future climate conditions
91. Introduction 4. Downsc. Development3. GCM evaluation 5. Downsc. Application2. Research Objectives 6. Conclusions
10. GCM5
Reanalysis
- in which real observations are
assimilated
- only available for present
= ?
GCM
GCM1GCM2
GCM4
GCM6
reanalysis
GCMs
If PDF score > 0.7 ok!
101. Introduction 4. Downsc. Development3. GCM evaluation 5. Downsc. Application2. Research Objectives 6. Conclusions
11. Figure: Probability density function (PDF) scores of the wind speed PDF at ~80 m (1979 - 2005).
111. Introduction 4. Downsc. Development3. GCM evaluation 5. Downsc. Application2. Research Objectives 6. Conclusions
PDF score
12. Figure: Lowest altitude for which all GCMs have PDF scores > 0.7 and remain >0.7 up to ~1500 m in representing the wind speed PDF
during summer day. White: no GCM has a level with a PDF score > 0.7. Gray: PDF score at 1500 m <0.7 and layers underneath have PDF
scores > 0.7. The surrounding graphs show on the left axis the probability density for the reanalysis minus the probability density for the
GCM at each bin for MIROC (green), CanESM (blue), NorESM (yellow), ISPL (pink), HADGEM (red) and CNRM (grey). ERA-Interim reanalysis
wind speed histograms are plotted in grey, and their frequency values are shown on the righthand y-axis. x-axes show wind speed (m s–1).
Wind speed
Wind speed
DensityDensity
121. Introduction 4. Downsc. Development3. GCM evaluation 5. Downsc. Application2. Research Objectives 6. Conclusions
Reanalyse PDF
GCM PDF
13. Figure: Lowest altitude for which all GCMs have PDF scores > 0.7 and remain >0.7 up to ~1500 m in representing the wind speed PDF
during winter day. White: no GCM has a level with a PDF score > 0.7. Gray: PDF score at 1500 m <0.7 and layers underneath have PDF
scores > 0.7. The surrounding graphs show on the left axis the probability density for the reanalysis minus the probability density for the
GCM at each bin for MIROC (green), CanESM (blue), NorESM (yellow), ISPL (pink), HADGEM (red) and CNRM (grey). ERA-Interim reanalysis
wind speed histograms are plotted in grey, and their frequency values are shown on the righthand y-axis. x-axes show wind speed (m s–1).
131. Introduction 4. Downsc. Development3. GCM evaluation 5. Downsc. Application2. Research Objectives 6. Conclusions
14. GCM
Downscaling
Wind power (Pout)
GCM1
GCM2
GCM4
GCM5
GCM6
1. Evaluate all GCMs
- past
3. Apply the downscaling on
GCM ensemble
- past
-future
2. Develop downscaling to go
from one GCM to rotorheight
wind climate in Cabauw
- past
Sub Objectives
PDF
Main Objective Estimation of the change in wind power (Pout) in
Europe under future climate conditions
141. Introduction 4. Downsc. Development3. GCM evaluation 5. Downsc. Application2. Research Objectives 6. Conclusions
15. Small scale
wind climate
Large scale
information
Statistical relationship between
small-scale and large-scale
Yi=βi,1.X1+ βi,2.X2+ε
Possible predictors (X)
PDF is defined by λ and k
GCM
Downscaling
PDF
151. Introduction 4. Downsc. Development3. GCM evaluation 5. Downsc. Application2. Research Objectives 6. Conclusions
17. Figure: GCM PDF – Observed PDF
without downscaling
with downscaling
OBSERVATIE PDF
GCM PDF
Winter day Winter night
Summer day Summer night
171. Introduction 4. Downsc. Development3. GCM evaluation 5. Downsc. Application2. Research Objectives 6. Conclusions
Winter day
18. Figure: GCM PDF – Observed PDF
without downscaling
with downscaling
Winter day Winter night
Summer day Summer night
Yi=βi,1.X1+ βi,2.X2+ε
Predictors:
•wind speed from ~ 1000m
Predictors:
•wind speed from ~ 1000m
•temperature gradient between in and out ABL
181. Introduction 4. Downsc. Development3. GCM evaluation 5. Downsc. Application2. Research Objectives 6. Conclusions
19. GCM
Downscaling
Wind power (Pout)
GCM1
GCM2
GCM4
GCM5
GCM6
1. Evaluate all GCMs
- past
3. Apply the downscaling on
GCM ensemble
- past
-future
2. Develop downscaling to go
from one GCM to rotorheight
wind climate in Cabauw
- past
Sub Objectives
PDF
Main Objective Estimation of the change in wind power (Pout) in
Europe under future climate conditions
191. Introduction 4. Downsc. Development3. GCM evaluation 5. Downsc. Application2. Research Objectives 6. Conclusions
20. Past (1979-2005) Future (2020-2049)
Downscaling
Pout
(GCM1) Pout
(GCM2)
Pout
(GCM3)
Pout
(GCM4)
Pout
(GCM5)
GCM1
GCM2
GCM3
GCM4
GCM5
Pout
(GCM1) Pout
(GCM2)
Pout
(GCM3)
Pout
(GCM4)
Pout
(GCM5)
GCM1
GCM2
GCM3
GCM4
GCM5
201. Introduction 4. Downsc. Development3. GCM evaluation 5. Downsc. Application2. Research Objectives 6. Conclusions
22. Ensemble mean
change in power
output
68 %
221. Introduction 4. Downsc. Development3. GCM evaluation 5. Downsc. Application2. Research Objectives 6. Conclusions
Lower bound
of ensemble
Upper bound
of ensemble
Conceptual example
23. Change in Pout(kW)
WINTERDAY
(1979-2005 to 2020-2049)(for a 2300kW turbine)
68 %
231. Introduction 4. Downsc. Development3. GCM evaluation 5. Downsc. Application2. Research Objectives 6. Conclusions
24. Change in Pout(kW)
WINTERDAY
(1979-2005 to 2020-2049)(for a 2300kW turbine)
SUMMERDAY
68 %
241. Introduction 4. Downsc. Development3. GCM evaluation 5. Downsc. Application2. Research Objectives 6. Conclusions
25. Pout : Exctractable power output (W)
U: Wind speed (m/s)
Cp: Power coefficient
ρ: air density
R: rotor diameter
251. Introduction 4. Downsc. Development3. GCM evaluation 5. Downsc. Application2. Research Objectives 6. Conclusions
U Cp
26. Figure: Effect of varying λ and k parameters on Weibull PDF
261. Introduction 4. Downsc. Development3. GCM evaluation 5. Downsc. Application2. Research Objectives 6. Conclusions
Is the change in Pout different
when only mean wind is taken
into account?
U
27. Ensemble mean change in λ Ensemble mean change in k Ensemble mean change in Pout
WINTERDAY
271. Introduction 4. Downsc. Development3. GCM evaluation 5. Downsc. Application2. Research Objectives 6. Conclusions
Conceptual example
28. 281. Introduction 4. Downsc. Development3. GCM evaluation 5. Downsc. Application2. Research Objectives 6. Conclusions
1. Evaluate all GCMs
3. Apply the downscaling on
GCM ensemble
2. Develop downscaling to go
from one GCM to rotorheight
wind climate in Cabauw
Sub Objectives
Summer: Small-scale bias in GCMs
Winter: Little small-scale bias in GCMs
Summer: Added value
Winter: Little added value
Possible changes in power output by 2020-2049:
•Significant decrease in Mediterranean
•Insignificant small increase in Northwestern
Europe
Sub Conclusions
29. Bedankt voor jullie aandacht
Arenberg Doctoral School of Science, Engineering &Technology
Faculty of Science
Earth & Environmental Science
Agentschap voor wetenschap
en technologie
30. • Added value of downscaling on representation of
rotorheight windclimate:
– Summer: Small-scale bias in GCMs added value
– Winter: No small-scale bias in GCMs little added
value
• Possible changes in power output by 2020-2049:
– Significant decrease in Mediterannean (~16%)
– Insignificant small increase in Northwestern Europe
301. Introduction 4. Downsc. Development3. GCM evaluation 5. Downsc. Application2. Research Objectives 6. Conclusions
31. Is the change in Pout dependent on
the turbine type?
WINTERDAY
311. Introduction 4. Downsc. Development3. GCM evaluation 5. Downsc. Application2. Research Objectives 6. Conclusions
32. Is there still an added value of downscaling?
Representation of past rotor height wind speed
Yes, during summer
Representation of change (future-past) in power output
Impossible to check …
WINTERDAY
With
downscaling
Without
downscaling
SUMMERDAY
321. Introduction 4. Downsc. Development3. GCM evaluation 5. Downsc. Application2. Research Objectives 6. Conclusions
Change in Pout(kW) (for a 2300kW turbine)
34. • Effect of downscaling
– Representation of present climate
• Summer: Small-scale bias in GCMs added value
• Winter: No small-scale bias in GCM little added value
– Climate change signal
• No effect
• Possible changes in power output by 2020-2049
– Significant decrease in Mediterannean (~16%)
– Insignificant small increase in Northwestern Europe
341. Introduction 4. Downsc. Development3. GCM evaluation 5. Downsc. Application2. Research Objectives 6. Conclusions