Climate Forecasting Unit
State-of-the-Art Climate Forecasting for
Wind EnergyMelanie Davis, Francisco Doblas-Reyes, Fabian Lienert
CLIMRUN General Assembly, ENEA, Rome, July 2013
Climate Forecasting UnitPresentation Outline:Climate forecasting for wind energy
Problem: How can climate variability be a risk in wind energy decisions?
Solution: How can climate forecasting minimise this risk?
Methodology: Climate forecasting of wind speed, a seasonal example.
Caveats/Further research: What are the limitations and potential for wind energy forecasting?
Conclusions: State-of-the-art climate forecasting for wind energy, current status.
Climate Forecasting Unit
Problem: How can climate variability be a risk in wind energy decisions?
Mea
n W
ind
Spe
ed (
m/s
)
- Reduced uncertainty of future wind variability
- Identify likelihood of extreme events
1980 1990 2000 2010 2020
5
0
10
15
Time in yrs(lines represent 1st May./yr)
JJA '13
Seasonal Variability in Wind Resource at Site X
Observations
ForecastUncertainty
High
HighLow
Climatology Uncertainty
High
HighLow
Problem: Climate variability risk in wind decisions
Climate Forecasting Unit
Problem: Climate variability risk in wind decisions
Operational decisions (Wind farm/grid operator, trader)
Planning decisions (Policy maker, energy planning, grid development)
Investment decisions
Energy generation – balancing resources, energy trading, extremes, insurance?Maintenance – offshore most vulnerable
Market strategies – incentives, energy mixSpatial planning – balancing resources, reinforce/redesign distribution network
Site selection – robust resource assessments, portfolio designRevenue – robust projections, volatility over time, insurance?
(debt financing, throughout project)
-30 years
PAST Observations
FUTUREPredictionsP
RE
SE
NT
Weather Forecasts
Hours/days/weeks
ClimateForecasts
Months to seasons(1month-1year)
Seasonal Annual-Decadal
Inter/multi-annual (1-30years)
Multi-decadal(30+years)
HindcastsClimate Change
Climate Forecasting Unit
Solution: Climate forecasting of wind resources
- Robust assessments- Contingency plans
- Early-warning systems
- Monitoring- Mobilise resources- Prepare measures
- Instruction - Action
Operational decisions Planning decisions
Investment decisions
GUIDANCE/RISK MANAGEMENT
ACTION/RISK MINIMISATION
Climate Forecasting UnitMethodology:Climate forecasting of wind speed
Stage A: Wind Resource Assessment
Wind energy potential: Where does the highest wind occur? Wind energy volatility: Where does the wind vary the greatest?
Stage B: Wind Forecast Skill Assessment
Validation of the climate forecasts: How well can it reproduce the wind resources and its variability over past timescales
Stage C: Operational Wind Forecasts
Probabilistic forecast of future wind resource information
Climate Forecasting Unit
Spring 10m wind speed from 1981-2011 (ERA-Interim) in m/s
Stage A: Wind Resource Assessment Wind energy potential: Where is it the windiest?
Methodology: Wind Forecasts Stages
Climate Forecasting Unit
Spring 10m wind inter-annual variability from 1981-2011 (ERA-Interim) in m/s
Stage A: Wind Resource Assessment Wind energy volatility: Where does the wind vary the greatest?
Methodology: Wind Forecasts Stages
Climate Forecasting Unit
Europe
Stage A: Wind Resource Assessment
Spring 10m wind resource availability Spring 10m wind inter-annual variability
Areas of interest: Patagonia/
E.BrasilCentral Sahara, Sahel
China/Mongolia/N. Russia
W. Australia/Tasmania
S.America Africa Asia Australia
N.Mexico/N.Canada
N.America
UK/Baltic Sea
Where is wind resource potential and variability the highest?
Methodology: Wind Forecasts Stages
Climate Forecasting Unit
Stage B: Wind Forecast Skill Assessment1St validation of the climate forecast system:
Spring 10m wind resource ensemble mean correlation(ECMWF S4, 1 month forecast lead time, once a year from 1981-2010)
Methodology: Wind Forecasts Stages
Perfect Forecast
Same as Climatology
Worse than
Clima-tology
Can the wind forecast mean tell us about the future wind resource variability at a specific time?
Climate Forecasting Unit
Stage B: Wind Forecast Skill Assessment2nd validation of the climate forecast system:
Spring 10m wind speed continuous ranked probability skill score(ECMWF S4, 1 month forecast lead time, once a year from 1981-2010, no calibration)
Perfect Forecast
Same as Climatology
Worse than
Clima-tology
Can the wind forecast distribution tell us about both the magnitude of the wind resource variability, and its uncertainty at a specific time?
Methodology: Wind Forecasts Stages
Climate Forecasting Unit
Europe
Stage B: Wind Forecast Skill Assessment
Areas of interest: E.Brasil
N.ChileIndonesia/W.India
W. Australia
S.America Africa Asia Australia
Mexico/S.Canada
N.America
N.Spain/S.E Europe
Spring 10m wind resource magnitude and its uncertainty forecast skill
Spring 10m wind resource variability forecast skill
Wind resource variability forecast skill only
Both wind resource magnitude and its uncertainty skill
KenyaSomalia
Where is wind forecast skill highest?
Methodology: Wind Forecasts Stages
Climate Forecasting Unit
Europe
Stage B: Wind Forecast Skill AssessmentWhere is wind forecast skill highest?
Areas of Interest:(Forecast skill)
E.BrazilN.Chile
Indonesia/W.India
W.
S.America Africa Asia Australia
Mexico/S.Canada
N.America
N.Spain/S.E Europe
Magnitude + uncertainty forecast skillVariability forecast skill
KenyaSomalia
Stage A: Wind Resource AssessmentWhere is wind resource potential and volatility highest?
Europe
Wind resource inter-annual variability m/sm/s
S.America Africa Asia AustraliaN.America
Patagonia/E.Brazil
C.Sahara, Sahel
China/ Mongolia/N.Russia
W.Australia/Tasmania
N.Mexico/N.Canada
UK/Baltic Sea
Areas of Interest: (Resources)
Methodology Conclusion: Global Wind Forecasts in Spring
Mexico
N.Mexico/
E.BrasilMexico
E.Brasil
W.Australia
W.Australia
Climate Forecasting Unit Wind resource availability
Climate Forecasting Unit
Probabilistic forecast of (future) spring 2011,10m wind resource most likely tercile(ECMWF S4, 1 month forecast lead time)
N.America
MexicoMexico
Areas of Interest Identified:(Resources and Forecast Skill)
S.America
E.BrasilE.Brasil
W.
Australia
W.Australia
S.America
Stage C: Operational Wind Forecasts
Methodology: Wind Forecasts Stages
Climate Forecasting Unit
Probabilistic forecast of spring 2011,10m wind resource most likely tercile(ECMWF S4, 1 month forecast lead time)
N.America
MexicoMexico
Areas of Interest Identified:(Resources and Forecast Skill)
S.America
E.BrasilE.Brasil
W.
Australia
W.Australia
S.America
Stage C: Operational Wind Forecasts
Methodology: Wind Forecasts Stages
Climate Forecasting Unit
1. 10m wind not representative of wind turbine hub height.
Caveats and further research:Climate forecasting for wind energy
2. Lack of relevant, observational wind data for robust validations of forecast skill: reanalysis data used instead.
3. Seasonal wind forecasts assessed with a single climate model with 15 ensemble members: a multi-model, calibrated approach is needed with more ensemble members.
1. Multi-model approach needed for a more robust forecast skill assessment.
2. Seasonal wind forecasts to be made down to site-specific scales.
3. Collaborations undertaken to formulate seasonal wind power forecasts with simple wind energy models to issue theoretical power predictions.
Caveats
Further research
4. Explore the potential of decadal wind forecasts for wind energy sector.
Climate Forecasting Unit
1. Wind forecasting over seasonal to decadal timescales can help to minimise risk of future wind variability on operational, planning and investment decisions
Conclusions:Climate forecasting for wind energy
2. Seasonal wind forecasting is an emerging climate service within the renewable energy sector, whilst decadal wind forecasts are yet to be explored.
3. Some global regions are more vulnerable to wind resource variability over seasonal timescales than others
4. Although wind forecast skill is limited in some regions, there are others that show good potential (more so for predicting the resource variability than magnitude)
5. Based on points 3, 4, regions where operational Spring wind forecasts demonstrate the greatest value from research to date includes: Mexico, E.Brasil, W.Australia.
6. Seasonal and decadal wind forecast research to date includes several caveats, and there is scope for significant improvement with further research and better observational datasets.
Climate Forecasting Unit
Join the initiative at: www.arecs.org ✔ Seasonal and decadal, wind and solar forecast information✔ Provide feedback, register your needs✔ Receive a quarterly seasonal wind forecast newsletter
Advancing Renewable Energy with Climate Services (ARECS)
Climate Forecasting Unit
The research leading to these results has received funding from the European Union Seventh Framework Programme (FP7/2007-2013) under the following projects:
CLIM-RUN, www.climrun.eu (GA n° 265192)
EUPORIAS, www.euporias.eu (GA n° 308291)
SPECS, www.specs-fp7.eu (GA n° 308378)
THANK [email protected]