sónia liléo, phd wind resource analyst - r&d manager o2 vind ab stockholm, sweden
DESCRIPTION
Investigation on the use of NCEP/NCAR, MERRA and NCEP/CFSR reanalysis data in wind resource analysis. Sónia Liléo, PhD Wind resource analyst - R&D manager O2 Vind AB Stockholm, Sweden [email protected]. Olga Petrik Master thesis student Royal Institute of Technology Stockholm, Sweden - PowerPoint PPT PresentationTRANSCRIPT
Investigation on the use of NCEP/NCAR, MERRA and NCEP/CFSR
reanalysis data in wind resource analysis
Sónia Liléo, PhD
Wind resource analyst - R&D manager
O2 Vind AB
Stockholm, Sweden
Olga Petrik
Master thesis student
Royal Institute of Technology
Stockholm, Sweden
Why the need of reanalysis data in wind resource analysis?
Interannual variability of the wind speed
Need to long-term correct
the wind measurements
Long-term series of
wind data are needed
Reanalysis datasets may be used as reference dataseries for the
long-term correction of wind measurements.
Reanalysis dataset
Institution VintageTime interval
available
Horizontal resolution( lat x lon)⁰ ⁰
Vertical level
Temporal resolution
(h)
NCEP/NCAR NCEP 19951948 – present
(Monthly releases; 1 week delay)
5/2 x 5/20.995 sigma
level(1)
6(instan-taneous)
MERRA NASA 20091979 – present
(Monthly releases; 1.5 months delay)
1/2 x 2/3 50 m1
(time averaged)
NCEP/CFSR NCEP 2009
1979 - Dec 2009(planned to be
available on real time)
1/2 x 1/20.995 sigma
level(1)
1(instan-taneous)
(1) The 0.995 sigma level corresponds to a level of 99.5% of the surface pressure, that is equivalent
to approximately 42m a.g.l. for standard atmospheric conditions.
The reanalysis datasets analyzed in this study are the following,
There are two essential requirements that reanalysis datasets
have to fulfil in order to be used as long-term reference data
in wind resource analysis.
1. Good degree of correlation with wind measurements
2. Temporal consistency
These aspects have been investigated for the reanalysis
datasets NCAR, MERRA and CFSR.
1. Correlation analysis of NCAR, MERRA and CFSR
reanalysis wind data with wind measurements
Wind speed measurements from 24 masts (13 met masts and 11 co-locations on telecommunication masts) distributed rather uniformly over Sweden have been used in this analysis.
0 5 10 15 20 250
2
4
6
8
10
12
14
16
18
Correlation with CFSR for MastAmboke_70
CFSR wind speed (m/s)
Mas
t w
ind
sp
eed
(m
/s)
Mast wind speed (m/s)y=0.68338*x + 1.6753, R=0.86898
1. Correlation analysis of NCAR, MERRA and CFSR
reanalysis wind data with wind measurements
Wind speed measurements from 24 masts (13 met masts and 11 co-locations on telecommunication masts) distributed rather uniformly over Sweden have been used in this analysis.
0 5 10 15 20 250
2
4
6
8
10
12
14
16
18
Correlation with CFSR for MastAmboke_70
CFSR wind speed (m/s)
Mas
t w
ind
sp
eed
(m
/s)
Mast wind speed (m/s)y=0.68338*x + 1.6753, R=0.86898
1. Correlation analysis of NCAR, MERRA and CFSR
reanalysis wind data with wind measurements
Wind speed measurements from 24 masts (13 met masts and 11 co-locations on telecommunication masts) distributed rather uniformly over Sweden have been used in this analysis.
The correlation coefficient, R, of the linear regression fit between wind speed measurements from each mast and wind speed data from the nearest located reanalysis NCAR, MERRA and CFSR grid points have been analyzed.
Mast R-value NCAR data
R-value MERRA data
R-value CFSR data
Improvement in R-value MERRA as compared to
NCAR (%)
Improvement in R-value CFSR as compared to
NCAR (%)
M1 0.731 0.872 0.870 19.3 19.0M2 0.716 0.874 0.865 22.0 20.8M3 0.642 0.821 0.806 28.0 25.6M4 0.715 0.835 0.825 16.8 15.3M5 0.807 0.885 0.895 9.6 10.9M6 0.672 0.880 0.855 31.0 27.2M7 0.799 0.873 0.869 9.3 8.8M8 0.738 0.841 0.850 14.0 15.3M9 0.826 0.856 0.865 3.6 4.7M10 0.701 0.799 0.819 13.9 16.7M11 0.806 0.858 0.880 6.4 9.2M12 0.733 0.826 0.804 12.6 9.6M13 0.762 0.853 0.863 12.0 13.3M14 0.773 0.860 0.841 11.2 8.8M15 0.670 0.850 0.822 26.9 22.6M16 0.799 0.849 0.853 6.2 6.8M17 0.635 0.843 0.833 32.7 31.2M18 0.762 0.848 0.848 11.3 11.2M19 0.675 0.817 0.805 21.0 19.3M20 0.700 0.815 0.813 16.3 16.1M21 0.703 0.814 0.797 15.8 13.5M22 0.759 0.815 0.826 7.4 8.8M23 0.695 0.814 0.797 17.1 14.7M24 0.636 0.748 0.629 17.6 -1.0
Mean (%) 15.9 14.5
Stdev (%) 7.7 7.3
2. Analysis of the temporal consistency of NCAR, MERRA and CFSR
reanalysis wind speed data
kmin = k-value of the CFSR 64.5⁰N 21⁰E grid point.
Corresponds to the minimum k-value of all the NCAR, MERRA and CFSR grid points.
NCAR, MERRA and CFSR consistency maps
k/kmin for each of the NCAR, MERRA and CFSR grid points.
2.1. Procedure
10 12.5 15 17.5 20 22.5 2552.5
55
57.5
60
62.5
65
67.5
70
Longitude (degrees)La
titud
e (d
egre
es)
NCAR map of Sweden
-500
0
500
1000
1500
2000
2500
k/kmin
2500
-500
NCAR
2.2. NCAR, MERRA and CFSR consistency maps
k/kmin
10 12.5 15 17.5 20 22.5 2552.5
55
57.5
60
62.5
65
67.5
70
Longitude (degrees)La
titud
e (d
egre
es)
NCAR map of Sweden
-500
0
500
1000
1500
2000
2500
k/kmin
2500
-500
NCAR
10 10.7 11.3 12 12.7 13.3 14 14.7 15.3 16 16.7 17.3 18 18.7 19.3 20 20.7 21.3 22 22.7 23.3 2455
55.5
56
56.5
57
57.5
58
58.5
59
59.5
60
60.5
61
61.5
62
62.5
63
63.5
64
64.5
65
65.5
66
66.5
67
67.5
68
68.5
69
69.5
70
Longitude (degrees)
Latit
ude
(deg
rees
)
MERRA map of Sweden
-400
-300
-200
-100
0
100
k/kmin
100
-400
MERRA
2.2. NCAR, MERRA and CFSR consistency maps
MERRA data show predominantly weak downward long-term trends.
This result is in accordance with the downward long-term trend observed in the mean wind speed in Sweden during the period of 1951-2008 as reported by Wern et al.
Wern, L. and Bärring L., “Sveriges vindklimat 1901-2008. Analys av förändring i geostrofisk vind”, Meteorologi Nr 138/2009 SMHI, 2009
k/kmin
k/kmin
10 12.5 15 17.5 20 22.5 2552.5
55
57.5
60
62.5
65
67.5
70
Longitude (degrees)La
titud
e (d
egre
es)
NCAR map of Sweden
-500
0
500
1000
1500
2000
2500
k/kmin
2500
-500
NCAR
10 10.7 11.3 12 12.7 13.3 14 14.7 15.3 16 16.7 17.3 18 18.7 19.3 20 20.7 21.3 22 22.7 23.3 2455
55.5
56
56.5
57
57.5
58
58.5
59
59.5
60
60.5
61
61.5
62
62.5
63
63.5
64
64.5
65
65.5
66
66.5
67
67.5
68
68.5
69
69.5
70
Longitude (degrees)
Latit
ude
(deg
rees
)
MERRA map of Sweden
-400
-300
-200
-100
0
100
k/kmin
100
-400
MERRA
10 10.5 11 11.5 12 12.5 13 13.5 14 14.5 15 15.5 16 16.5 17 17.5 18 18.5 19 19.5 20 20.5 21 21.5 22 22.5 23 23.5 24 24.5 2555
55.5
56
56.5
57
57.5
58
58.5
59
59.5
60
60.5
61
61.5
62
62.5
63
63.5
64
64.5
65
65.5
66
66.5
67
67.5
68
68.5
69
69.5
70
Longitude (degrees)
Latit
ude
(deg
rees
)
CFSR map of Sweden
-1000
-500
0
500
1000
1500
k/kmin
1500
-1000
CFSR
2.2. NCAR, MERRA and CFSR consistency maps
k/kmin
k/kmin
k/kmin
2.3. Results
Reanalysis data
Range of k/kmin Mean value of |k/kmin| Standard deviation of |k/kmin|
NCAR [-755.6 ; 2784.8] 939.9 806.8
MERRA [-488.9 ; 184.5] 198.4 111.1
CFSR [-1136.8 ; 1719.5] 394.0 309.7
MERRA wind speed data show significantly weaker long-
term trends than NCAR and CFSR.
80% weaker long-term trend in average than NCAR.
50% weaker long-term trend in average than CFSR.
How does temporal inconsistency of reference wind data
influence the estimate of energy production?
3. Influence of the choice of reanalysis data on the energy production estimate - Case Study
Grid points
Distance from the mast (km)
R-value on wind speed
Trend
(k/kmin)
Energy correction factor
Relative difference in the energy estimate compared to using NCAR 57.5 N 15 E ⁰ ⁰
NCAR57.5 N 15 E⁰ ⁰
66 0.806 +1393 0.93 -
MERRA 57.5 N 14.7 E⁰ ⁰
61 0.817 -412 1.06 +14%
CFSR57.5 N 14.5 E⁰ ⁰
60 0.852 -106 1.10 +18%
MERRA58.0 N 14.7 E⁰ ⁰
10 0.858 -458 1.07 +15%
CFSR58.0 N 14.5 E⁰ ⁰
0 0.880 -79 1.09 +17%
Higher correlation coefficients for closer located grid points
Low temporal consistency
Mainly due to the difference in temporal consistency
Due to the closer location of the grid point and to the higher temporal consistency of the reanalysis data
There are two essential requirements that reanalysis datasets have to fulfil in order to
be used as long-term reference data in wind resource analysis.
1. Good degree of correlation with wind measurements
2. Temporal consistency
The higher spatial and temporal resolutions of MERRA and CFSR reanalysis wind data allow a better representation of the local wind climate.
An average improvement of 16% in correlation coefficient with local wind measurements is obtained for MERRA and 15% for CFSR when compared to NCAR.
The higher spatial and temporal resolutions of MERRA and CFSR reanalysis wind data allow a better representation of the local wind climate.
An average improvement of 16% in correlation coefficient with local wind measurements is obtained for MERRA and 15% for CFSR when compared to NCAR.
NCAR data show for some grid points large temporal inconsistencies that affect considerably the energy production estimates.
The relative difference in energy estimate is for a specific analyzed case about 14%, caused mainly by the difference in temporal consistency of the reanalysis data used.
NCAR data show for some grid points large temporal inconsistencies that affect considerably the energy production estimates.
The relative difference in energy estimate is for a specific analyzed case about 14%, caused mainly by the difference in temporal consistency of the reanalysis data used.
Conclusions
Similar analysis performed on the reanalysis wind direction would be of great
interest.
Future Work
How to correctly judge the uncertainty inferred by long-term trends in the energy
estimate should be further investigated.
The causes of the large temporal inconsistency observed in some grid data should
be analyzed in more detail.
The analysis of the reanalysis dataset ERA-Interim (not publicly available for
commercial uses) developed by ECMWF (European Centre for Medium Range
Weather Forecasts), would also be of great interest.
The NCEP/NCAR reanalysis data used in this investigation was provided by the
NOAA/OAR/ESRL PSD, Boulder, Colorado, USA.
Acknowledgements
The NCEP/CFSR data are from the NOAA’s National Operational Model Archive
and Distribution System (NOMADS) which is maintained at NOAA’s National Climatic
Data Center (NDCD).
The authors would also like to acknowledge the Global Modeling and Assimilation
Office (GMAO) and the GES DISC (Goddard Earth Sciences Data and Information
Services Center) for the dissemination of MERRA.
Thank you!