Download as pdf or txt
Download as pdf or txt
You are on page 1of 6

Evaluation of Wind Energy Resources and Wind

Power Generation Based on SAR-Retrieved Wind in


the Eastern Sea Area of Yancheng, Jiangsu, China

Jie Jiang', Yongxue Liu", Saishuai Zhao', Minxi Zhou', Wangyu Cheng', Manchun Li'
'Jiangsu Provincial Key Laboratory of Geographic Information Science and Technology, Nanjing University, Nanjing, China
*Corresponding author, e-mail: yongxue@nju.edu.cn

Abstract-Evaluation of wind energy resources and wind power [5]. As a result, SAR images with a wide range of real-time and
generation can provide decision basis for wind energy construction. dynamic ocean surface information have a unique advantage in
Based on the study area in the eastern sea area of Yancheng, 8 offshore wind field retrieval [6]. A number of published studies
scenes of ERS-2 SAR images in 2007 are collected to retrieve wind have addressed the wind field retrieval methods and proved the
field and the retrieval results are verified by QuikSCAT wind data. feasibility of SAR-retrieved wind [5, 7-lO]. However, wind
Then, average wind power density is figured out to evaluate wind energy resources and wind power generation have not been
energy resources and unit capacity of wind power generator is further evaluated upon wind speed retrieved from SAR images.
calculated to evaluate wind power generation. The results show
that: (1) it is feasible and accurate to retrieve wind speed based on Accordingly, the purpose of this study is to combine wind
CMOD5 model; (2) regional variations of wind energy resources field retrieval with evaluation of wind energy resources and
and wind energy storage in different seasons should be taken into wind power generation. Firstly, wind field data is derived from
account; (3) average wind power density is from 1.22 to multi-temporal SAR images. Secondly, wind energy resources
929.51 Wlm2, and wind energy resources are abundant in three and wind power generation are evaluated to provide decision­
areas, the unit capacity of which could reach to 442.84, 1181.02 making basis for the construction of offshore wind power
and 1491.31 kW. generation.
Keywords-ERS-2 SAR images; wind field retrieval; wind energy The structure of the paper is as follows: Section 2
resources evaluation; wind power generation evaluation introduces the study area and dataset. Section 3 describes wind
retrieval, wind speed verification and wind energy evaluation
I. INTRODUCTION
methods. Section 4 analyses wind retrieval, wind speed
With the exhaustion of fossil energy resources and verification and evaluation results. Section 5 concludes the
seriousness of environmental and climatic issues, clean and paper.
renewable energy has become energy developmental focus of
countries around the world [1]. As a kind of clean and II. STUDY AREA AND MATERIALS
renewable energy, wind energy has been paid more and more
A. Study Area
attention globally. Due to the smaller surface roughness, 120"E 121°E
offshore wind energy resources are more abundant (about 3-4 N
times) than land wind energy resources, so it has greater
potential to develop offshore wind energy resources [2]. 's'l A �
Evaluation of wind energy resources and wind power
{;':<-
generation is the precondition for reasonable exploitation and "'..
utilization of offshore wind energy resources, the key of which )�",clle"g

is to obtain reliable wind direction data, wind speed data, and <f'
'9
temporal variation of wind field data.

The offshore wind field retrieval method has matured ,/C

recently based on the information sources of remote sensing
images and techniques of image processing and information
extraction [3]. Compared with ship data, island station data and
offshore buoy data, remote sensing images can provide
offshore wind field data with long time series and large areas.
So remote sensing images have played a more and more
significant role in the observation and research of offshore
wind filed [4]. Among them, Synthetic Aperture Radar (SAR),
not affected by weather and imaging time, can offer all­ Figure I. The location map
weather, all-time and high-resolution imaging of ocean surface

Supported by the National Natural Science Foundation of China (NO.


41171325 and JII03408) and the Program for New Century Excellent Talents
in University (NCET-12-0264)
Considering the abundance of wind energy resources and Moreover, the spatial resolution of ERS-2 SAR images is 30m,
construction costs, the study area located in the eastern sea area the pixel size is 12.5mx 12.5m and the pixel depth is 16 bit. In
of Yancheng, Jiangsu Province, China was selected. The this study, 8 scenes of ERS-2 SAR images (2007-01-07, 2007-
latitude range is from 32°53'N to 33°4'N and the longitude 02-11, 2007-04-22, 2007-05-11, 2007-08-05, 2007-09-09,
range is from 120053'E to 121°4'E. Climate of this area transits 2007-10-14, 2007-12-23) are used to retrieve wind direction
from warm temperate monsoon climate to northern subtropical and wind speed.
monsoon climate. The study area also belongs to China's
eastern coastal regions with advanced economy and large
(2) QuikSCAT data. QuikSCAT satellite is a marine
scientific satellite launched by National Aeronautics and Space
population. Therefore, the evaluation of wind energy resources
Administration (NASA) in June, 1999, the SeaWinds detector
and wind power generation is beneficial to electrical energy
of which can obtain wind direction and wind speed 10m height
supply for urban construction and economic development.
above the sea surface [11]. Obtained wind field data have
B. Dataset 25km spatial resolution and could cover more than 90% of the
The datasets used in this study can be organized into the world's ocean [12]. In order to verify retrieved wind speed, 8
following two categories: scenes of QuikSCAT wind data were collected on the same day
as the ERS-2 SAR images.
(1) ERS-2 SAR images. ERS-2 satellite carrying synthetic
aperture radar, wind scatterometer and some other payloads III. METHODS
was launched by Europe Space Agency (ESA) in 1995.

: :
,---------------- ,---------------- ,-----------------' ,----------------- ,-----------------
Evaluation of wind Evaluation of wind :
RS.2 SA�
Preprocessing Wind retrieval Wind speed verification
,
i ;---Relrieved� ,
,
power generation

D Images I

:
! �nd speed

1
/ ,
I
,
,
I
Power exponent
calculation of
wind energy
profile
i G;;�o� ] I
I
I

:: L .
lion

:I
, . Wind power
Wind speed
� . :
density at
covered the :V indpowcr 70m
: whole sludy density al
:--.; :--.; � _:"':":':':'_..J
area 10m , •
,
, Unit capacity
,
. , calculation of
,
I
Verification wind power
according to ,
QuikSCA T wind I
I
dam I
I
,
I
I
energy I
resources :
----------______ 1

Figure 2. Technology framework of the study

A. Preprocessing terrain changes result in surface area changes of


backscattering section; (2) terrain changes result in local
Data preprocessing mainly involves ERS-2 SAR images,
incidence angle changes [14]. Consequently, in this study we
including data filtering, radiometric calibration, and terrain
used Range-Doppler algorithm to correct terrain by the 90m
correction.
spatial-resolution SRTM DEM data downloaded from Europe
Radar images record amplitude information and phase Space Agency website. Range-Doppler algorithm principle is
information of backscattering resulted from interaction shown in equation (2):
between radar waves and surface features. For comparing
o 0 sin ()DEM
SAR images of different times, patterns, and sensors,
(JRange-Doppler = (JEllipsoid
SIll ()Ellipsoid

radiometric calibration is needed to transform original


amplitude information to backscattering coefficient. The (2)
formula is as follows [13]:
where (lRange-Doppler is backscattering coefficient afer terrain
(70 1- K + 10 ]ogIO[( R�) �
s n an GSYSO 1 (1)
O
correction, a EII,pso,d is backscattering coefficient before terrain
correction, 8DEM is local incidence angle, 8E1lipSOid is nominal
=
� smao
Gs s
y incidence angle.
where 1=1 Olog/O(DN/), DNij is the intensity of pixel (i, j),
K is the radiometric calibration constant, Rn is the slant
B. Wind Retrieval
distance of pixel (i, j), Ro is the referable slant distance, Gsys is Since evaluation of wind energy resources and wind power
the system gain of calibrated SAR image, GsysO is the system generation is based on retrieved wind speed, which builds on
gain of determined K. known wind direction, we should fIrstly retrieve wind
direction. There are wind stripes (I-8km interval) parallel to
Terrain differences also have a significant impact on marine wind directions in SAR images and these stripes are
backscattering coefficient, which specifically shows that: (1) caused by instabilities of marine atmospheric boundary layers
[S]. So in this study we calculated low-frequency band to which are also 10m height above the sea surface, to verify
retrieve wind direction by the following equation [10]: retrieved wind speeds. As wind speed retrieval results are
N N
discrete point data, spatial interpolation should be conducted
Y. '"' '"' . e-2:r1(jI+km)IN to get wind speeds covering the whole study area. In this
I,m �� X),k
=

j�l k�l study, we chose inverse distance weighting (mW)


(3) interpolation to interpolate wind speed by the following
formula [16]:
where Y is the low-frequency band of SAR image, X is the
gray value of SAR image, I, m =1,2, . . . . . . ,N.
Retrieved wind direction above still exists 1 80 ° wind
direction confusion. Hence we took QuikSCAT wind data on
the same day into consideration to determine the final where Zo is the estimated value of point 0, Zi is the value of
direction. control point i, dl is the distance between point 0 and control
Wind speed retrieval method originated in the C-band point i, n is the number of control points, k is the specified
scatterometer model, which is a geophysical model describing power.
the relationship between wind speed 10m height above the sea Therefore, retrieved wind speeds are fustly conducted
surface and backscattering coefficient on the moderately stable mw spatial interpolatin and then verified by QuikSCAT wind
atmospheric conditions [IS]. Among these scatterometer data on the same day.
models, CMOD4, CMOD-IFR2 and CMODS models are more
successful [8]. We chose CMODS model to retrieve wind D. Evaluation of Wind Energy Resources
speed and its basic form can be expressed as: Wind power density defmed as wind power per area
perpendicular to wind direction was used in this study to
aO =Bo(1+B, COS¢+B2 COS2¢)'6
evaluate wind energy resources. Moreover, wind power
(4) density is an important parameter to measure aplenty and
where (l is the backscattering coefficient, cP is the angel potential of wind energy resources in a region. The formula of
between instrumental azimuth and wind direction, Bo, BJ and average wind power density during a specific timeframe is as
B2 are functions between wind speed V and incidence angle 8. follows [17]:
If x=(8 -40)/2S, then 1 n
L
1=1 P
v,'
_

Dwp =-

Bo=1 0ao+a,v /(a2v,soY 2n

(S) (11)

/(s,so) =
{ (so)a g(so)
where Dwp is the average wind power density, n is the
record number during a specific timeframe, VI is the wind
speed of record i, P is the air density and it can be defined as
g(s)
(6) equation 12 [17].

where g(s)=1 1( 1 +exp(-s)), a=so( 1 -g(so)), So=CJrCJ3X, P


2 2 3
Y=C9+CJl;X+CIlX , aO=cJ+c2X+C3X +c+x , aJ=c5+C!iX, a2=c7+cs,X. P= RT
(12)
and
where P is the average atmospheric pressure, R is a
_ CI4(1+X)-CI6V(0.S + x-tanh[4(x+Cl6 +c17v)]) constant (2871 kg-I. KI), T is the annual average degree kelvin
BI- -

1+ exp(0.34(v- CI8 )) (KO).


(7)
E. Evaluation of Wind Power Generation
B2 = (-d, +d,vJ exp(-v2 ) The retrieved wind speed is 10m height above the sea
(8) surface while the height of wind turbine hub is mostly 60-70m
above the sea surface. In consequence, wind power density at
n
{:+b(Y_I) Y<Yo 70m should be calculated before wind power generation
v2 =
evaluation. Generally, average wind speed and average wind
Y;:::Yo
power density satisfy the following two equations [18]:
(9)
- -z
where y=(v+vo)/vo'10=cJ9, n=C20, a=Yo- (yo- l)/n, b=lI[n(yo- vn=vo( nt
2 zo
1)"-1], VO=C2J+C22X+C23X , dJ=c24+c25x+c2!iX , d2=c27+c2s,X.
(13)
28 coefficients concerning the formulas above can be Z
assigned values according to table. N
Wn = W0 (Zn)
C. Wind Speed Verification o
Wind speeds retrieved from SAR images are 10m height (14)
above the sea surface, so we used QuikSCA T wind data,
where Vo ' Wo is average speed, average wind power 1 3 1 2 1 --2
q=[(-pv )x(-nD )]xc =-nc W70D
2 4 p 4 p
density at the height of Zo, vn ' W" is average speed, average (15)
wind power density at the height of Zn, a is power exponent of
wind speed according to height change (this is also called where q is the unit capacity of wind power generator, p is
power exponent of wind speed profile for short), N is power the air density, W 70 is the average wind power density at 70m,
exponent of wind power density according to height change cp is the efficiency of wind power generator (generally 25%-
(this is also called power exponent of wind energy profile for 50% [22]) and in this study the value of cp is 25%, D is the
short). impeller diameter of wind power generator (generally 70-80m
of wind turbine wheel hub at the height of 70m) and in this
Values of a and N are related to surface roughness and the study the value of D is 70m.
power exponent of wind speed profile becomes smaller from
inland to coastal area [18]. According to China's architectural IV. RESULTS AND DISCUSSIONS
structure load specification, the value of a is 0.12 and N is
about three times as a, so in this study the value of N is 0.36 A. Wind Speed Retrieval Results
[19, 20]. We firstly processed data filtering, radiometric calibration
and terrain correction of ERS-2 SAR images by NEST
Unit capacity of wind power generator could be figured
software. Secondly, we figured out low-frequency band to
out based on utilization efficiency of wind power generator
retrieve wind direction existing 180 0 confusion. Thirdly, we
and average wind power density at 70m calculated according
determined the final wind direction according to QuiSCAT
to power exponent of wind energy profile. The equation is as
wind data on the same day. Fourthly, based on retrieved wind
follows [21]:
direction, we retrieved wind speed upon CMOD5 model. Wind
speed retrieval results are shown in Figure 3.
120'55'E 120'55'E 121'0'E 120'55'E 121'0'E 120'55'E 121'0'E

N
L..._...L-
....J _-'-
---'__.... �
Wind speed ( mls )
10
A
20
_1I1111I::==-___ kl11
I 2 J 4 5 6 7 8 9 10 II 12 13 14 15 16 17 18 19 20

Figure 3. Wind speed retrieved from SAR images


(a)2007-01-07; (b)2007-02-11;(c)2007-04-22; (d)2007-05-11;
(e)2007-08-05; (02007-09-09; (g)2007-1O-14; (h)2007-12-23

As illustrated in Figure 3, wind speeds retrieved from In this study, we used QuikSCAT wind data to verify wind
ERS-2 SAR images are mainly between 0 to 20 m/s. From speed retrieved from ERS-2 SAR images. Due to the low
spatial distribution aspect, wind speeds of different places spatial resolution of QuiSCAT wind data (generally 2 to 3
change vastly and randomly, but there are 3 to 8 areas of points in the study area), we verified 19 points in all during 8
larger wind speeds every month. From a time scale aspect, in time periods. Figure 4 shows the wind speed verification
2007, the wind speed of study area decreases first and results.
increases later in seasons. Specifically, the wind speed
decreases gradually from January to September and increases
later. In addition, the wind speed in September is lowest for
the wind speed is almost between 1 to 2m/s. As a result, when
we develop and utilize wind energy resources, we should pay
more attention to regional variations of wind energy resources
and wind energy storage in different seasons.
B. Wind Speed V erification Results
6 300-400 7.0 Very good
7 400-1000 9. 4 Very good

1200SS'E 121"aE 120"SS'E 121"O'E

(b)

!'
z

:::

Figure 4. V erification of wind speed retrieved from SAR images

Figure 4 shows that wind speeds retrieved from ERS-2


Classification of wind power density Unit capacity ofwiod powcr gcncrator( kW)
SAR images are close to QuiSCAT wind speeds. The retrieval
929.51 '?442.84 1 1 81.02 1491.31
I 2 J 4 ;-S
errors are small, for their absolute values are all between 0 to 1
m/s. Specifically, the number of absolute values of errors Figure 5. Evaluation results of wind energy resources and wind power generation
among 0 to 0.5m/s is 16, accounting for 84.21 %. The number (a)Wind power density; (b) Classification of wind power density and
unit capacity of wind power generator
of absolute values of errors among 0.5 to Im/s is 3, accounting
for 15.79%. Furthermore, retrieval errors are mainly caused by
v. CONCLUSIONS
CMOD5 model errors, IDW interpolation errors and time
errors between ERS-2 SAR images and QuikSCAT wind data. Based on the study area in the eastern sea area of
In a word, it is feasible and relatively accurate to retrieve wind Yancheng, Jiangsu Province, this study firstly retrieved wind
speed from ERS-2 SAR images based on low-frequency band field from 8 scenes of ERS-2 SAR images in 2007. Then,
method and CMOD5 model. wind speed retrieval results were verified by QuikSCAT wind
data. Lastly, wind energy resources and wind power
c. Wind Energy Resources and Wind Power Generation generation were evaluated upon retrieved wind speed. The
Evaluation Results main conclusions are summarized below.
In order to evaluate wind energy resources and wind power
generation, we firstly figured out wind power density at 10m (1) It is feasible to retrieve wind direction according to
covering the whole study area and classified wind power low-frequency band method and retrieve wind speed based on
density at 10m according to Table I. Secondly, we evaluated CMOD5 model from ERS-2 SAR images. Compared with
wind energy resources based on wind power density and its QuikSCA T wind data, the retrieval results are relatively
classification. Thirdly, we picked up areas of abundant wind accurate.
energy resources. Fourthly, upon wind power density at 70m (2) Retrieved wind speeds are mainly from 0 to 20 m/s.
reckoned by power exponent of wind energy profile, we The wind speed of different area changes largely and
calculated unit capacity of wind power generator according to randomly in space. Besides, the wind speed of study area
Formula 15 to evaluate wind power generation. The evaluation decreases first and increases later in seasons. Therefore, we
results are shown in Figure 5. should pay more attention to regional variations of wind
As shown in Figure 5, average wind power density of the energy resources and wind energy storage in different seasons.
study area is from l.22 to 929.51 W/m2 and the level of wind (3) The average wind power density of the study area is
power density is from 1 to 7 according to Tabel I. From spatial between l.22 to 929.51W/m2 and wind power density level is
distribution aspect, the wind energy resources are abundant in from 1 to 7. Moreover, the wind energy resources are
three areas of upper center and left-bottom center. abundant in three areas of upper center and left-bottom center
Consequently it has huge potential to construct wind power and the unit capacity of wind power generator could reach to
generation in these three areas. Furthermore, if wind power 442.84, 118l.02 and 149l.31 kW.
construction are carried out, the unit capacity of wind power
generator calculated according to Formula 14 and Formula 15 ACKNOWLEDGMENT
could reach to 442.84,118l.02 and 149l.31 kW. This research is supported by the National Natural Science
Foundation of China (NO. 41171325 and 11103408) and the
TABLE I. WIND POWER DENSITY LEVEL (IOM HEIGHT ABOVE
THE SEA SURFACE) Program for New Century Excellent Talents in University
(NCET-12-0264).
Reference value of Application
Wind power
Level annual average wind potential in wind REFERENCES
density(W1m2)
s�eed�m/s2 �ower generation
1 <100 4.4 [I] Q. Yan, Y. C. Chen, A. J. Wang, W. J. Yu, and Q. S. Chen,
2 100-150 5. 1 "Development obstacles of new energies in China and countermeasures:
A review on global current situation," Acta Geoscientica Sinica, vol.
3 150-200 5. 6 Relatively good
31(5),2010,pp. 759-767.
4 200-250 6.0 Good
5 250-300 6.4 Very good
[2] H. L. Xin, "Aspect on the development of offshore wind energy in sensor data from QuikSCAT," Journal of Tropical Meteorology, vol.
China," Periodical of Ocean University of China, vol. 40 (6), 2010, pp. 19(5I),2003,pp. 107-117.
147-152. [13] R. Q. Zhang, W. Zhang, Z. L. Yuan, "Study on the preprocessing
[3] P. Beaucage, M. Bernier, G. Lafrance,J. Choisnard, "Regional mapping techniques of multi-temporal ERS-2 SAR images being used in crops
of the offshore wind resource: Towards a significant contribution from classification," Remote Sensing Technology and Application, vol. 15(1),
space-borne synthetic aperture radars," IEEE Journal of Selected Topics 2000,pp. 60-62.
in Applied Earth Observations and Remote Sensing, vol. I (1), pp. 48- [I4] Y. S. Bao,J. Z. Wang,Y. J. Zhang,J. J. Qu, W. M. Zhang, C. Chen, "A
56,March 2008. semi-empirical model for correction of terrain influences in SAR
[4] K. W. Lu,J. Y. Hu,X. Y. Yang,"Spatial patterns in seasonal variability backscattering," Acta Geodaetica et Cartographica Sinica, vol. 40(4),
of sea surface wind over the South China Sea and its adjacent ocean," 20II,pp. 483-488.
Journal of Tropical Oceanography,vol. 31(6),2012,pp. 41-47. [15] F. M. Monaldo, D. R. Thompson, R. C. Beal, "Comparison of SAR­
[5] T. Wang, "Derivation of wind field in Huanghai Sea by synthetic derived wind speed with model predictions and ocean buoy
aperture radar," Transactions of Oceanology and Limnology,vol. 10(4), measurements," IEEE Transactions on Geoscience and Remote Sensing,
2007,pp. 10-23. vol. 39(12),pp. 2587-2600,December 2001.
[6] L. Zhang, H. Q. Shi, Z. Y. Long, H. D. Du, Z. Y. Zhang, "Overview of [I6] X. B. Deng, "Comparison between two space interpolation methods
the ocean wind retrieval from space-borne SAR images," Marine based on ArcGIS," Geospatiallnformation,vol. 6(6),2008,pp. 85-87.
Science Bulletin, vol. 31 (6),2012,pp. 713-720. [17] Z. H. Meng, H. Xu, H. Z. Du, "Evaluation of the wind energy resources
[7] J. Horstmann, W. Koch, S. Lehner, R. Tonboe, "Wind retrieval over the on the southeast coast of China," Journal of Natural Resources,vol. 6(I),
ocean using synthetic aperture radar with C-band HH polarization," 1991,pp. 1-12.
IEEE Transactions on Geoscience and Remote Sensing, vol. 38(5), pp. [I8] Z. C. Wang, L. Y. Ding, T. S. Huang, "Using 10m wind speed to
2122-2131,September 2000. calculate average wind power density at wheel height of wind turbines,"
[8] B. R. Furevik, O. M. Johannessen, A. D. Sandvik, "SAR-retrieved wind Meteorological Science and Technology,vol. 40(4),2012,pp. 680-684.
in polar regions-comparison with in situ data and atmospheric model [I9] S. C. Huang,J. Ren,B. M. Wang, W. Qian, W. L. Wang,"Evaluation of
output," IEEE Transactions on Geoscience and Remote Sensing, vol. wind energy at 80m in Jiangsu," Journal of the Meteorological Sciences,
40(8),pp. 1720-1732,August 2002. vol. 29(4),2009,pp. 4519-4523.
[9] G. T. Song, Y. J. Hou, P. Qi, "Wind vector retrieval using dual [20] H. Y. Shen, X. Wu, J. F. Xie, C. Liu, Z. H. Jiang, "Research on
polarization imagery of ASAR," Progress in Natural Science, vol. algorithm of wind energy parameters in surface layer with height,"
16(11),pp. 1183-1187,November 2006. Meteorological Monthly,vol. 35(7),2009,pp. 54-60.
[10] J. S. Yang, W. G. Huang, C. B. Zhou, B. Fu, A. Q. Shi, D. L. Li, [21] Y. C. Deng, Z. Yu, S. Liu, "Estimation of wind farm capacity based on
"Coastal ocean surface wind retrieval from SAR imagery," Journal of wake experiment," Acta Scientiarum Naturalium Universitatis
Remote Sensing, vol. 5(1),200I,pp. 13-16. Sunyatseni,vol. 49(6),2010,pp. 53-57.
[11] X. Fu, T. Y. Zhang, F. 1. Yu, "Application of QuikSCAT wind in [22] 1. M. Tian, S. J. Liang, T. L. Xue, "Wind power utilization research of a
typhoon storm surge simulation," Marine Forecasts, vol. 29(4), 2012, small wind generator," Journal of Electric Power, vol. 25(4),2010, pp.
pp.18-22. 303-305.
[12] C. X. Liu, X. C. He, "The analysis on the statistical character of
QuiSCA T scatterometer winds and strong wind frequency using remote

You might also like