30710So222FuregvikSatelliteTracked Satellite-based wind maps as guidance for siting offshore wind farms B. R. Furevik1, H. A. Espedal1, T. Hamre1, C. B. Hasager2, O. M. Johannessen1, B. H. Jørgensen2 and O. Rathmann2
(1) Nansen Environmental and Remote Sensing Center, Edv. Griegsvei 3A, N-5059 Bergen, Norway. Phone: +47 5529 7288, fax: +47 5520 0050, email:
[email protected] (2) Risoe National Laboratory, Wind Energy Department, Frederiksborgvej 399, DK-4000 Roskilde, Denmark. Phone: +45 4677 5014, fax: +45 4677 5970, email:
[email protected]
ABSTACT Offshore wind farms have started to contribute important supplies of renewable energy. The energy production of a wind farm depends upon the local wind climate and so may be predicted in advance. Usually, the prediction is based on at least one year of accurate wind measurements. Satellite Synthetic Aperture Radar (SAR) wind mapping can be a useful tool in selecting optimal sites and may therefore increase the cost-effectiveness of planning wind farms, e.g. in feasibility studies. In the WEMSAR project1 wind fields from SAR, in situ measurements and model output from three test-sites have been analysed [1]. Subsequently, a WEMSAR tool for effectively retrieving wind data from SAR images and utilising them in the WAsP micrositing model, has been developed. Testing of the WEMSAR tool at Horns Rev offshore wind farm in Denmark is ongoing.
1 - Introduction In situ wind measurements are obtained from groundmounted instruments, such as cup- or sonic anemometers, which usually provides time series of averaged values at 1 or 10 minutes intervals. However, such measurements do not resolve the spatial variations in the wind field and so it may be difficult to estimate wind conditions at a nearby site. This problem is traditionally handled by using mathematical flow models, called micrositing models, such as the Wind Atlas analysis and application Program (WAsP2) developed at Risø National Laboratory, Denmark [2]. For accurate estimation of the wind energy potential in a given area, WAsP requires long-term wind statistics, knowledge of topography and an accurate time series of wind speed and direction during a whole year. Such measurements are expensive. It is therefore important to choose the most suitable site for the meteorological mast, which can be easier done if a fast estimate of the wind conditions can be obtained. We propose to use satellite Synthetic Aperture Radar (SAR) wind retrieval in selecting the location for the more detailed in situ observations [3]. The SAR is an active remote sensing instrument carried onboard several present satellites, i.e. the European satellites ERS-2 and Envisat and the Canadian satellite Radarsat-1. The SAR antenna sends pulses of microwave radiation (at C-band) to the side of the satellite track and records the signal scattered back to the antenna. Over the sea, the backscattered energy is related to the amount of wind-generated capillary and short wavelength gravity waves on the sea surface. Therefore, a relationship exists between the wind speed and the normalised radar backscattering cross section over the sea. The objective of the work reported here is to utilise this relationship to derive high spatial resolution offshore wind maps from SAR images for use in offshore wind farm siting.
1 2
http://www.nersc.no/~wemsar http://www.wasp.dk/
The SAR images used in this study are from the ERS-1 and ERS-2 satellites with a ground coverage of 100 km x 100 km. The images are calibrated to an absolute reference to obtain values for the normalised radar cross section (NRCS). These values are translated into wind speed at 10 m above sea level using an empirical C-band model function and the local incidence angle and wind direction. The wind direction is derived from the direction of wind streaks on the sea surface, originating from atmospheric roll vortices or Langmuir cells in the upper ocean. A processing tool for the SAR data is being developed to run as an add-on to the WasP program. The tool will be tested against data from the Danish west coast in the area of the worlds largest offshore wind farm, Horns Rev that has been in operation since December 2002.
2 - Approach Three sites were chosen for the study, Figure 1. These are the west coast of Norway, the west coast of Denmark and La Maddalena shallow in the Bonifacio strait between Corsica and Sardinia. We will describe the wind retrieval technique from satellite radar images, then give a brief description of the three sites and show comparisons between in situ data, model data and SAR data from the Norwegian site and of in situ data and SAR data at the Italian site. Finally, the tool developed for using SAR data in wind resource estimation will be described. The three models used in our study are: (i) algorithms used for wind vector retrieval from SAR, (ii) a mesoscale atmospheric non-hydrostatic model and, (iii) a micrositing model (Table 1). Model-runs for chosen days with SAR coverage are used for validation of the SAR retrieved wind, as a comparison with the in situ data. The mesoscale model is the Karlsruhe Atmospheric Mesoscale Model (KAMM2), which is normally used for calculating the regional wind resource maps from global re-analysis data (eg. ECMWF or NCEP) [4,5,6]. The micrositing model, WasP, has been run in a non-standard way, i.e. on single few-hours cases, in order to provide detailed flow information in the coastal areas for spatial comparison with the SAR retrieved wind fields. WAsP has furthermore normalised the wind speed down to 10 m above sea level (a.s.l) in agreement with the estimations from SAR. Table 1 – Overview of models used in the study. For the SAR data the C-band transfer models and a technique for retrieving the wind direction from the images has been used. MODEL
DESCRIPTION Empirical relation to transfer the normalized radar backscatter crosssection (NRCS) values into wind speed. Technique for deriving the direction of the main peak in the FFT-derived lowwavenumber image spectrum.
INPUT
OUTPUT
Wind direction, radiation incidence angle, NRCS values.
Wind speed at 10 m a.s.l. with 400m x 400m pixel size.
SAR images at 100m x 100m pixel size
Wind direction in areas of 12.5 km x 12.5 km over the image.
Karlsruhe Atmospheric Mesoscale Model (KAMM2)
Three-dimensional nonhydrostatic atmospheric mesoscale model
Hindcast wind field, bottom boundary conditions
Wind atlas analysis and application program (WAsP)
Micrositing model based on a mathematical flow model.
Wind speed and wind direction from meteorological measurements
(SAR) CMOD-IFR2 CMOD4
(SAR) Wind direction estimation
2.1 Wind retrieval from SAR images
All wind components, pressure, temperature. Wind climate and wind resources at a specific location. Wind speed at 10 m a.s.l.
Within the incidence-angle range of the SAR, the reflected energy from the sea surface is proportional to the surface wind speed. The short gravity waves responsible for the backscatter of the radar signal are created instantly by the wind stress and are therefore good indicators of the sea-surface wind speed. From a large number of collocations between scatterometer and buoy/model data, empirical relations between the wind speed (normalized to 10m above the surface) and the NRCS have been established [7,8]. These relations are known as C band models (CMOD). Their application and accuracy for SAR is reported in several papers [9-17]. The accuracy for wind speed typically lies within ±2 m/s for medium wind (5-18 m/s) and at ±20° for derived wind directions. The NRCS is obtained through calibration of the SAR precision images (PRI). The approach for calibrating ERS SAR PRI is described by Laur et al. [18] and the European Space Agency 3 (ESA) have made a package of routines available for their users for this purpose . SAR PRI has an initial pixel size of 12.5 m × 12.5 m. However, to obtain satisfactory statistical significance the images are transformed to a pixel size of 400 m × 400 m, which still allows detailed mapping of the wind [16]. The wind directions are estimated from FFT-derived image spectra over a SAR scene [15,16]. This wind direction field is used as input to the CMOD algorithm to obtain the wind speed; see Table 1.
2.2 The Norwegian site The in situ data at the Norwegian site are obtained by the Norwegian Meteorological Institute (met.no) at the island of Hellisøy (Figure 5). The weather station on Hellisøy is situated on top of a 10 m high mast mounted beside a house, and surrounded by several other masts and buildings. Such obstacles, the topography and the very rough terrain affect the wind climate. The anemometer is 33 m above sea level and records the average wind speed over ten minutes every hour. The speed-up effect caused by the terrain varies for the different wind directions and may be considerable under certain conditions [19]. WAsP and KAMM2 have been run on cases from 7 of the 49 days with SAR coverage [15,20]. 2.3 The Danish site The in situ data for the Danish site are obtained from the offshore mast at Horns Rev in the period May 1999 to October 2001 prior to installation of the worlds largest offshore wind farm [21]. These high quality data have been analysed extensively [22,23,28] and are used for the final validation of the WEMSAR tool [24]. 2.4 The Italian site In situ data for the Italian site are obtained from the Meteorological station Mezzo Passo, which is a 10m offshore mast situated in the narrow strait between Sardinia mainland and the Maddalena Island (Figure 2). Wind measurements are available as 10-min. mean speeds for the period January 1997 – June 1999. Nine SAR PRI images have been analysed at this site.
3
http://earth.esa.int/services/best/
Figure 1 – Location of the sites for study and validation in Norway, Denmark and Italy.
3 - Results 3.1 Comparison – absolute values The comparison of wind speed and wind direction from the in situ observations with those obtained from SAR, from the Norwegian site was discussed in Furevik and Espedal 2002 [11]. The SAR wind speed may be compared with in situ data at Hellisøy (rms of 3.1 m/s), with ECMWF data (rms of 1.6 m/s) outside Hellisøy [15] and with in situ data at La Maddalena (rms of 2.8 m/s) (Figure 3). The conclusion from comparing Hellisøy SAR wind vector retrievals with the in situ observations is that the best results for wind speed is obtained when the direction was retrieved from the images and used in CMOD-IFR2 [15]. At Maddalena the wind directions from SAR are in fairly good agreement with in situ measurements but the SAR wind speed values differ somewhat from observed. This can be explained by the location of the in situ measurement station in between the islands, while the SAR average is taken in more open waters (within the box in Figure 2).
Figure 2: Site 2 – La Maddalena Shallow.
Figure 3 – Comparison of wind speed (left) and wind direction (right) retrieved from ERS SAR images and from in situ observations at La Maddalena, Italy. Input wind directions are derived from the SAR images. At Hellisøy, the WAsP model was used to calculate the wind speed down from the mast level (33 m a.s.l.) to the level of the SAR wind speed estimations at 10m over the sea. The WAsP results are valid until about 10 km from the mast at Hellisøy. An area of 5.5 km by 5.5 km outside Hellisøy/Fedje was chosen to make a comparison of the SAR wind speeds with the model results of KAMM2 and WAsP (Figure 5). The comparison for the 7 cases with model runs is shown in Figure 4. For only one situation the SAR retrieved wind speed (12.3m/s) does not agree with either WAsP or KAMM2 wind speeds (9.0m/s and 5.8m/s respectively). The case corresponds to 21st of October 1997, a quite stationary situation with wind speed 10 m/s and Northwesterly wind direction as measured on Hellisøy. But the SAR image seems to be influenced by oceanic phenomena giving areas of higher and lower wind speed values in different parts over the open sea. Thus, in the area used for averaging the wind speed is relatively high (maximum at 14m/s) compared to the surrounding area with a wind speed of about 10m/s. For the rest of the cases the agreement is good for the SAR with at least one of the models. In two cases SAR underestimates compared to WAsP; one case is shown in Figure 5 while the other corresponds to 22nd of June 1996. Both are situations that are not stationary in terms of the weather situation, since the wind speed is changing at the time of the SAR acquisition. This means that the wave field responsible for the radar return may not in equilibrium with the wind speed above.
Figure 4: Wind speed from SAR using CMOD-IFR2 plotted against * wind speed from WAsP model and ° wind speed from KAMM2 model. 3.2 Comparison – spatial features
Figure 5 shows a comparison between WAsP output and the SAR retrieved wind from February 14 1996, during on-shore wind conditions. The wind speed from SAR is steadily increasing towards the coast from 6 m/s offshore to 10m/s close to the coast, but clear lines corresponding to atmospheric roll vortices indicate the South-westerly wind direction. At 2100 UTC, the anemometer at 33 m asl on Hellisøy recorded south-westerly wind 18.1 m/s 199°. The WAsP run used wind speed and direction data from the Hellisøy weather mast at the time of the satellite passage. From these data WAsP predicts 14 m/s wind speed offshore, while the SAR wind speed is less at about 10 m/s and even lower further out as mentioned above. From the time series of Figure 5, note that wind speed was steadily increasing that day which may be the reason for this unstationary SAR wind field. The KAMM2 model predicts a wind speed of 9-10 m/s offshore, which agrees well with the SAR/CMOD-IFR2 close to the coast. All three methods show a decrease in wind speed in the lee of Fedje and Hellisøy islands. A profile through the three data sets (SAR, WAsP and KAMM2) upwind from the coast across Fedje to the sea is indicated in all three plots and shown in Figure 5 (iii). Based on the 18.1 m/s recording at Hellisøy, WAsP gives an offshore wind speed of 14.5 m/s. KAMM2 and CMODIFR2 agree on a somewhat lower wind of 10m/s offshore. Leeward of Fedje all three indicate a drop of 1-2m/s in wind speed. Note that the SAR/CMOD-IFR2 wind profile is broken since SAR can not retrieve wind over land.
Fedje H ellisø y
(i) ERS SAR. CMOD IFR2 wind speed with used input wind directions.
(ii) Part of SAR wind map geocoded to WAsP grid.
(iii) Profiles from land across Fedje to sea. * indicates the wind speed observation. (iv) WAsP wind speed (m/s)
(v) KAMM2 (m/s) WAsP area indicated.
(vi) Hellisøy weather station
Figure 5. Site 2 – Hellisøy. (i) ERS SAR wind field over the Norwegian site, February 14 1996. The plot in (iii) show the profiles through (ii) SAR retrieved wind field, (iv) model output from WAsP at 10 m a.s.l. and (v) model output from KAMM2. (vi) Plots of wind speed, wind direction and temperature at the Hellisøy weather station. The vertical bar indicates the time of the SAR passage. Please note that the WAsP results are only valid to about 5-10 km from the Hellisøy weather station.
3.3 The WEMSAR tool During a 3-year EC project, a first version of a WEMSAR tool has been developed for retrieving wind fields from ERS SAR images and integrating these data into the WAsP programme [1]. The tool consists of the two parts sketched in Figure 6.
Figure 6: The WEMSAR tool consisting of a SAR wind retrieval part and a statistical part.
The first part of the tool is the wind retrieval from ERS SAR images. This module reads calibrated image files and the associated header files, obtained using the ESA ERS SAR toolbox. The images are reduced to approximately 1000x1000 pixels with a pixel size of 100 m by 100 m. The wind is then calculated using the technique for deriving the wind direction, CMOD4 and CMOD-IFR2. The output is binary files (floating-point data format) of wind speed and wind direction with a pixel size of 400 m by 400 m. These files are read into the second part of the tool, which is the statistical module. In this module all the satellite wind fields are treated together to provide input to WAsP. The add-on tool to WAsP calculates wind speed and wind
direction for the available satellite samples. As there will only be a limited number of satellite images available (typically less than 100), the statistics have to be described carefully [25] The SAR wind retrieval module is sketched in more detail in Figure 7. If the user inserts a wind direction from in situ observations, one single wind direction is used over the whole 100 km by 100 km image. Otherwise, the wind directions are calculated from SAR image spectra in boxes of 12.5 km by 12.5 km over the image. Before the C-band models are called, the user has the opportunity to remove erroneous vectors and to correct for the 180° directional ambiguity due to the wind direction retrieval technique [10,12,15,16]. Improved geophysical model functions and other types of SAR data e.g. ENVISAT ASAR and RADARSAT ScanSAR imagery will be implemented in a future software version.
Figure 7: The SAR wind retrieval module of the WEMSAR tool. The input is calibrated image files, and the output is files of wind speed from two models CMOD-4 and CMOD-IFR2 and wind direction. Pink boxes indicate where the user must provide input. The wind statistical module (see Figure 6) has been developed as add-on software to the WAsP programme [2]. The wind statistical tool is described in detail in [26]. The basic functionalities are: • to area-average the relevant footprint area of the SAR wind map into a wind speed and wind direction, representative for the selected site for each wind map [23,24], and • to calculate an observed wind climatology from the series of wind maps. The methodology of using very few samples (around 50-100 wind maps, compared to e.g. oneyear of hourly data totalling 8769 observations) is treated with advanced statistics and fitting functions to infer the Weibull A and k parameters, as well as error estimates. The result of this procedure is saved as a so-called Observed Wind Climate (OWC), an ASCII-file forming the interface to WAsP. Using the OWC, WAsP can calculate the wind resource maps based on satellite data.
4 - Conclusions The aim of the study is to utilise the advantages of remote sensing in finding the best locations for offshore wind farms. ERS SAR scenes from sites located in Norway, Italy, and Denmark have been studied. A total of 78 scenes for all sites were analysed. For a selected number of cases, the KAMM2 mesoscale model and WAsP model have been used to calculate the wind pattern over the sea in the coastal zone. Comparison results between SAR wind maps, in-situ data, KAMM2 and WAsP model results for the Norwegian and Italian sites have been reported. When direction was retrieved from the images, the wind speed in general agrees well with the measured wind directions. In an early open-ocean validation study Vachon and Dobson [6] found the best agreement between buoy wind speed and SAR retrieved wind speed when using in situ wind directions as input to the C-band models. Our result may reflect the problems of comparing offshore SAR wind measurements with onshore in situ data. However, as at present the accuracy of SAR wind estimates is not adequate for it to be used alone in wind resource assessments. The comparison of spatial features show a fair agreement, and all three methods (SAR, WAsP and KAMM2) show a decrease in wind speed of 1-2 m/s in the lee of Fedje and Hellisøy islands. Comparing SAR surface wind speeds with wind speed from WAsP runs based on the Hellisøy observations normalised to 10m a.s.l. was in fairly good agreement i.e to within ±2 m/s for four cases. The reason for the disagreement (SAR underestimating in two cases and overestimating in one case) for the last three cases seems to be due to unstationar weather situations or oceanic influence on the images. On the other hand, in two of these cases the SAR agree well with the KAMM2 wind speed, which may indicate that WAsP runs may not take the special conditions over the ocean (atmospheric boundary layer, sea surface roughness) into account as well as an atmospheric model. The fact that the SAR winds agree with one model or the other in six of the seven cases suggests that the surface measurements carried out using the SAR may add useful information to revealing a site. A prototype software, the so-called WEMSAR Tool, has been developed for utilising SAR retrieved wind measurements in WAsP. The structure and functionality has been described. The major advantage of the software is that it can be used in offshore areas where no suitable in-situ wind observations are available for wind resource estimation. The only input needed is a series of around 70 ERS SAR satellite wind maps. These can be purchased from ESA. The final validation of the prototype of the WEMSAR tool at Horns Rev site is ongoing. A future version of the WEMSAR tool will also be able to handle RADARSAR and ENVISAT SAR data. This will significantly increase the temporal coverage of SAR data worldwide for potential sites, providing necessary data for high-resolution wind climatology. Moreover, combining SAR (high spatial resolution wind; 400m, with improving temporal resolution due to more satellites) and scatterometer (high temporal resolution; every day but with low spatial resolution), it can be possible to obtain a fast first estimate of the wind distribution along a coastline. Acknowledgements The study has been carried out during the EC project ‘Wind energy mapping using Synthetic Aperture Radar (WEMSAR)’, project number ERK6-CT99-00017. Thanks are due to G. Gaudiosi at ENEA for providing in situ data from La Maddalena and to L. C. Christensen, NEG Micon, for valuable comments. ERS SAR data are provided by the European Space Agency, through AO3-281, AO3-153 and EO-1356. Finally, funding from the Norwegian Research Council is acknowledged.
References [1] The WEMSAR Consortium, WEMSAR Final Report, NERSC Technical Report no. 237, Nansen Environmental and Remote Sensing Center, Edv. Greigsvei 3a, N-5059 Bergen, Norway, March 2003.
[2] Mortensen, N. G, D. N. Heathfield, L. Landberg, O. Rathmann, I. Troen, and E. L. Petersen (2000) Wind atlas analysis and application program: WAsP 7.0 Help facility, User’s guide, Risø National Laboratory, Frederiksborgvej 399, Roskilde, Denmark, ISBN 87-55026672. [3] Furevik, B. R., C. B. Hasager, R. Barthelmie, H. A. Espedal, B. H. Jørgensen, O. Rathmann, S. Sandven, G. Gaudiosi, L. C. Christensen, S. Pryor and O. M. Johannessen , Satellite-based wind maps – are they useful for siting offshore wind farms, Proc OWEMES 2003 (Naples) [4] Frank, H., and L. Landberg (1997) Modelling the wind climate of Ireland, BoundaryLayer Meteorology, vol. 85, 359-378. [5] B. H. Jørgensen, B. Furevik, C. B. Hasager, P. Astrup, O. Rathmann, R. Barthelmie, S. Pryor. Off-shore wind fields obtained from mesoscale modeling and satellite SAR images, Offshore Wind Energy, EWEA Special Topic Conference, Brussels, Belgium, 2001 [6] B. H. Jørgensen, B. Furevik, C. B. Hasager, P. Astrup, O. Rathmann, R. Barthelmie, S. Pryor. Developments in mesoscale modeling and SAR imaging of Off-shore wind maps, Global Wind Power, EWEA Conference, Paris, France, 2002. [7] Quilfen, Y., B. Chapron, T. Elfouhaily, K. Katsaros and J. Tournadre (1998) Observation of tropical cyclones by high-resolution scatterometry, Journal of Geophysical Research, vol. 103, nr. C4. [8] Stoffelen, A., and D. Anderson (1997) Scatterometer data interpretation: Estimation and validation of the transfer function CMOD4, Journal of Geophysical Research, vol. 102, no. C3. [9] Scoon, A., I. S. Robinson and P. J. Meadows (1996) Demonstration of an improved calibration scheme for ERS-1 SAR imagery using a scatterometer wind model, Int. J. Remote Sens., 17 (2), 413-418. [10] Vachon, P. W, and F. W Dobson (1996) Validation of wind vector retrieval from ERS-1 SAR images over the ocean, The Global Atm. and Ocean Sys., 5, 177-187, 1996. [11] Fetterer, F., D. Gineris and C. C. Wackerman (1998) Validating a scatterometer wind algoritm for ERS-1 SAR, IEEE transaction on Geoscience and Remote Sensing, vol. 36, no. 2. [12] Korsbakken, E., J. A. Johannessen and O. M Johannessen (1998) Coastal wind field retrieval from ERS synthetic aperture radar images, J. Geophys. Res., vol. 103, no. C4, April 15. [13] Furevik, B., and E. Korsbakken (2000) Comparison of derived wind speed from SAR and Scatterometer during the ERS tandem phase, IEEE Trans., of Geosc. Rem. Sens., vol. 38, no. 2. [14] Horstmann, J., W. Koch, S. Lehner and R. Tonboe (2000) “Wind retrieval over the ocean using Synthetic Aperture Radar with C-band HH polarisation”, IEEE transactions on Geoscience and Remote sensing, vol. 38, no. 5, 2122-2131. [15] Furevik, B. R., and H. A. Espedal (2002) Wind energy mapping using Synthetic Aperture Radar, Canadian Journal of Remote Sensing, Vol 28, no. 2, pp. 196-204. [16] Furevik, B., O. M. Johannessen and A. D. Sandvik (2002) SAR-retrieved wind in polar regions – comparison with in situ data and atmospheric model output, IEEE Trans., of Geosc. Rem. Sens., Vol. 40, no. 8. [17] Thompson, D., F. Monaldo, R. C. Beal, N. S. Winstead, W. G. Pichel and P. ClementeColon (2001) Combined estimates improve high-resolution coastal wind mapping, EOS, Transactions of the American Geophysical Union, vol. 82, no. 41, October. [18] Laur, H., P. Bally, P. Meadows, P. J. Sanchez, B. Schaettler and E. Lopinto (1996) Derivation of the backscattering coefficient σ0 in ESA ERS SAR PRI products, Doc. ES-TN-RSPM-HL09, issue 2, rev. 2, Eur. Space Res. Inst., Frascati, Italy, June 28. [19] Jackson, P. S., and Hunt, J. C. R. (1975). Turbulent wind flow over a low hill, Quarterly Journal of the Royal Meteorological Society, Vol. 101, pp. 929-955. [20] Hasager, C.B., P. Astrup, R J. Barthelmie, E. Dellwik, B. H. Jørgensen, N. G. Mortensen, M. Nielsen, S. Pryor, O. Rathmann, (2002) Validation of satellite SAR offshore wind speed maps to in-situ data, microscale and mesoscale model results. Technical report. Risø-R1298(EN), p. 271. [21] Sommer, A., (2003) Offshore measurements of wind and waves at Horns Rev and Laesoe, Denmark. Proc. OWEMES 2003, 10-12 April 2003, Naples, Italy, pp. 65-79. [22] Hasager, C.B., B. R. Furevik, S. C. Pryor, R. J. Barthelmie (2002) Offshore wind resources quantified from satellite SAR: Methodology and technical aspects. In: Proceedings from the 1. International symposium on recent advances in quantitative remote sensing, Torrent (ES), 16-20 Sep 2002. Sobrino, J.A. (ed.), (Publicacions de la Universitat de Valencia, Valencia, 2002) p. 778-782.
[23] Hasager, C.B.; N. O. Jensen, M. Nielsen, B. R. Furevik (2002) SAR satellite image derived wind speed maps validated with in-situ meteorological observations and footprint theory for offshore wind resource mapping (poster). In: Proceedings of the 2002 global windpower conference and exhibition (CD-ROM), Paris (FR), 2-5 Apr 2002. (European Wind Energy Association, Brussels, 2002). [24] Hasager, C.B.; Nielsen, M.; Rathmann, O.; Furevik, B.R.; Hamre, T., Offshore wind maps from ERS-2 SAR and wind resource modelling. In: IGARSS 2003 (on CD-ROM). IEEE international geoscience and remote sensing symposium, Toulouse (FR), 21-25 Jul 2003. (IEEE, Piscataway, NJ, 2003) 3 p. [25] Barthelmie, R.J. and S. C. Pryor (2003) Can satellite sampling of offshore wind speeds realistically represent wind speed distributions. Journal of Applied Meteorology, 42(1), p 83-94. [26] Hasager, C. B., Rathmann, O., Nielsen, M., Barthelmie, R., Pryor, S. C., Dellwik, E., and B.R. Furevik 2003 Offshore wind resource assessment based on satellite wind field maps. Proceedings from 2003 EWEC - European Wind Energy Conference. 16-19 June 2003, Madrid, Spain, CD-ROM, 11 p. (in press). [27] Hasager, C.B, H.P.Frank, B.R. Furevik (2002) On offshore wind energy mapping using satellite SAR. Canadian Journal of Remote Sensing, 28/1, 80-89 [28] Hasager, C.B., Furevik, B., Dellwik, E., Sandven, S. Frank, H.P., Jensen, N.O., Astrup, P., Joergensen, B.H., Rathmann, O., Barthelmie, R.; Johannessen, O., Gaudiosi, G.; Christensen, L.C., (2001) Satellite images used in offshore wind resource assessment. 2001 Proceedings of the European Wind Energy Conference and exhibition (EWEC), Copenhagen (DK), 2-6 July 2001, 673-677
