measurements of wind and turbulence profiles with scanning doppler lidar for wind energy...

42 IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING, VOL. 1, NO. 1, MARCH 2008

Measurements of Wind and Turbulence ProfilesWith Scanning Doppler Lidar for Wind

Energy ApplicationsRod Frehlich, Member, IEEE, and Neil Kelley

Abstract—High-quality profiles of mean and turbulent statisticsof the wind field upstream of a wind farm can be produced usinga scanning Doppler lidar. Careful corrections for the spatial fil-tering of the wind field by the lidar pulse produce turbulence es-timates equivalent to point sensors but with the added advantageof a larger sampling volume to increase the statistical accuracy ofthe estimates. For a well-designed lidar system, this permits ac-curate estimates of the key turbulent statistics over various sub-domains and with sufficiently short observation times to monitorrapid changes in conditions. These features may be ideally suitedfor optimal operation of wind farms and also for improved resourceassessment of potential sites.

Index Terms—Doppler measurements, remote sensing, windenergy.

I. INTRODUCTION

T HE deployment of wind energy technology in the U.S.and Europe has increased rapidly along with the size

and capacity of wind turbines. These larger machines requiredetailed wind resource measurements at higher and higheraltitudes. Accurate wind speed, wind direction, and turbulencestatistics are required for wind resource assessment and effi-cient wind farm operation. Typically, in situ sensors mountedon towers extending 50 to 60 m above ground are used to char-acterize the atmospheric conditions. These measurements arelimited in both height and horizontal coverage and, therefore,require a longer time average to provide sufficient statistical ac-curacy. Wind resource assessment usually requires a minimumof one year of wind speed data from candidate wind energy siteswith minimal data loss (preferably less than 15%). Typically,

Manuscript received February 9, 2008; revised June 19, 2008. First publishedSeptember 30, 2008; current version published October 15, 2008. This work wassupported in part by the National Science Foundation (NSF) under Grants ATM-0522004 and ATM-0646401, in part by a renewable energy seed Grant from theUniversity of Colorado, and in part by the Army Research Office (ARO) underGrant W911NF-06-1-0256, Walter Bach program manager. The NREL workwas supported by the U.S. Department of Energy under Contract DE-AC36-99GO10337.

R. Frehlich is with Cooperative Institute for Research in the EnvironmentalSciences (CIRES), University of Colorado, Boulder, CO 80309 USA (e-mail:[email protected]).

N. Kelley is with the National Wind Technology Center (NWTC), NationalRenewable Energy Laboratory (NREL), Golden, CO 80401-3393 USA (e-mail:[email protected]).

Color versions of one or more of the figures in this paper are available onlineat http://ieeexplore.ieee.org.

Digital Object Identifier 10.1109/JSTARS.2008.2001758

the tower measurements are interpolated to higher hub-heightsfor power predictions. Both continuous wave and pulsed co-herent Doppler lidar are being evaluated to help provide neededresource information. Large scale wind resource mapping usinga 3-D scanning lidar has recently been evaluated for resourceassessment over a 10 10 km domain by Hannon et al. [1]with agreement in a monthly mean of lidar derived windspeedand direction to anemometer measurements of 0.02 m/s and1.7 , respectively. Other remote sensing sources of data suchas sodar and radar profilers suffer from limited height andhorizontal coverage.

There is also an immediate need to determine if modern lidartechnology can be adapted to provide useful information in tur-bine operations to increase power capture and to reduce struc-tural loads based on more accurate profiles of turbulence statis-tics. In many situations, information from a single upwind toweris often insufficient to fully describe the spatial aspects of orga-nized turbulent flows that induce significant structural loads onwind turbines [2].

Coherent Doppler lidar has become a mature technology thatprovides high-resolution measurements of wind speed and windstatistics [3]–[7]. The most promising technology for providingaccurate profiles of the mean and turbulent wind field statisticsin the boundary layer is an eye-safe solid-state scanning co-herent Doppler lidar [5]–[7]. A short laser pulse with a Gaussiantemporal profile is transmitted as a collimated beam. The pho-tons scattered by aerosol particles are converted into an oscil-lating signal with a Doppler frequency shift that is propor-tional to the radial velocity of those aerosol particles illumi-nated by the lidar pulse over the processing range-gate. Variousprocessing algorithms have been developed to produce accu-rate radial velocity estimates [3]–[15], [39]. The statistical ac-curacy of the radial velocity estimates is well documented bycomputer simulations [8]–[19], [39] and from statistical anal-ysis of Doppler lidar data [20]. Profiles of turbulent statisticshave also been produced from Doppler lidar radial velocity sta-tistics by removing the contribution from estimation error [17]and by careful correction for the spatial filtering of the velocityfield by the lidar pulse [13], [21]–[26], as well as using the lidarsignal spectral width [26].

Recent investigations in the effects of atmospheric turbulenceon wind energy generation have indicated the need for moreaccurate measurements of turbulence profiles, especially in thenighttime stable boundary layer (SBL) [27]–[31]. The statisticalaccuracy of profiles provided by tower measurements is limitedby the number of independent samples of the turbulent eddies

1939-1404/$25.00 © 2008 IEEE

FREHLICH AND KELLEY: MEASUREMENTS OF WIND AND TURBULENCE PROFILES WITH SCANNING DOPPLER LIDAR 43

[32] and, therefore, can be improved by using a 3-D measure-ment volume provided conditions are locally homogeneous andstationary over the analysis domain.

During a one-week period in early March 2007, a Lock-heed-Martin Coherent Technologies, Inc., WindTracer scanningcoherent Doppler lidar was operated at the National Wind Tech-nology Center (NWTC) of the National Renewable EnergyLaboratory (NREL) which is located south of Boulder, CO.Because of its location downwind of the Colorado Front RangeMountains, the NWTC provides a unique operating environ-ment to evaluate new wind turbine designs. While the NWTCwind regime is insufficient for profitable power generation onan annual basis, it provides frequent periods of both steady andvery turbulent winds that are ideal for test purposes. NRELresearchers have developed a sophisticated wind turbine sim-ulation capability that includes a stochastic inflow simulatorthat mimics the turbulence characteristics of the site up to atower height limit of 80 m. In the future, turbines with rotorsextending up to as much as 150 m may be installed. Thepurpose of measuring the wind and turbulence characteristicsusing the scanning lidar is to verify the turbulence simulator forall heights of interest for larger turbines.

The NWTC operates an 82-m meteorological tower at thewestern (nominally upwind) edge of its site approximately0.35 km to the west-southwest of the WindTracer lidar (seeFig. 1). The tower is equipped with cup anemometers and criti-cally damped (0.6 damping ratio) direction vanes at elevationsof 2, 5, 10, 20, 50, and 80 m elevations. Aspirated platinumresistance temperature detectors are mounted at the 2-, 50-, and80-m levels with precision temperature differences measuredbetween the 2- and 50-m and 50- and 80-m levels. The data iscollected at a rate of one sample every two seconds from whicha series of statistical quantities (means, standard deviations,and extremes tagged with the time and wind direction) arecalculated over a period of one minute and then stored.

A real-time postprocessing program, located on one ofNREL’s servers, applies quality control criteria to the 1-minstatistics and derives several additional parameters. These in-clude the surface friction velocity , an estimate of the surfaceroughness, the gradient Richardson number stability parameterbetween the height differences of 2 and 50 m, 50 and 80 m, and2 and 80 m, and the mean vertical wind speed shear between2 and 80 m.

II. SCANNING DOPPLER LIDAR

The Doppler lidar used in this study is a 2- m eye-safe Wind-Tracer lidar manufactured by Coherent Technologies, Inc. (CTI)[5], [6]. The lidar parameters are a Gaussian pulse width

m (full width at half maximum), a range-gate lengthm, an azimuth scan rate of 2.5 per second, a measure-

ment time per radial velocity profile s using 100 lidarpulses per radial beam, and an azimuth spacing . Thescanning geometry is similar to the one employed at the Pent-agon Shield file campaign [22], [23], i.e., the azimuth anglevaried from 45 –135 to sample a domain with a slight down-ward slope of 1 . The elevation angle varied from to

Fig. 1. Doppler lidar radial velocity � versus azimuth angle � and range foran elevation angle � � � and for a 36-s time interval at 4:47 UTC (21:47 LT)on March 9, 2007. The NREL tower is also marked.

produce the maximum amount of high-quality data below an al-titude of 250 m. Range-gate distances from 384 to 1896 m wereused for all the lidar analysis and the transverse dimensions ofthe lidar sensing volume for each radial velocity estimate is lessthan m which is much less than therange-gate length m and the assumption of zerois valid [17].

Fig. 1 shows an example of Doppler lidar radial velocity datafor a single azimuth scan at an elevation of 4 from a high windperiod of interest for wind energy applications. The large tur-bulent eddies are clearly visible as orange to red colored fea-tures. There are some random outliers beyond a range of approx-imately 1.8 km because of the low aerosol content. Therefore,a median filter was used to remove outliers in the processingdomain. An example of the radial velocity as a function of az-imuth angle for a fixed range-gate in Fig. 1 is shown in Fig. 2as well as the best-fit mean velocity assuming a constant windspeed and direction [22], [34]. The turbulent component of theDoppler lidar radial velocity is defined as the deviations fromthe best-fit model. The large random scatter is mostly producedfrom the atmospheric variability because the estimation error ofeach measurement is typically less than 0.1 m/s. These largevelocity variations produce large estimation errors in all thevelocity statistics. However, the 3-D sampling of the scanning


Fig. 2. Radial velocity (bullets) � versus azimuth angle� from the data shownin Fig. 1 for a range-gate distance � � �� m, elevation angle � � � , andaltitude � � �� m. The best-fit radial velocity [22] is shown as a dashed lineand the resulting wind speed and direction are 12.2 m/s and 278 , respectively.

Doppler lidar provides a larger number of independent samplesof the atmospheric processes compared with tower data, and,therefore, the lidar produces the most accurate statistics, whenthe turbulent field is approximately homogeneous and stationaryover the analysis space-time domain.

The turbulent component of the radial velocity for each range-gate is the transverse velocity component as a function of the arcdistance where is in radians. This is also the velocitycomponent incident on the rotor blades of a wind turbine if theturbine axis is aligned to the mean wind vector. The spatial sta-tistics of the turbulent fluctuations in the plane of the rotor aredefined by the transverse velocity variance , the outer scaleof turbulence (for the von-Kármán model) that defines thetypical size of the eddies, and the energy dissipation rate thatdescribes the small scale statistics.

Estimates of the turbulence statistics ( , , ) can be pro-duced from analysis of the structure functions of radial velocityin either the azimuth direction (azimuth structure functions)[22], [23] or the radial direction (longitudinal structure func-tion) [21], [23]–[26]. The critical turbulence statistics of thevelocity field are produced from these structure functions ofthe Doppler lidar radial velocity fluctuations using the best fitto a von-Kármán turbulence model assuming: 1) the structurefunctions satisfy the universal description of the von-Kármánmodel in the horizontal plane, 2) the spatial averaging by thefinite extent of the lidar pulse and processing algorithms areremoved by careful analysis, and 3) the contribution from theestimation error in the radial velocities is removed, which isnot as important for the higher winds and higher turbulenceregimes of interest for wind energy production. Fig. 3 showsan average Doppler lidar azimuth structure function with thebest-fit von-Kármán model for the case of a well resolvedturbulent eddy size . Fig. 4 shows a more challenging caseof a larger outer scale that reflects the large eddies shownin Fig. 2. There are larger deviations from the best-fit model

Fig. 3. Lidar azimuth structure function (o) using all data in the height interval40–60 m, best-fit model including spatial filtering by the lidar pulse (solid line),and the corresponding von-Kármán model (dashed line). The processing domainis � � � � �� and �� m and the resulting best-fit param-eters are wind speed � is 10.4 m/s, wind direction is 291 , � � ��m s , � �� m, and � � �� m/s.

Fig. 4. Lidar azimuth structure function (o) using all data in the height interval120–140 m, best-fit model including spatial filtering by the lidar pulse (solidline), and the corresponding von-Kármán model (dashed line). The processingdomain is � � � � �� and �� m and the resultingbest-fit parameters are wind speed � is 12.4 m/s, wind direction is 288 , � �� m s , � �� m, and � � �� m/s.

than those shown in Fig. 3, which is expected because there arefewer independent samples of the larger eddies.

III. LIDAR DERIVED PROFILES

Profiles of wind speed and direction are produced by com-bining the best-fit to the radial velocity as a function of azimuthangle (see Fig. 2) for all range-gates that occupy a height in-terval of dimensions 20 m [22], [34] within the processing do-main. Similarly, profiles of the turbulence statistics are gener-ated from the structure function analysis using radial velocitydata occupying the same specified height interval of 20 m. Fig. 5


Fig. 5. Profiles of wind speed, direction, �, � , and � from lidar analysis (bul-lets) for the same processing domain as Figs. 3 and 4. The results from the towerdata are indicated by (o).

Fig. 6. Profiles of wind speed, direction, �, � , and � from lidar analysis fortwo azimuth angle subsectors: 50 –90 (bullets) and 90 –130 (o).

shows the wind speed and turbulence profiles for the time in-terval 4:37–5:00 UTC (21:37–22:00 LT) on March 9, 2007 aswell as results from the nearby NREL tower data, which agreewell with lidar derived results, especially at higher altitudeswhere the effects of the terrain variations are less important. Thewind speed and direction estimated from the single range-gateshown in Fig. 2 is in good agreement with the average profileof Fig. 5, even with the large variations in the radial velocitywith azimuth . Most of the scatter in the profile of wind direc-tion is estimation error produced by the high turbulence level

and large outer scale . Note that from the tower data isthe standard deviation of wind speed measurements. The bulkRichardson number is 0.039 over the height interval of 2 to 80 m,i.e., dynamically unstable conditions. Note that the size of theeddies increase with height and above the 180 m altitude,it is difficult to quantify . The standard metric of turbulence

at a hub height of 90 m exceeds the IEC Class Avalue of 0.16 for high turbulence [35].

The same lidar analysis was performed in two subsectors ofthe domain, i.e., azimuth angles from 50 –90 and 90 –130 ,to investigate the spatial variability of the profiles (see Fig. 6).There are noticeable differences in the wind speed and direction

Fig. 7 Profiles of wind speed, direction, �, � , and � from lidar analysis fortwo consecutive volume scans: 5:29–5:36 UTC (22:29–22:36 LT) (bullets) and5:36–5:43 UTC (22:36–22:43 LT) (o).

Fig. 8. Profiles of wind speed, direction, �, � , and � from lidar analysis (bul-lets) and tower data (o) for a case of large wind direction shear.

above 60 m. However, the turbulence profiles are in agreement.Note that the size of the eddies is smaller than in Fig. 5 be-cause of the smaller processing domain. These profiles of meanwinds and turbulence statistics can be used for resource assess-ment of potential sites (see also Hannon et al. [1]).

The profiles from two consecutive lidar volume scans shownin Fig. 7 demonstrate the rapid change in conditions that couldbe a critical requirement for optimal control of wind turbines.Although the wind speed at a hub height of 90 m dropped bya factor of 2 over a 7 min. time interval, the turbulence pro-files are similar except for a noticeable increase in from 90to 180 m, which reflects the changing conditions on the largerscales of turbulence. An important feature of the 3-D scanninglidar analysis is the ability to provide accurate profiles in a shortobservation time compared with tower and sodar data.

Fig. 8 shows a large directional shear for the time interval6:37–6:45 UTC (23:37–23:45 LT) on March 10, 2007 with abulk Richardson number of 0.49 over the height interval 2–80 m,i.e., stable conditions. Typically, large directional shear is as-sociated with lighter winds, lower turbulence conditions, andsmaller eddy size . The lidar derived wind speed and direc-tion agree well with the tower data, especially at higher altitudes.Note that is much lower than in Figs. 5–7 and that the velocity


Fig. 9. Lidar azimuth structure function (o), best-fit model including spatialfiltering by the lidar pulse (solid line), and the corresponding von-Kármán model(dashed line) for an altitude � � �� m from Fig. 8 and with best-fit wind speed� � �� m/s, wind direction is 196 , � � ��m �s , � � ��m, and� � �� m/s.

standard deviation from the tower data is lower than the lidardata above 30 m which may be due to the larger processing do-main of the lidar analysis. These are challenging conditions forthe lidar algorithm because a large correction for the lidar pulsesmoothing is required as shown in Fig. 9 as the large differencebetween the best-fit model and the von-Kármán model. Clearly,a Doppler lidar with a pulse width of order 25 m would providemore accurate estimates of the outer scale and have sufficientestimation error for accurate estimates of the energy dissipationrate , especially for the higher turbulence conditions of interestfor wind energy applications.

IV. SUMMARY AND DISCUSSION

High-quality wind speed, wind direction, and turbulence pro-files were produced from analysis of Doppler lidar data thatsampled a 3-D volume from near the surface to 240 m aboveground. Small variations in the wind speed and direction wereobserved by splitting the domain into two angular sectors. Thestatistical properties of profiles from sub-domains degrade asthe size of the analysis domain is reduced. Further research isrequired to quantify the statistical properties of these profilesas a function of the lidar parameters, scanning pattern, pro-cessing algorithms, and spatial statistics of the turbulent velocityfields to determine the value of scanning Doppler lidar for re-source assessment [1]. A shorter lidar pulse width and a smallerrange-gate would provide more accurate estimates of the turbu-lence statistics and permit profiles of the longitudinal velocitystatistics, i.e., the statistics of the radial velocity in the direc-tion of the lidar beam [23]. In addition, the assumption of a uni-versal model for the turbulence statistics requires more investi-gation. More realistic turbulence descriptions [36]–[38] shouldimprove the estimation of the important turbulence parameters.Recent results indicate that the estimation of energy dissipationrate is insensitive to spatial filtering of the lidar data [23] and,

therefore, may also be insensitive to the form of the universalturbulence model.

The improved statistical accuracy of the volume averagedDoppler lidar derived profiles is ideally suited for monitoringrapid changes in atmospheric conditions for timely control ofwind farms (see Fig. 6). More research is required to determinethe optimal update rate, lidar design, analysis algorithms, andturbine control algorithms for wind energy applications. Longterm observations of winds and turbulence at an operationalwind farm are needed to better understand the affect of atmo-spheric conditions on power production and maintenance issuessuch as outages from excessive loading and vibration.

ACKNOWLEDGMENT

The authors would like to thank J. Crescenti, J. Sharp,A. Oliver, G. Poulos, R. Sharman, and S. Hannon for usefuldiscussions. They would also like to thank K. Barr, S. Hannon,and G. Pelk of Lockheed Martin Coherent Technologies, Inc.,for providing the Doppler lidar data.

REFERENCES

[1] S. M. Hannon, K. Barr, J. Novotny, J. Bass, A. Oliver, and M. Anderson,“Large scale wind resource mapping using a state-of-the-art 3-D scan-ning lidar,” presented at the European Wind Energy Conf., Brussels,Mar. 31–Apr. 3 2008, Paper CS4.2, unpublished.

[2] N. D. Kelley, B. J. Jonkman, and G. N. Scott, The Impact of CoherentTurbulence on Wind Turbine Aeroelastic Response and Its SimulationNational Renewable Energy Lab., Golden, CO, NREL/CP-500-38074,2005.

[3] R. T. Menzies and R. M. Hardesty, “Coherent Doppler lidar for mea-surements of wind fields,” Proc. IEEE, vol. 77, no. 3, pp. 449–462, May1989.

[4] M. R. Huffaker and R. M. Hardesty, “Remote sensing of atmosphericwind velocities using solid-state and CO coherent laser systems,”Proc. IEEE, vol. 84, no. 2, pp. 181–204, Feb. 1996.

[5] S. W. Henderson, C. P. Hale, J. R. Magee, M. J. Kavaya, and A.V. Huffaker, “Eye-safe coherent laser radar system at 2.1 �m usingTm,Ho:YAG lasers,” Opt. Lett., vol. 16, pp. 773–775, 1991.

[6] S. W. Henderson, P. J. M. Suni, C. P. Hale, S. M. Hannon, J. R. Magee,D. L. Bruns, and E. H. Yuen, “Coherent laser radar at 2 �m usingsolid-state lasers,” IEEE Trans. Geosci. Remote Sens., vol. 31, no. 1,pp. 4–15, Jan. 1993.

[7] C. J. Grund, R. M. Banta, J. L. George, J. N. Howell, M. J. Post, R.A. Richter, and A. M. Weickmann, “High-resolution Doppler lidar forboundary layer and cloud research,” J. Atmos. Ocean. Technol., vol. 18,pp. 376–393, 2001.

[8] D. S. Zrnic, “Estimation of spectral moments of weather echoes,” IEEETrans. Geosci. Electron., vol. GE-17, no. 4, pp. 113–128, Oct. 1979.

[9] B. J. Rye and R. M. Hardesty, “Discrete spectral peak estimation inincoherent backscatter heterodyne lidar. I. Spectral accumulation andthe Cramer–Rao lower bound,” IEEE Trans. Geosci. Remote Sens., vol.31, no. 1, pp. 16–27, Jan. 1993.

[10] B. J. Rye and R. M. Hardesty, “Discrete spectral peak estimation in in-coherent backscatter heterodyne lidar. II. Correlogram accumulation,”IEEE Trans. Geosci. Remote Sens., vol. 31, no. 1, pp. 28–35, Jan. 1993.

[11] R. G. Frehlich and M. J. Yadlowsky, “Performance of mean frequencyestimators for Doppler radar and lidar,” J. Atmos. Ocean. Technol., vol.11, pp. 1217–1230, 1994.

[12] R. G. Frehlich, “Simulation of coherent Doppler lidar performancein the weak signal regime,” J. Atmos. Ocean. Technol., vol. 13, pp.646–658, 1996.

[13] R. G. Frehlich, “Effects of wind turbulence on coherent Doppler lidarperformance,” J. Atmos. Ocean. Technol., vol. 14, pp. 54–75, 1997.

[14] A. Dabas, “Semiempirical model for the reliability of a matched filterfrequency estimator for Doppler lidar,” J. Atmos. Ocean. Technol., vol.16, pp. 19–28, 1999.

[15] J.-L. Zarader, A. Dabas, P. H. Flamant, B. Gas, and O. Adam, “Adap-tive parametric algorithms for processing coherent Doppler-lidar sig-nals,” IEEE Trans. Geosci. Remote Sens., vol. 37, no. 6, pp. 2678–2691,Nov. 1999.


[16] R. G. Frehlich, “Estimation of velocity error for Doppler lidar mea-surements,” J. Atmos. Ocean. Technol., vol. 18, pp. 1628–1639, 2001.

[17] R. Frehlich and L. Cornman, “Estimating spatial velocity statisticswith coherent Doppler lidar,” J. Atmos. Ocean. Technol., vol. 19, pp.355–366, 2002.

[18] R. G. Frehlich, “Velocity error for coherent Doppler lidar with pulseaccumulation,” J. Atmos. Ocean. Technol., vol. 21, pp. 905–920, 2004.

[19] V. A. Banakh and C. Werner, “Computer simulation of coherentDoppler lidar measurements of wind velocity and retrieval of turbulentwind statistics,” Opt. Eng., vol. 44, pp. 1–19, 205, paper 071205.

[20] R. G. Frehlich, S. Hannon, and S. Henderson, “Coherent Doppler lidarmeasurements of winds in the weak signal regime,” Appl. Opt., vol. 36,pp. 3491–3499, 1997.

[21] R. G. Frehlich, S. Hannon, and S. Henderson, “Coherent Doppler lidarmeasurements of wind field statistics,” Bound.-Layer Meteorol., vol.86, pp. 233–256, 1998.

[22] R. Frehlich, Y. Meillier, M. L. Jensen, B. Balsley, and R. Sharman,“Measurements of boundary layer profiles in an urban environment,”J. Appl. Meteorol. Climatol., vol. 45, pp. 821–837, 2006.

[23] R. Frehlich, Y. Meillier, and M. L. Jensen, “Measurements of boundarylayer profiles with in situ sensors and Doppler lidar,” J. Atmos. Ocean.Technol., in press.

[24] V. A. Banakh and I. N. Smalikho, “Estimation of the turbulence energydissipation rate from pulsed Doppler lidar data,” Atmos. Ocean. Opt.,vol. 10, pp. 957–965, 1997.

[25] F. Davies, C. G. Collier, G. N. Pearson, and K. E. Bozier, “Dopplerlidar measurements of turbulent structure function over an urban area,”J. Atmos. Ocean. Technol., vol. 21, pp. 753–761, 2004.

[26] I. Smalikho, F. Kopp, and S. Rahm, “Measurement of atmospheric tur-bulence by 2-�m Doppler lidar,” J. Atmos. Ocean. Technol., vol. 22,pp. 1733–1747, 2005.

[27] W. Blumen, R. Banta, S. P. Burns, D. C. Fritts, R. Newsom, G. S.Poulos, and J. Sun, “Turbulence statistics of a Kelvin-Helmholtz billowevent observed in the night-time boundary layer during the coopera-tive atmosphere-surface exchange study field program,” Dyn. Atmos.Oceans, vol. 34, pp. 189–204, 2001.

[28] R. M. Banta, Y. L. Pichugina, and R. K. Newsom, “Relationship be-tween low-level jet properties and turbulence kinetic energy in the noc-turnal stable boundary layer,” J. Atmos. Sci., vol. 60, pp. 2549–2555,2003.

[29] D. C. Fritts, Conditions Contributing to Adverse Loading of Wind Tur-bines in the Nocturnal Boundary Layer: Final Rep. NREL/SR-500-37809, 2005.

[30] R. M. Banta, Y. L. Pichugina, and W. A. Brewer, “Turbulent velocity-variance in the stable boundary layer generated by a nocturnal low-leveljet,” J. Atmos. Sci., vol. 63, pp. 2700–2719, 2006.

[31] N. D. Kelley, B. J. Jonkman, and G. N. Scott, The Great Plains Tur-bulence Environment: Its Origins, Impact, and Simulation NREL/CP-500-40176, 2006.

[32] D. H. Lenschow, J. Mann, and L. Kristensen, “How long is longenough when measuring fluxes and other turbulence statistics,” J.Atmos. Ocean. Technol., vol. 11, pp. 661–673, 1994.

[33] T. Warner, P. Benda, S. Swerdlin, J. Knievel, E. Argenta, B. Aronian,B. Balsley, J. Bowers, R. Carter, P. Clark, K. Clawson, J. Copeland,A. Crook, R. Frehlich, M. Jensen, Y. Liu, S. Mayor, Y. Meillier, B.Morley, R. Sharman, S. Spuler, D. Storwold, J. Sun, J. Weil, M. Xu,A. Yates, and Y. Zhang, “The Pentagon Shield field program: Towardcritical infrastructure protection,” Bull. Amer. Meteorol. Soc., vol. 88,no. 2, pp. 167–176, 2007.

[34] K. A. Browning and R. Wexler, “The determination of kinematic prop-erties of a wind field using Doppler radar,” J. Appl. Meteorol., vol. 7,pp. 105–113, 1968.

[35] International Electrotechnical Commission, IEC, Wind Turbines—Part1: Design Requirements IEC 61400-1 Ed. 3., 2005.

[36] D. H. Lenschow and L. Kristensen, “Applications of dual aircraft for-mation flights,” J. Atmos. Ocean. Technol., vol. 5, pp. 715–726, 1988.

[37] L. Kristensen, D. H. Lenschow, P. Kirkegaard, and M. Courtney, “Thespectral velocity tensor for homogeneous boundary-layer turbulence,”Bound.-Layer Meteorol., vol. 47, pp. 149–193, 1989.

[38] J. Mann, “The spatial structure of neutral atmospheric surface-layerturbulence,” J. Fluid Mech., vol. 273, pp. 141–168, 1994.

[39] R. G. Frehlich and M. J. Yadlowsky, “Errata,” J. Atmos. Ocean.Technol., vol. 12, pp. 445–446, 1995.

Rod Frehlich (M’84) received the M.Sc. degreein physics from the University of Saskatchewan,Saskatoon, SK, Canada, in 1977, for studies onthe stochastic properties of the aurora borealis, andthe Ph.D. degree in applied physics/electrical engi-neering from the University of California, San Diego,in 1982, for research on laser propagation throughatmospheric turbulence.

From 1984 to 1985, he studied theoretical methodsof wave propagation in random media at the La JollaInstitute, San Diego. Since 1985, he has been with

the Cooperative Institute for Research in the Environmental Sciences (CIRES),University of Colorado, Boulder. In 1998, he became a part-time visiting scien-tist at the Research Applications Laboratory, National Center for AtmosphericResearch, Boulder. The focus of his research has been on interdisciplinary prob-lems of atmospheric science, which include measurements and forecasting ofturbulence, Doppler lidar measurements of wind fields and turbulence, high-ratein situ measurements of atmospheric turbulence, advanced data assimilationto include the effects of atmospheric turbulence, wave propagation in randommedia, and measurements of winds from space.

Neil Kelley is a meteorologist specializing in ap-plying atmospheric science in support of engineeringproblems. Since 1980, he has been a PrincipalScientist specializing in wind energy researchwith the National Renewable Energy Laboratory(NREL), Golden, CO. He has authored papers andreports on the generation, impact, and control ofwind turbine low-frequency noise. Since 1988, hehas concentrated his research on the influence ofatmospheric turbulence on wind turbine structuralloads and vibratory response.

Mr. Kelley holds active memberships in the American Meteorological So-ciety, the American Association for the Advancement of Science, and SigmaXi, the Scientific Research Society. In 1982, he was the recipient of an NRELOutstanding Achievement Award for his work on wind turbine low-frequencynoise, and, more recently, he received an Outstanding Achievement Award fromthe U.S. Department of Energy for his research on the effects of turbulence onwind turbines.

measurements of wind and turbulence profiles with scanning doppler lidar for wind energy...

Documents