{"Bibliographic":{"Title":"An evaluation of wind measurements by four Doppler sodars","Authors":"","Publication date":"1984","Publisher":""},"Administrative":{"Date created":"08-17-2023","Language":"English","Rights":"CC 0","Size":"0000084291"},"Pages":["AN EVALUATION OF\nWIND MEASUREMENTS\nBY FOUR DOPPLER SODARS\nAOBAOBAOBACB\nA\nINITIAL\nthe\nB","851\nB6\nno.5\nAN EVALUATION OF\nWIND MEASUREMENTS\nBY FOUR DOPPLER SODARS\nJ. c/ Kaimal\nJ.E. Gaynor\nP.L. Finkelstein\nM.E. Graves\nT.J. Lockhart\nReport Number Five\nJuly 1984\nNOAA\nBoulder Atmospheric Observatory\nAND\nATMOSPH\nPHERIC\nNOAA\nU.S. Department of Commerce\nAMOUNT\nNational Oceanic and Atmospheric Administration\nEnvironmental Research Laboratories\nCOMMUNITY\nLIBRARY\nA NOAA publication available from NOAA/ERL, Boulder, CO 80303.\nNOV 27 1984\nN.O.A.A.\nU. S. Dept. of Commerce","","AN EVALUATION OF WIND MEASUREMENTS BY FOUR DOPPLER SODARS\nby\nJ. C. Kaimal and J. E. Gaynor\nNOAA/ERL/Wave Propagation Laboratory\nBoulder, Colorado 80303\nP. L. Finkelstein*\nEnvironmental Protection Agency\nResearch Triangle Park, North Carolina 27711\nM. E. Graves\nNorthrop Services, Inc.\nResearch Triangle Park, North Carolina 27709\nT. J. Lockhart**\nMeteorology Research, Inc.\nAltadena, California 91001\nThis study was conducted for\nU.S. Environmental Protection Agency\nunder Interagency Agreement No. DW1-3F2A059\nWave Propagation Laboratory\nEnvironmental Research Laboratory\nU.S. Department of Commerce\nBoulder, Colorado 80303\n*\nOn assignment from National Oceanic and Atmospheric Administration\nPresent Affiliation: Meteorological Standards Institute\nFox Island, Washington 98333","NOTICE\nAcquisition of the information provided in this document\nwas funded in part by the United States Environmental\nProtection Agency under Interagency Agreement No. DW1- -\n3F2A059. This study was conducted jointly by NOAA/\nEnvironmental Research Laboratory and the Environmental\nProtection Agency.\nMention of a commercial company or product does not consti- -\ntute an endorsement by NOAA/Environmental Research Labora-\ntories or the Environmental Protection Agency.\niv","CONTENTS\nAbstract\nvii\nFigures\nviii\nTables\nxii\nAcknowledgments\nxiii\n1. Introduction\n1\n2. Description of Instrumentation\n3\n3. Description of Field Program\n9\n4. Measurement of the Standard Deviation of W\n17\n5.\nMeasurement of Wind Speed\n47\n6. Measurement of Wind Direction\n65\n7.\nSodar Rawinsonde Comparisons\n81\n8.\nCharacteristics of Sodar W Spectra\n87\n9.\nConcluding Remarks\n97\nReferences\n99\nAppendix\n101\nV","","ABSTRACT\nMeasurements of wind speed, wind direction, and the vertical component of\nturbulence, from four different commercially available Doppler sodars, are\ncompared with similar measurements from in situ sensors on a 300 m instru-\nmented tower. Results indicate that the four sodars measure wind speed and\ndirection accurately and with reasonably high precision. The sodars tended to\noverestimate the vertical component of turbulence at night and to underesti-\nmate it during the day. Precision in those measurements was considerably\npoorer than for the averaged speeds and directions. Analysis of the vertical\nwind from the sodars indicates that the measurement inaccuracies arise from a\ncombination of aliasing and spatial averaging.\nvii","FIGURES\nPage\nNumber\n1 Plot plan showing sodar antenna deployment in the BAO test area\n10\nContour map of the immediate BAO terrain showing location of the\n2\n11\ntower and the sodar test area\n3 (a) View of the sodar test area looking north from County Road 8\n12\n13\n(b) Instrumentation on the BAO 300 m tower\n4 Comparison of 100 m °W values from the AV sodar and the BAO\n26\nsensor\n5 Comparison of 100 m o w values from the RAD sodar and the BAO\n27\nsensor\n6 Comparison of 100 m °W values from the REM sodar and the BAO\n28\nsensor\nComparison of 100 m o W values from the XON sodar and the BAO\n7\n29\nsensor\n8 Comparison of 200 m ow values from the AV sodar and the BAO\n30\nsensor\nComparison of 200 m ow values from the RAD sodar and the BAO\n9\n31\nsensor\n10 Comparison of 200 m O w values from the REM sodar and the BAO\n32\nsensor\nComparison of 200 m o W values from the XON sodar and the BAO\n11\n33\nsensor\n12 Comparison of 300 m O values from the AV sodar and the BAO\nW\n34\nsensor\n13 Comparison of 300 m ow values from the RAD sodar and the BAO\n35\nsensor\n14 Comparison of 300 m ow values from the REM sodar and the BAO\n36\nsensor\n15 Comparison of 300 m ow values from the XON sodar and the BAO\n37\nsensor\nviii","16\nComparison of 100 ow values from RAD (mode: BI) sodar and the\nBAO sensor\n38\n17\nComparison of 200 m o values from RAD (mode: BI) sodar and\nthe\nW\nBAO sensor\n39\n18 Comparison of 300 m o W values from RAD (mode: BI) sodar and the\nBAO sensor\n40\n19 Comparison of 100 m o values from RAD (mode: MONO) sodar and\nW\nthe BAO sensor\n41\n20 Comparison of 200 m o W values from RAD (mode: MONO) sodar and\nthe BAO sensor\n42\n21\nComparison\nof\n300\nvalues\nfrom\nm\nRAD\n(mode:\nMONO)\nsodar\nand\no\nW\nthe BAO sensor\n43\n22\nComparison of 100 m o W values from RAD (mode: MULTI) sodar and\nthe BAO sensor\n44\n23\nComparison of 200 m values from RAD (mode: MULTI) sodar and\nthe BAO sensor\n45\n24\nComparison of 300 m O. W values from RAD (mode: MULTI) sodar and\nthe BAO sensor\n46\n25\nComparison of 100 m wind speeds from AV sodar and BAO sensors\n53\n26\nComparison of 100 m wind speeds from RAD sodar and BAO sensors\n54\n27 Comparison of 100 m wind speeds from REM sodar and BAO sensors\n55\n28 Comparison of 100 m wind speeds from XON sodar and BAO sensors\n56\n29\nComparison of 200 m wind speeds from AV sodar and BAO sensors\n57\n30 Comparison of 200 m wind speeds from RAD sodar and BAO sensors\n58\n31 Comparison of 200 m wind speeds from REM sodar and BAO sensors\n59\n32\nComparison of 200 m wind speeds from XON sodar and BAO sensors\n60\n33\nComparison of 300 m wind speeds from AV sodar and BAO sensors\n61\n34 Comparison of 300 m wind speeds from RAD sodar and BAO sensors\n62\n35\nComparison of 300 m wind speeds from REM sodar and BAO sensors\n63\n36\nComparison of 300 m wind speeds from XON sodar and BAO sensors\n64\n37\nComparison of 100 m wind directions from AV sodar and BAO\nsensors\n69\nix","Comparison of 100 m wind directions from RAD sodar and BAO\n38\n70\nsensors\nComparison of 100 m wind directions from REM sodar and BAO\n39\n71\nsensors\n40 Comparison of 100 m wind directions from XON sodar and BAO\n72\nsensors\nComparison of 200 m wind directions from AV sodar and BAO\n41\n73\nsensors\n42 Comparison of 200 m wind directions from RAD sodar and BAO\n74\nsensors\nComparison of 200 m wind directions from REM sodar and BAO\n43\n75\nsensors\n44 Comparison of 200 m wind directions from XON sodar and BAO\n76\nsensors\nComparison of 300 m wind directions from AV sodar and BAO\n45\n77\nsensors\n46 Comparison of 300 m wind directions from RAD sodar and BAO\n78\nsensors\n47 Comparison of 300 m wind directions from REM sodar and BAO\n79\nsensors\n48 Comparison of 300 m wind directions from XON sodar and BAO\n80\nsensors\nComparison of wind speeds and wind directions from rawinsonde\n49\n83\nand sodar measurements\n50 Wind speed and wind direction profiles from sodar (AV),\nrawinsonde, and tower measurements in a stable boundary layer\n84\n51 Wind speed and wind direction profiles from sodar (AV),\nrawinsonde, and tower measurements in a convective boundary\n85\nlayer\n52 (a) Schematic representation of distortions introduced in the\nW spectrum from attenuation due to spatial averaging and\nfrom aliasing.\n(b) Shift in spectral behavior with height and its implications\n89\nfor sampling and aliasing errors\n53 Sodar and sonic anemometer W spectra at (a) 150 m and (b) 200 m\n91\ncompared for morning conditions\n54 Sodar and sonic anemometer W spectra at 300 m compared for (a)\n92\nmorning and (b) nighttime conditions\nX","55 Time series of W corresponding to spectral forms shown in Fig. 54(b)\nin the frequency range 01.0 m/s.\n21","Regression analysis for °W\nTable 3 .\nN\nB1\nVendor\nBO\nHeight\np\n0.67\n178\n0.28\n0.85\nRAD\n100 m\n0.68\n139\n0.10\n0.89\nREM\n0.63\n190\n0.18\n0.89\nAV\n0.51\n171\n0.46\n0.67\nXON\n0.67\n144\n0.39\n0.77\nRAD\n200 m\n0.75\n119\n0.12\n0.88\nREM\n0.64\n167\n0.20\nAV\n0.84\n0.63\n146\n0.26\nXON\n0.83\n0.69\n158\nRAD\n0.72\n0.39\n300 m\n136\n0.79\nREM\n0.84\n0.17\n214\n0.68\nAV\n0.70\n0.19\n0.80\n157\n0.82\n0.13\nXON\n= estimate of correlation coefficient\np\nBO = intercept term\nB1 = slope term\nN = number of observations\n22","4.3 Sodar Modes\nThe sodar systems had two distinct operational modes: monostatic (MONO)\nand bistatic (BI). AV and REM employed the former mode; XON represented the\nlatter mode; RAD rotated daily among MONO, BI, and a combination of the two\nmodes called multistatic (MULTI). .\nA comparison of RAD modal data with simultaneous sonic standard deviation\nvalues yields the b, C, and S results of Table 4. The sample bias (b) is\nsignificantly nonzero in all modes at all heights. It is equivalent among\nmodes at each height at a probability level of 0.05, except for a significant\ndifference between MONO and MULTI biases at 200 m.\nComparability (c) and standard deviation (s) do not show a uniform ranking\nof modes with height. The large c-value for MONO at 200 m is partly due to\nthe large bias and the presence of two ow values slightly greater than 2 m/s\nin the subset. The s-values are inconsistent in their equivalence from height\nto height, possibly because of the sparseness of the data.\nScatter diagrams of individual 20 min average values of sodar ow versus\nsonic o are shown by height and by mode in Figs. 16-24. The BI mode has the\nW\nlargest departures from the 1:1 line. Figures 16-18 do not include sonic >\n1.0 m/s. Thus the plots involving BI mode are not strictly comparable with\nthe others.\nIn summary, there is a consistent tendency in all sodar systems to\noverestimate ow at low sonic readings (in stable conditions). Conversely,\nthe\nsodar systems underestimated °W when the true value was higher (unstable\nconditions), In all cases the regression lines had positive y-intercepts and\nslopes less than unity.\n23","On average, sample biases were generally significantly positive, but AV\ndid not show a significant bias at 100 m and was marginally biased at 200 m\nand 300 m. REM was the only vendor with a negative bias at 100 m, but at 300\nm, REM had a positive bias, as did all the others.\nREM performed very well with respect to C and S at 200 m, but results for\nthese statistics were not consistent with height. In fact, a different vendor\nemerged with the lowest, i.e., best, values of C and S at each level.\nMode switching in the RAD system revealed a significant amount of bias in\neach mode (BI, MONO, MULTI) at 100 m, 200 m, and 300 m. However, as Table 4\nindicates, there was little difference among modes.\n24","Table 4. Radian modal w compared with sonic\nw\nHeight\nMode\nb (m/s)\nC (m/s)\nS (m/s)\nN\n100 m\nMONO\n0.14\n0.28\n0.24\n68\nBI\n0.14\n0.24\n0.19\n44\nMULTI\n0.10\n0.22\n0.20\n66\n200 m\nMONO\n0.30\n0.50\n0.40\n58\nBI\n0.19\n0.35\n0.29\n30\nMULTI\n0.15\n0.26\n0.21\n56\n300 m\nMONO\n0.25\n0.36\n0.25\n57\nBI\n0.23\n0.40\n0.32\n48\nMULTI\n0.19\n0.39\n0.34\n53\nb = bias (accuracy)\nC = comparability\nS = standard deviation of differences (precision)\nN = number of observations\n25","2.5\nComparison of 100 m ow values from the AV sodar and the BAO sensor.\nAV-100m\n2.0\n1.5\nSonic (m/s)\n+\n+\n+\n1.0\n+\nLINE\nINTERCEPT-0 AND SLOPE-1 (45° ANGLE)\n# HO\n0.5\nHOOO\nFigure 4.\n0000 #\nNIGHT TIME OBSERVATIONS 0\n+\nstoot\nDAY TIME OBSERVATIONS\n0\nREGRESSION LINE\n0.0\n1.0\n0.5\n0.0\n2.5\n2.0\n1.5","2.5\nRAD-100m\nComparison of 100 m o W values from the RAD sodar and the sensor.\n2.0\n1.5\nSonic (m/s)\n+\n+\n+\n+\ntot\n1.9\n+\n+\n+\nHA\n+\n+\n#\n#\n++\n+\nH+\n+\n#\nINTERCEPT-0 AND SLOPE-1 (45° ANGLE) LINE\n0.5\nFigure 5.\nNIGHT TIME OBSERVATIONS 0\n+\nDAY TIME OBSERVATIONS\nREGRESSION LINE\n0.0\n2.5\n2.0\n1.5\n1.0\n0.5\n0.0\n27","2.5\nComparison of 100 m °W values from the REM sodar and the BAO sensor.\nREM-100m\n2.0\n1.5\nSonic (m/s)\n+ +\nI\n+\n+\n+\n+\nto\n1.0\nIt\nINTERCEPT-0 AND SLOPE-1 (45° ANGLE) LINE\n0.5\nFigure 6.\nNIGHT TIME OBSERVATIONS\nDAY TIME OBSERVATIONS\n0\nREGRESSION LINE\n0.0\n0.0\n1.0\n0.5\n2.5\n2.0\n1.5","2.5\nXON-100m\nComparison of 100 m ow values from the XON sodar and the BAO sensor.\n2.0\n1.5\nSonic (m/s)\n+\n+\n+\nthe # ++\n#\n++ # #\n+\n1.0\n#\n+\n+++\n+\n++\nINTERCEPT-0 AND SLOPE-1 (45° ANGLE) LINE\n0.5\nFigure 7.\n+\nNIGHT TIME OBSERVATIONS\nDAY TIME OBSERVATIONS\nREGRESSION LINE\n0.0\n2.5\n2.0\n1.5\n1.0\n0.5\n0.0","2.5\nComparison of 200 m o values from the AV sodar and the BAO sensor.\nAV-200m\n2.0\n+\n+\n1.5\nSonic (m/s)\n+\n+\n+\n+\nto\n+\n+\n#\n+++\n1.0\n+\n#\n+\n+\n#+\n+\nIf\n+\n+ +\n+\n+\n++++\n+\nINTERCEPT.0 AND SLOPE-1 (45° ANGLE) LINE\n+\n0.5\nFigure 8.\nNIGHT TIME OBSERVATIONS 0\n+\nDAY TIME OBSERVATIONS\nREGRESSION LINE\n0.0\n0.0\n0.5\n2.0\n1.5\n1.0\n2.5","2.5\nRAD-200m\n. Comparison of 200 m o ow values from the RAD sodar and the BAO sensor.\n2.0\n++\n1.5\n+\nSonic (m/s)\n+\n+\n1.0\n+\n+\n+\n+\n#8-7\n+\nINTERCEPT-0 AND SLOPE-1 (45° ANGLE) LINE\n+\n+\n0.5\nFigure 9.\n+\nNIGHT TIME OBSERVATIONS\n+\nDAY TIME OBSERVATIONS\nREGRESSION LINE\n0.0\n2.5\n2.0\n1.5\n1.0\n0.5\n0.0","2.5\nFigure 10. Comparison of 200 m ow values from the REM sodar and the BAO sensor.\nREM-200m\n2.0\n1.5\nSonic (m/s)\n+\n+\n++\n+\n1.0\n+\n+\n#\n+\n+\n+\nto\nINTERCEPT.0 AND SLOPE=1 (45° ANGLE) LINE\n+\nI\n+\n0.5\nNIGHT TIME OBSERVATIONS 0\nDAY TIME OBSERVATIONS +\nREGRESSION LINE\n0.0\n0.0\n0.5\n1.0\n2.0\n1.5\n2.5\n32","2.5\nXON-200m\nComparison of 200 m °W values from the XON sodar and the BAO sensor.\n2.0\n+\n1.5\n+\nSonic (m/s)\n+\n+\n+\n,4'+\n#\n+\n#H\n+\n+\n1.0\n++\n+\n#+++\n+\n+\n+\n#\n+\n+\n+\nINTERCEPT-0 AND SLOPE-1 (45° ANGLE) LINE\n++++\n+\n0.5\n+\nFigure 11.\n+\nNIGHT TIME OBSERVATIONS\nDAY TIME OBSERVATIONS\nREGRESSION LINE\n0.0\n2.5\n2.0\n1.5\n1.0\n0.5\n0.0","2.5\nFigure 12. Comparison of 300 m ow values from the AV sodar and the BAO sensor.\nAV-300m\n2.0\n+\n+\n1.5\nSonic (m/s)\n+\n+++\n+\n+\n+\n#\n+\n+\n+\n+\n1.0\n+\n+\n+\nINTEPCEPT-0 AND SLOPE=1 (45° ANGLE) LINE\n+\n0.5\n+\nNIGHT TIME OBSERVATIONS\nDAY TIME OBSERVATIONS\nREGRESSION LINE\n0.0\n0.0\n1.0\n0.5\n2.0\n1.5\n2.5","2.5\nRAD-300m\nComparison of 300 m o ow values from the RAD sodar and the BAO sensor.\n2.0\n+\n+\n1.5\n+\nSonic (m/s)\n+\n++\n+\n+\n+\n+\n+\n+\n+\n+\n+\n+\n1.0\n+\n+\n+\n+\n+\n+\n+\nINTERCEPT.0 AND SLOPE-1 (45° ANGLE) LINE\n+\n+\n0.5\n+\n+\nFigure 13.\nNIGHT TIME OBSERVATIONS 0\nDAY TIME OBSERVATIONS\nREGRESSION LINE\n0.0\n2.5\n2.0\n1.5\n1.0\n0.5\n0.0","2.5\nREM-300m\nsensor.\nFigure 14. Comparison of 300 m ow values from the REM sodar and the BAO\n2.0\n+\n1.5\nSonic (m/s)\n+ + +\n++\n++\n++\n+\n+\n1.0\n+\n+\n+\n+\nINTERCEPT=0 AND SLOPE-1 (45° ANGLE) LINE\n0.5\n+ +\nNIGHT TIME OBSERVATIONS a\nDAY TIME OBSERVATIONS\nREGRESSION LINE\n0.0\n0.0\n0.5\n1.0\n2.0\n1.5\n2.5","NIGHT DAY TIME OBSERVATIONS\nREGRESSION LINE\nINTERCEPT-@ OBSERVATIONS AND LINE\nTIME\n2.0\n0.0\n0.5\n1.0\n1.5\n2.5\n0.0\nFigure\n16. 100 BI)\n+\n0.5\n1.0\nSonic (m/s)\n1.5\n2.0\nRAD-100m\n2.5","NIGHT TIME OBSERVATIONS 0\nINTERCEPT.0 AND SLOPE-1 (45° ANGLE) LINE\nDAY TIME OBSERVATIONS\nREGRESSION LINE\n0.0\n0.5\n1.0\n1.5\n2.0\n2.5\n0.0\nFigure 15.\n+\n+\n0.5\nComparison of 300 m ow values from the XON sodar and the BAO sensor.\n++++\n+ +\n+\n0\n++\n1.0\n#\n+\nSonic (m/s)\n+\n+\n+\n1.5\n2.0\nXON-300m\n2.5","2.5\nsensor.\nRAD-200m\nComparison of 200 m °W values from RAD (mode: BI) sodar and the BAO\n2.0\n1.5\nSonic (m/s)\n1.0\n+\nINTERCEPT-0 AND SLOPE-1 (45° ANGLE) LINE\n0.5\nNIGHT TIME OBSERVATIONS 0\nFigure 17.\n+\nDAY TIME OBSERVATIONS\nREGRESSION LINE\n0.0\n2.5\n2.0\n1.5\n1.0\n0.5\n0.0","2.5\nComparison of 300 m ow values from RAD (mode: BI) sodar and the BAO sensor.\nRAD-300m\n2.0\n1.5\nSonic (m/s)\n+\n1.0\n+\n+\nINTERCEPT-0 AND SLOPE=1 (45° ANGLE) LINE\n0.5\nNIGHT TIME OBSERVATIONS 0\n000\nFigure 18.\n+\nDAY TIME OBSERVATIONS\nREGRESSION LINE\n0.0\n0.0\n1.0\n0.5\n2.5\n2.0\n1.5","sensor.\n2.5\nRAD-100m\nComparison of 100 m °W values from RAD (mode: MONO) sodar and the BAO\n2.0\n1.5\nSonic (m/s)\n+\n1.0\n++\n#\n+\n+\nINTERCEPT-0 AND SLOPE-1 (45° ANGLE) LINE\n0.5\nNIGHT TIME OBSERVATIONS 0\n+\nFigure 19.\nDAY TIME OBSERVATIONS\nREGRESSION LINE\n0.0\n2.5\n2.0\n1.5\n1.0\n0.5\n0.0","sensor.\n2.5\nRAD-200m\nComparison of 200 m °W values from RAD (mode: MONO) sodar and the BAO\n2.0\n++\n1.5\nSonic (m/s)\n+\n1.0\n+\n+\nINTERCEPT=0 AND SLOPE=1 (45° ANGLE) LINE\n7\n+++\n4\n0.5\nNIGHT TIME OBSERVATIONS D\nFigure 20.\n+\nDAY TIME OBSERVATIONS\nPEGRESSION LINE\n0.0\n0.0\n0.5\n1.5\n1.0\n2.5\n2.0","sensor.\n2.5\nRAD-300m\nComparison of 300 m ow values from RAD (mode: MONO) sodar and the BAO\n2.0\n+\n1.5\nSonic (m/s)\n+\n1.0\n+\nINTERCEPT-0 AND SLOPE=1 (45° ANGLE) LINE\n+\n++\n+\n+\n+\n0.5\n+\nNIGHT TIME OBSERVATIONS 0\n+\nFigure 21.\nDAY TIME OBSERVATIONS\nREGRESSION LINE\n0.0\n2.5\n2.0\n1.5\n1.0\n0.5\n0.0","sensor.\n2.5\nRAD-100m\nFigure 22. Comparison of 100 m °W values from RAD (mode: MULTI) sodar and the BAO\n2.0\n1.5\nSonic (m/s)\n+\n+\n+\n1.0\n+\n+\n+\n#\n+\n+\nINTERCEPT.0 AND SLOPE-1 (450 ANGLE) LINE\n0.5\nNIGHT TIME OBSERVATIONS 0\n+\nDAY TIME OBSERVATIONS\n+\n0\nREGRESSION LINE\n0.0\n0.0\n1.0\n0.5\n2.5\n2.0\n1.5","Comparison of 200 m o w values from RAD (mode: MULTI) sodar and the BAO sensor.\n2.5\nRAD-200m\n2.0\n1.5\n+\nSonic (m/s)\n+\n1.0\n+\n+\n+\nANGLE) LINE\n+\n+\n+\n0.5\nINTERCEPT-0 AND SLOPE-1 (45°\n+\nNIGHT TIME OBSERVATIONS\nFigure 23.\nDAY TIME OBSERVATIONS\nREGRESSION LINE\n0.0\n2.5\n2.0\n1.5\n1.0\n0.5\n0.0","sensor.\n2.5\nRAD-300m\nComparison of 300 m values from RAD (mode: MULTI) sodar and the BAO\n2.0\n1.5\nSonic (m/s)\n0\n+\n+\n+\n+\n+\n+\n1.0\n+\nINTERCEPT-0 AND SLOPE-1 (45° ANGLE) LINE\n0.5\n+\nFigure 24.\nNIGHT TIME OBSERVATIONS\nDAY TIME OBSERVATIONS\nREGRESSION LINE\n0.0\n1.0\n0.5\n0.0\n2.5\n2.0\n1.5","5. MEASUREMENT OF WIND SPEED\nMeasurements of wind speed (S) were obtained at sampling rates of 1\ndatum/15 S (REM, RAD, XON) and 1 datum/24 S (AV), and average S was computed\nin the field over 20 min intervals. The four sodars cycled in sequence, so\nthat a maximum of about 1440/4, or 360, values could be obtained by each ven-\ndor in the 20 days of the experiment. At the three heights under con-\nsideration (100 m, 200 m, 300 m), AV had complete data and the other\nmanufacturers ranged in completeness as follows: RAD, 60% to 96%, REM, 55% to\n72%; XON, 91% to 92%, depending on the variable.\nTo examine the accuracy and precision of the sodars, simultaneous obser-\nvations of wind speed were recorded from sonic anemometers at the same three\nheights on a tower that was about 600 m from the sodar systems. The sonic\nsystems had a sampling rate of 10 Hz, and they are regarded as the reference\ninstruments in the evaluation. However, owing to a wind shadow zone created\nby the tower, extending +40° from north for the sonic instruments, reference\ndata in this sector were obtained by the propeller-vane at the BAO tower. A\ncomparison of sonic and propeller wind speed measurements on the tower showed\nthat the instruments were approximately equivalent. The resulting sonic and\npropeller data sets are about equal in size at 200 m, but the sonic set has\none-third more data at 100 m and four-thirds more data at 300 m. Considering\nomissions in the reference data, these completeness percentages resulted: AV,\n83% to 91%; RAD, 55% to 88%; REM, 51% to 66%; XON, 75% to 84%.\n47","5.1 Sodar Reference Differences\nValues of sample bias (b), comparability (c), and standard deviation (s)\nand coefficient of variation (s' ) for the differences between sodar and\nreference values are presented in Table 5 for combined sodar observations at\neach height as well as for the sodar record of each vendor. Propeller wind\nspeeds were excluded when the wind speed was less than 1 m/s.\nThe estimates of bias in Table 5 show mostly negative values at 100 m\nand a composite value near -0.4 m/s. Since the difference is taken as (sodar\n- reference), this means that the sodar systems tend to register too low. An\nexception is RAD, which does not have a significant bias at 100 m.\nAt 200 m, the vendors all record too high, and at 300 m, RAD and XON again\nrecord too high whereas AV is slightly negative and REM unbiased. Biases were\ncomputed for day (0600-1800 hours) and night (1800-0600 hours) values. Most\ndifferences between day and night are insignificant.\nThe comparability (c) of sodar wind speeds with reference values is also\ngiven in Table 5. Precision is represented by standard deviation (s) and per -\ncentage deviation (s'). The s' values range from about 15% to 35% around com-\nposite values near 25%.\n5.2 Individual 20 Minute Average Values\nAdditional information about the characteristics of sodar measurements of\nS can be sought in the scatter diagrams of sodar 20 min average values plotted\nagainst reference values. Such plots are presented by height and by vendor in\nFigs. 25-36.\nEach chart has a broken line at 45° from the origin representing a slope\n48","of 1. The estimates of the correlation coefficient, slope, and intercept are\ngiven in Table 6.\nThe agreement between sodar and tower wind speed measurements is obviously\nquite good. Differences between manufacturers can be deduced by the reader.\n5.3 Sodar Modes\nA comparison between RAD modal data and corresponding 20 min reference\naverages yields the b, , C, , and S results of Table 7. The monostatic mode had\nsignificantly higher bias than the other two modes at all heights.\nThe comparability and standard deviation show a distinct advantage to the\nbistatic mode; MONO has the greatest magnitudes of C at 200 m and 300 m, and\nMULTI has the greatest magnitudes of S at 100 m and 300 m.\n49","Table 5. Sodar wind speed compared with reference wind speed\nS (m/s)\ns' (%)\nN\nC (m/s)\nb (m/s)\nVendor\nHeight\n1179\n1.21\n28\n-0.42\n1.28\nAll\n100 m\n21\n327\n0.90\n1.03\nAV\n-0.50\n315\n1.18\n28\n0.02\n1.18\nRAD\n14\n236\n0.60\n0.62\n-0.12\nREM\n301\n1.56\n37\n1.88\n-1.04\nXON\n23\n1019\n0.96\n0.14\n0.98\nAll\n200 m\n0.72\n17\n298\n0.72\nAV\n0.05\n35\n258\n1.00\n1.47\nRAD\n0.31\n194\n0.72\n17\nREM\n0.12\n0.73\n17\n269\n0.71\n0.70\nXON\n0.09\n27\n1005\n1.24\n1.23\nAll\n0.16\n300 m\n328\n1.15\n1.15\n25\nAV\n-0.10\n37\n198\nRAD\n0.29\n1.71\n1.69\n183\n0.74\n0.74\n17\nREM\n0.02\n1.12\n25\n296\nXON\n0.44\n1.20\n= bias (accuracy)\nb\nC = comparability\nS = standard deviation of differences (precision)\nis = S expressed as a percentage of average value of reference wind speed\nN = number of observations\nAV = Aerovironment\nRAD = Radian\nREM = Remtech\nXON = Xontech\n50","Table 6. Regression analysis for wind speed\nHeight\nVendor\nN\nBO\np\nB1\n100 m\nAV\n0.94\n-0.03\n0.89\n327\nRAD\n0.90\n0.41\n0.91\n315\nREM\n0.97\n-0.22\n1.02\n236\nXON\n0.82\n-0.07\n0.77\n301\n200 m\nAV\n0.96\n-0.02\n1.02\n298\nRAD\n0.87\n0.04\n1.07\n258\nREM\n0.96\n0.34\n0.95\n194\nXON\n0.96\n0.02\n1.01\n269\n300 m\nAV\n0.93\n0.45\n0.88\n328\nRAD\n0.85\n1.02\n0.84\n198\nREM\n0.96\n0.18\n0.96\n183\nXON\n0.93\n0.46\n0.97\n296\n= estimate of correlation coefficient\na\nBO = intercept term\nB1 = slope term\nN = number of observations\n51","Table 7. . Sodar Radian modes: Accuracy and precision for wind speed\nS (m/s)\nN\nC (m/s)\nb (m/s)\nMode\nHeight\n0.54\n90\n0.56\n-0.15\nBI\n100 m\n128\n1.53\n1.54\n-0.10\nMULTI\n1.04\n97\n1.18\n0.31\nMONO\n78\n0.90\n0.90\n-0.01\nBI\n200 m\n110\n1.62\n1.64\n0.27\nMULTI\n70\n1.63\n1.79\n0.74\nMONO\n1.26\n1.26\n66\n0.08\nBI\n300 m\n1.88\n92\nMULTI\n0.12\n1.89\n40\n1.96\n1.68\nMONO\n1.01\nb = bias (accuracy)\nC = comparability\nS = standard deviation of differences (precision)\nN = number of observations\n52","25\nAV-100m\nFigure 25. of 100 m wind speeds from AV sodar and BAO sensors.\n20\nSonic / Propvane (m/s)\n15\n10\nINTERCEPT-0 AND SLOPE=1 145 ANGLE LINE\n5\nNIGHT TIME OBSERVATIONS & D\nDAY TIME OBSERVATIONS\nREGRESSION LINE\n7\n6\n5\n6\n25\n20\n15\n10\n53","25\nRAD-100m\n20\nSonic/Propvane (m/s\n15\nwind\n10\n26.\n5\nFigure\nOBSERVATIONS\nINTERCEPT-@\nREGRESSION\n0\nTIME\n5\n0\n25\n20\n15\n10\nDAY\n54","25\nREM-100m\nFigure 27. Comparison of 100 m wind speeds from REM sodar and sensors.\n20\nSonic / Propvane (m/s)\n15\n+\n10\n+\nINTERCEPT=@ AND SLOPE.1 (45 ANGLE) LINE\n5\nHIGHT TIME OBSERVATIONS 2\nDAY TIME OBSERVATIONS\nREGRESSION LINE\n+\n0\n5\n0\n25\n20\n15\n10","25\nXON-100m\n20\n10\n5\n25","25\nAV-200m\n20\nSonic/Propvane in.kl\n15\n10\n5\n0\nREGRESSION\n5\n0\n25\n20\n15\n10\nDAY\n57","25\nRAD-200m\n30. Comparison of 200 m wind speeds from RAD sodar and BAO sensors.\n20\nSonic / Propvane (m/s)\n15\n+\n10\nINTERCEPT-0 AND SLOPE=1 (450 ANGLE) LINE\n5\nFigure\nNIGHT TIME OBSERVATIONS\nDAY TIME OBSERVATIONS\nREGRESSION LINE\n+\n1\n0\n5\n0\n20\n15\n10\n25","25\nREM-200m\nsensors.\n31. Comparison 200 m wind speeds from REM sodar and BAO\n20\nSonic / Propvane (m/s)\n15\n10\n+\nINTERCEPT-@ AND SLOPE-1 (150 ANGLE) LINE\n5\nFigure\nNIGHT TIME OBSERVATIONS\nDAY TIME OBSERVATIONS\n+\nREGRESSION LINE\n++\n7\n0\n5\n0\n25\n20\n15\n10","25\nXON-200m\nsensors.\n32. m wind from XON BAO\n20\nSonic/Propvane (m/s)\n15\n10\nINTERCEPT-@ AND 0 ANGLE) LINE\n5\nFigure\n+\nOBSERVATIONS\nDAY TIME OBSERVATIONS\nREGRESSION LINE\nTIME\n0\n5\n0\n10\nNIGHT\n20\n15\n25\n60","25\nAV-300m\n20\nSonic/Propvane im/si\n15\n10\n33.\n5\nFigure\nOBSERUATIONS\nOBSERVATIONS\nAND\nDAY REGRESSION\n0\n5\n0\n25\n20\n15\n10\n19","25\nRAD-300m\n+\n20\n34. of wind from RAD and\n+\nSonic / Propvane (m/s)\n15\nof\n0\n10\nINTERCEPT-0 AND D ANGLE) LINE\n5\nFigure\nOBSERVATIONS\nNIGHT DAY TIME OBSERVATIONS\nPEGPESSION LINE\nTIME\n0\n5\n0\n15\n25\n20\n10\n62","25\nREM-300m\nsensors.\n35. m wind REM sodar and BAO\n20\nSonic / Propvane (m/s)\n15\n10\n+\nINTERCEPT-@ OBSERVATIONS AND D LINE\n5\nFigure\nHIGHT DAY TIME OBSERVATIONS\nREGRESSION LINE\n+\nTIME\n7\n0\n5\n0\n25\n20\n15\n10\n63","25\nXON-300m\n20\n10\n5\nB\n25\n64","6. MEASUREMENT OF WIND DIRECTION\nAll four sodar vendors reported wind direction (D). At the three heights\nunder consideration (100 m, 200 m, 300 m), AV had complete data, and the other\nmanufacturers ranged in completeness as follows: RAD, 60% to 96%; REM, 55% to\n72%; XON, 91% to 92%. However, the wind shadow zone mentioned in Sec. 4\nlimited the amount of data that could be used. An investigation of the\npropeller-vane data indicated that they could not be substituted as reference\nvalues for wind direction. There were unexplained disparities between the\npropeller and sonic data that are still under investigation. Because sonic\ndata showed more consistent behavior and we therefore believe them to be more\nreliable, only the sonic wind direction measurements are used as reference\ndata, resulting in these completeness percentages for sodar/sonic differences:\nAV, 43% to 63%; RAD, 36% to 49%; REM, 31% to 39%; XON, 36% to 55%.\n6.1 Sodar Reference Differences\nValues of sample bias (b), comparability (c), and standard deviation (s)\nfor the differences between sodar and sonic reference values are presented in\nTable 8 for combined sodar observations at each height as well as for the\nsodar record of each vendor.\nThe values of bias in Table 8 show negative values at 100 m and 200 m for\nall vendors, but positive values at 300 m for all vendors except for RAD.\nHowever, most of these values are not significantly different from zero, con-\n65","sidering the variability of the wind direction and the number of cases\nincluded. Day and night biases given in Table 9 show some differences, but\nfor the most part they are not statistically significant.\nThe comparability (c) of sodar wind direction with sonic reference values\nis also given in Table 8. The low values at 200 m may be due to the loss of\ndata at that height during a storm on 13 September. Considering the scatter\nin data, it can be assumed that the vendors' measurements of wind directions\nare equivalent to each other.\n6.2 Individual 20 Minute Averaged Values\nScatter diagrams of 20 min averaged sodar values plotted against reference\nvalues are given in Figs. 37-48. For the most part, each sodar datum agreed\nwith the tower datum quite well. There were some notable exceptions, however,\nespecially during the night. XON seems to have predicted many more north\nwinds than actually occurred at 100 m and 300 m, but its agreement at 200 m is\nvery good.\nFrom the figures it seems that most of the vendors predicted more\nnortherly winds at night at 100 m than actually occurred, especially when the\ntower measured winds from the east and south. This does not seem to be the\ncase at 200 m or 300 m and may be due to local terrain effects, rather than\ninstrument problems\nCorrelation coefficients are given in Table 10, with intercepts and slopes\nof the linear regression lines. With two exceptions, p values are 0.9 or\ngreater.\n66","6.3\nSodar Modes\nThe three modes of operation for the RAD system showed no appreciable dif-\nference in their ability to measure wind direction.\nTable 8. Sodar wind direction compared with sonic wind direction\nHeight\nVendor\nb (deg)\nC (deg)\nS (deg)\nN\n100 m\nAll\n-4.41\n28.59\n28.25\n667\nAV\n-3.87\n26.70\n26.41\n187\nRAD\n-6.76\n27.06\n26.20\n177\nREM\n-2.03\n18.51\n18.40\n137\nXON\n-4.49\n37.85\n37.58\n166\n200 m\nAll\n-3.43\n23.22\n22.97\n523\nAV\n-0.79\n19.47\n19.45\n155\nRAD\n-7.86\n25.67\n24.43\n128\nREM\n-3.89\n24.67\n24.36\n110\nXON\n-1.85\n23.80\n23.73\n130\n300 m\nAll\n0.75\n29.59\n29.58\n697\nAV\n0.26\n28.56\n28.56\n227\nRAD\n-3.25\n29.98\n29.80\n131\nREM\n0.62\n19.70\n19.69\n142\nXON\n4.05\n35.96\n35.73\n197\nb = bias (accuracy)\nC = comparability\nS = standard deviation of differences (precision)\nN = number of observations\n67","Table 9. Test of day versus night wind direction bias\nN (night)\nb (night)\nN (day)\nb (day)\nVendor\nHeight\n294\n373\n-2.11\nAll\n-6.23\n100 m\n75\n112\n0.53\nAV\n-6.81\n86\n91\n-5.20\n-8.24\nRAD\n61\n76\n-0.66\n-3.13\nREM\n72\n94\n-2.41\nXON\n-6.09\n72\n94\n-0.84\nAll\n-5.75\n200 m\n69\n86\n0.06\n-1.48\nAV\n62\n66\n-4.19\n-11.31\nRAD\n53\n57\nREM\n-5.25\n-2.43\n62\n68\nXON\n-6.18\n2.89\n309\n2.24\n388\nAll\n-0.44\n300 m\n2.00\n131\n96\nAV\n-1.01\n61\n-4.20\n70\nRAD\n-2.42\n66\n76\n1.49\nREM\n-0.13\n86\n111\n7.66\nXON\n1.26\nTable 10. Regression analysis for wind direction\nN\nVendor\nB1\nHeight\nBO\np\n187\n-0.25\n0.98\nAV\n0.94\n100 m\n177\n5.20\n0.94\n0.92\nRAD\n137\n4.26\n0.98\nREM\n0.96\n166\n0.85\n8.29\n0.96\nXON\n155\n8.15\n0.95\nAV\n0.97\n200 m\n128\n14.07\n0.90\nRAD\n0.90\n110\n16.72\n0.91\nREM\n0.93\n130\n0.98\n2.37\nXON\n0.95\n1.05\n227\n-5.86\n300 m\nAV\n0.93\n131\n1.03\n-8.87\nRAD\n0.93\n142\n1.03\n-4.64\nREM\n0.97\n197\n1.09\n-8.79\nXON\n0.89\n= estimate of correlation coefficient\np\nBO = intercept term\nB1 = slope term\nN = number of observations\n68","360\n37. Comparison of 100 m wind directions from AV sodar and BAO sensors.\nAV-100m\n300\n240\nSonic (deg)\n180\n+\n+\n+\n120\nINTERCEPT=0 AND SLOPE-1 (45° ANGLE) LINE\n+\n60\nFigure\nNIGHT TIME OBSERVATIONS 0\n+\nDAY TIME OBSERVATIONS\nPEGRESSION LINE\n0\n0\n60\n360\n300\n240\n180\n120","360\nRAD-100m\n38. of 100 m wind directions from RAD sodar and BAO sensors.\n300\n240\n+\nSonic (deg)\n180\n120\nINTERCEPT=@ AND 0 ANGLE) LINE\n60\nFigure\nHIGHT TIME OBSERVATIONS\nDAY TIME OBSERVATIONS LINE\n0\n0\nPEGPESSION\n60\n180\n120\n360\n300\n240\n70","360\nREM-100m\n39. Comparison of 100 m wind directions from REM sodar and BAO sensors.\n300\n240\nSonic (deg)\n180\n+\n+\n120\n+\n#\n+\n+\nINTERCEPT-@ AND SLOPE-1 (45° ANGLE) LINE\n+\n60\n+\nNIGHT TIME OBSERVATIONS + 0\nFigure\nDAY TIME OBSERVATIONS\nREGRESSION LINE\n0\n0\n60\n360\n300\n240\n180\n120\n71","360\nXON-100m\nFigure 40. Comparison of 100 m wind directions from XON sodar and BAO sensors.\n300\n240\n+\nSonic (deg)\n+\n180\n+\n+\n+\n+\n+\n120\n+\n+\nINTERCEPT.0 AND SLOPE-1 (45° ANGLE ) LINE\n+\n+\n60\nNIGHT TIME OBSERVATIONS\nDAY TIME OBSERVATIONS\nREGRESSION LINE\n0\n0\n60\n240\n180\n120\n360\n300","360\nAV-200m\n41. Comparison of 200 m wind directions from AV sodar and BAO sensors.\n300\n240\n+\nSonic (deg)\n180\n+\n+\n120\nINTERCEPT-0 AND SLOPE-1 (45° ANGLE) LINE\n+\n++\n#\n60\n+\n+\n0\nFigure\n+\nNIGHT TIME OBSERVATIONS\nDAY TIME OBSERVATIONS\nPEGRESSION LINE\nT\n0\n0\n60\n360\n300\n240\n180\n120\n73","360\nRAD-200m\nsensors.\nm wind directions RAD sodar and BAO\n300\n+\n+\n240\nSonic (deg)\n180\n+\n120\nINTERCEPT-@ OBSERVATIONS AND SLOPE.1 (450 0 ANGLE) LINE\n60\n42.\nFigure\nNIGHT DAY TIME OBSERVATIONS\nPEGPESSION LINE\n0\n0\nTIME\n60\n180\n120\n360\n300\n240\n74","360\nREM-200m\nsensors.\n43. Comparison of 200 m wind directions from REM sodar and BAO\n300\n240\n+\nSonic (deg)\n+\n+\n180\n+\n+\n+\n+\n120\nINTERCEPT=0 AND SLOPE-1 0 (45° ANGLE) LINE\n+ +\n+\n60\nFigure\nHIGHT TIME OBSERVATIONS\nTIME OBSERVATIONS\nDAY REGRESSION LINE\n0\n0\n60\n360\n300\n240\n180\n120\n75","360\nXON-200m\n44. Comparison of 200 m wind directions from XON sodar and BAO sensors.\n300\n240\nSonic (deg)\n180\n+\n+\n+\n120\nINTERCEPT-0 AND SLOPE-1 (45° ANGLE) LINE\n+\n60\nNIGHT TIME OBSERVATIONS 0\nFigure\nDAY TIME OBSERVATIONS\nREGRESSION LINE\n0\n0\n60\n180\n120\n360\n300\n240","360\nAV-300m\nComparison of 300 m wind directions from AV sodar and BAO sensors.\n300\n+\n240\n+\n0\nSonic (deg)\n+\n180\n120\n+\n+\nINTERCEPT-0 AND SLOPE.1 (45° ANGLE) LINE\n+\n60\n+\nFigure 45.\nHIGHT TIME OBSERVATIONS 0\nDAY TIME OBSERVATIONS\nREGRESSION LINE\n0\n0\n60\n360\n300\n240\n180\n120\n71","360\nRAD-300m\nsensors.\n46. of 300 m wind directions from RAD sodar and BAO\n300\n+\not\n240\n+\nSonic (deg)\n180\n+\n+\n120\n+\nINTERCEPT.@ AND SLOPE-1 (45° ANGLE) LINE\n+\n60\nFigure\nNIGHT TIME OBSERVATIONS\nDAY TIME OBSERVATIONS\nREGRESSION LINE\n0\n0\n60\n360\n300\n240\n180\n120","REM-300m\n360\n47. 300 m wind directions from REM sodar and BAO sensors.\n300\n240\nSonic (deg)\n180\n+\n+\n+\n+\n++\n120\nDX\n+\n+\n+\n+++\n+\nINTERCEPT.@ AND SLOPE 1 ( 15 ANGLE) LINE\n#+\n60\nNIGHT TIME OBSERVATIONS 0\nFigure\nDAY TIME OBSERVATIONS\nREGRESSION LINE\n7\n0\n0\n60\n360\n300\n240\n180\n120","360\nXON-300m\n48. Comparison of 300 m wind directions from XON sodar and BAO sensors.\n300\n240\nSonic (deg)\n+\n180\n+\n+\n120\n+\nINTERCEPT.0 SLOPE-1 (45° ANGLE) LINE\n+\n+++\n60\n+\nFigure\nNIGHT TIME OBSERVATIONS\nDAY TIME OBSERVATIONS\nREGRESSION LINE\n0\n0\n60\n180\n120\n360\n300\n240","7. SODAR RAWINSONDE COMPARISONS\nAlthough our formal evaluation of the sodars was limited to the three\nobserving levels (100, 200 and 300 m) within the range of the tower instru-\nmentation, the vendors submitted data from higher levels in the atmosphere on\nsome occasions. These data were compared with the rawinsonde data obtained\nconcurrently. Values of sample bias, comparability, and correlation for the\nmeasured wind speeds and wind directions for each sodar are given in Table 11.\nOnly measurements at 200 m and above are included in the statistics because of\nthe limitations in the rawinsonde's accuracy below that height. These com-\nparisons serve two objectives: to assess at least in a qualitative sense the\nability of sodars to track the broad features in the speed and direction pro -\nfiles above 300 m; and to determine the magnitude and nature of the uncertain- -\nties encountered in comparing measurements from a single rawinsonde traverse\nwith time-averaged (20 min) sodar measurements.\nThe numbers in Table 11 are of the same order as the numbers in earlier\ntables comparing sodar and tower measurements. Agreement with the rawinsonde\nis better for some sodars than for others. The differences are not considered\nsignificant, given the fact that this experiment was designed, and the anten-\nnas set up, for performance evaluation in the first 300 m.\nSome idea of the agreement in the data can be obtained from the scatter\ndiagram in Fig. 49. The wind speed and direction measurements from one of the\nsodars (AV) is plotted against the rawinsonde measurements. The scatter is\n81","Table 11. Sodar wind speeds and directions for all heights\ncompared with rawinsonde speeds and directions\nN\nb\nSensor\nC\nVariable\np\n1.17\n0.86\n61\n0.37\nAV\nSpeed\n14\n0.92\n0.61\n-0.09\nRAD\n(m/s)\n19\n0.24\n0.91\n0.94\nREM\n2.34\n0.38\n31\n-1.25\nXON\n61\n4\n44\n0.90\nAV\nDirection\n0.89\n14\n18\n40\n(deg)\nRAD\n0.95\n19\nREM\n- 8\n37\n31\n39\n0.96\nXON\n15\nb = bias\nC = comparability\np = estimate of correlation coefficient\nN = number of observations\n82","10\nWind Speed\n8\n6\n4\n2\n0\n0\n2\n4\n6\n8\n10\nS sodar (m/s)\n360\nWind Direction\n270\n180\n90\n0\n0\n90\n180\n270\n360\nD sodar (deg)\nFigure 49. o\nComparison of wind speeds and wind directions from rawin-\nsonde and sodar measurements.\n83","Figure 50. Wind speed and wind direction profiles from sodar (AV), rawinsonde, and tower measurements\n360\n320\n280\n240\nDirection (deg)\n200\n160\n120\n80\n40\n0\n0\n400\n200\n1200\n1000\n800\n600\n10\n9\nin a stable boundary layer.\n8\nRawinsonde\n7\nTower\nSpeed (m/s)\nSodar\n6\n5\n4\n8 September 1982\n0403-0423 MST\n3\n2\n1\n0\n400\n200\n0\n1000\n800\n600\n1200","Figure 51. Wind speed and wind direction profiles from sodar (AV), rawinsonde and tower measurements\n360\n320\n280\nDirection (deg)\n240\n200\n160\n120\n80\n40\n0\n1200\n1000\n800\n600\n400\n200\n0\n10\n9\nin a convective boundary layer.\n8\nRawinsonde\n7\nTower\nSodar\nSpeed (m/s)\n6\n5\n4\n15 September 1982\n1204-1224 MST\n3\n2\n1\n0\n1200\n1000\n800\n600\n400\n200\n0","not significantly larger than in the plots against the tower measurements. It\nis apparent, however, that the wind speeds agree better when they are less\nthan 4 m/s, although the percentage error is approximately the same. The\nincrease in scatter above 4 m/s suggests that the larger spatial separation\nintroduced between sensing volumes at the larger wind speeds (with the rawin-\nsonde drifting farther away from the release point) is a factor to be\nrecognized when rawinsondes are used for evaluating sodar performance.\nAnother factor to be recognized (but not obvious in Fig. 49) is the possibi -\nlity of large wind direction differences in sodar rawinsonde comparison under\nlight wind conditions. Wind directions under such conditions tend to be\nhighly variable both spatially and temporally. These conditions occur fre-\nquently at the BAO when the winds are from E to SE. One cannot expect good\nagreement between the hear-instantaneous and the time-average measurements\nfrom the two systems during periods of light winds. This point is brought\nhome very clearly in the speed and direction profiles of Figs. 50 and 51.\nWhen wind speeds drop below 2 m/s, wind direction differences become large\nregardless of stability. When the wind speeds are larger, the agreement be-\ntween the rawinsonde and sodar profiles is good. The two cases presented here\nare perhaps more spectacular in terms of the wind speed effect on the com-\nparison than most of the other cases examined. Over more complicated terrain,\ndifferences in speed and directions between sodars and rawinsondes could be\nmuch larger. Caution must be exercised, therefore, in interpreting data that\ncompare sodars with rawinsondes. Nonetheless, this evaluation does indicate\nthat, under proper conditions, reasonable agreement can be expected between\nthe two sets of measurements as though both techniques measured bulk proper-\nties of the wind with reasonable accuracy.\n86","8. CHARACTERISTICS OF SODAR W SPECTRA\nThe spectra of the vertical wind speed derived from sodar Doppler measure-\nments should, in principle, correspond to spectra from the sonic anemometers,\nsubject to the effects of spatial averaging and aliasing. Spatial averaging\nattenuates fluctuations with scales smaller than the dimensions of the\nsampling volume. Aliasing folds energy left over at frequencies above no' the\nNyquist frequency (= 1/2 sampling rate), back into the available spectral\nbandwidth (0 In the presence of spatial averaging, the energy\nfolded back is reduced by the amount lost through averaging. A schematic\nrepresentation of the distortions introduced on a typical W spectrum for two\ndifferent sampling rates is given in Fig. (a). In this example, the atten-\nuation from spatial averaging is assumed to commence at frequency no = 0.02\nHz. The wavelength 1, corresponding to this frequency (11=U/n1, where U is\nthe mean horizontal wind component) is roughly 2 times the longest dimension\nin the sampling volume. (A sampling volume 40 m diam X 40 m long is assumed\nhere with U = 5 m/s. ) Because of the sharp spectral attenuation above 0.02\nHz, aliasing is confined primarily to the first fold, which merely raises the\nenergy near no by a factor of 2. For typical beamwidths used in most sodar\noperations, no n 1 22 22 0.02 Hz at 100 < Z < 300 m for moderate wind speeds.\nIn the convective boundary layer the percentage of spectral energy con-\ntained in frequencies above no increases as height, Z, decreases. Conse-\nquently the uncertainties in the observed spectral forms and in the measured\n87","variances also increase as the height decreases. Figure 52(b) shows the\nprogression of the spectrum on a typical day. The frequency at the n Sw(n)\nspectral maximum, n'm' is nearly constant above 0.25 Zj, , (where Zi is the\nboundary layer depth) and varies inversely with height below that. Within the\nheight range of most sodar systems, the wavelength at the spectral peak can be\napproximated by\n(Z < 0.25 Z )\n6Z,\n(5)\n(Z > 0.25 Z )\n1.5Z\n.\nSpectral energy in the observed bandwidth also drops with decreasing Z.\nThe attendant decrease in signal-to-noise ratio in the sodar measurements\nserves to increase further the uncertainty in the spectral and variance (ow2)\nestimates.\nIn the stable nocturnal atmosphere, the W spectral scales and intensities\nare more strongly controlled by stratification than by Z. Over flat terrain,\nwithin the stable boundary layer (Kaimal et al. 1972) one can approximate\nAm using\n(6)\n22 L, for << Z , ,\nwhere L is the Monin-Obukhov length. Within the height range of our com-\nparisons, ^m would be roughly an order of magnitude smaller than under\nunstable conditions. There is proportionally less energy within the spectral\nbandwidth, so one can expect to find larger uncertainties and errors in the\nnighttime spectra than in the daytime spectra. This may account for the\nincreased scatter in the nighttime °W values in Sec. 4. An improvement in\n88","1\n10\n(a)\no\n10\nAliased\nspectra\n=\n10-1\nE\nTrue\nspectrum\n-2\n10\nSpectrum\n3\nattenuated by\n10\nspatial averaging\n.4\n10-4\n10\n(b)\no\n10\n10-1\n-2\n10\nZ=200m\n100m\n50m\n-3\n10\n10\n10-4\n-3\n10-2\n10-1\n0\n1\n10\n10\n10\nn (Hz)\nFigure 52. (a) Schematic representation of distortions introduced in\nthe W spectrum from attenuation due to spatial averaging\nand from aliasing.\n(b) Shift in spectral behavior with height and its implica-\ntions for sampling and aliasing errors.\n89","accuracy is possible in the presence of strong gravity waves because of its\nlarge contribution to variance at frequencies below no.\nThe spectra presented in Figs. 53 and 54 were computed from time series\nprovided by AV. No significance is attached to the choice of AV. The outputs\nare treated as generic signals from a Doppler sodar. The absence of liftup at\nthe high end implies extensive influence of spatial averaging at frequencies\nbelow n 0 .\nThe sonic spectra in Fig. 54(a) and (b) illustrate the effect of stability\non spectral wavelengths and intensities at the 300 m level. The sodar\nspectrum shows poor agreement with the sonic spectrum at night; spectral\nlevels are greatly enhanced. The high ow levels at night in Sec. 4 can now\nbe traced back to this distortion. To determine how much of this distortion\ncomes from aliasing, the sonic time series was converted to grab samples every\n24 S. The resulting spectrum, also shown in Fig. 54(b), has the same shape as\nthe sodar spectrum, but one-half the energy.\nMore precise estimates of the contributions from aliasing and other fac- -\ntors, such as spatial averaging and noise, can be made from the variances\nlisted in Table 12. Sonic anemometer variances estimated over two bandwidths,\n0-5 Hz and 0-0.02 Hz, are listed alongside the sodar variances. Sodar vari- -\nances appear to be 10% - 15% lower than the full range (5 Hz) sonic variances\nduring the day but 15% - 20% higher than the sonic variances integrated to\n0.02 Hz (see Table 12). From Table 13 (last column) we find the sonic anemom-\neter variance in the band 0.02 < n < 5 Hz to be between 20% and 25% of the\ntotal variance (0 < n < 5 Hz) under convective conditions. If all that\nvariance were to be aliased back into the frequency range 0 < n < 0.02 Hz, the\nratio (ow2) 2 sodar/(ow2) son (0