SARS 2014 MONOSTATIC AND BISTATIC SODAR ERRORS DUE TO SOUND BEING CARRIED DOWNSTREAM (WIND DRIFT) Stuart Bradley and Alex Strehz Physics Department, University of Auckland SARS 2014 MOTIVATION Over the years there have been a number of papers on acoustic refraction influences on sodar winds. There have also been a few papers on wind drift, or the effect of the transmitted sound being carried downwind. Refraction and wind drift both result in scattering from a place other than straightline propagation, so will affect the scattering angle and hence the Doppler shift. Predicted errors due to these “beam wander” effects are large. But the errors are not seen in sodar data. However, we noticed large effects in two new sodar designs from our lab: a bi-static sodar and an ‘urban’ sodar which has a wide transmitter beam and narrow receiver beams. None of the earlier works developed a theory or model which included the finite beam width which sodars have. Also none of the earlier works produced measurements which unequivocally demonstrated beam wander. We felt it was time to really understand what is going on. [email protected] 2 SARS 2014 WIND DRIFT REFRACTION z Height z Height z t t For example, a constant sound speed gradient gives propagation in a circular arc of radius R c e.g. for a temperature gradient dT/dz = 0.01 K m-1, and transmitter zenith angle t = 18, then R = 2x105 m. Also z R so the ray zenith angle at the receiver has changed by only 0.06 for scattering from z = 100 m. [email protected] For constant wind, the upward propagation direction does not change, but the sound is blown downwind (blue arrows). A greater backward zenith angle, r , is needed for the reflected ray to reach the monostatic receiver, since the reflected ray is also blown downwind. dc sin t dz sin z sin t r Here 1 1 tan r tan t M cos t cos r where M = u/c is the Mach number. For t = 18, and u = 10 m s-1, t = 21, and change of 3. CONCLUSION: Wind drift is much bigger effect than refraction. 3 SARS 2014 2D WIND DRIFT Height z t r Assuming small angles, r t+2M Doppler shift f /fT = -(sinr + sint)M -2Msin t-2M 2 giving a fractional error of M/sint For example, if u = 10 m s-1 and t = 18, then u/u = 0.095. This is a 9.5% error in estimated wind [email protected] 4 SARS 2014 2D WIND DRIFT WITH FINITE BEAMS Height z Lower Transmit Doppler Receive beam Higher receive Doppler Shifted Transmit beam The spectral peak is from a region of lower transmit Doppler and higher receiver Doppler. If the beam widths are the same, there is no resulting error [email protected] Height z Lower Transmit Doppler Higher receive Doppler Receive beam Transmit beam With asymmetry, the wider beam ‘wins’. M 2r 2r Fractional wind speed error 1 1 sin t t2 t2 1 5 SARS 2014 MAGNITUDE OF ERROR (2D) Fortunately, monostatic sodars use the same transducers for transmit and receive so r= t. Fractional error in estimated wind 0.06 0.04 5 m/s 10 m/s 0.02 0 -0.02 -0.04 -0.06 -0.08 -0.1 0 Ray solution r= 0 0.5 1 r/t 1.5 2 M 2r 2r Fractional wind speed error 1 1 sin t t2 t2 [email protected] 1 6 SARS 2014 3D CASE In practice, beams are 3D. Allowing for a bi-static receiver baseline b, and a general Mach vector M, z 2 f r t R t 1 2 R M M 1 M M fT r t R t R R Mt t For a phased-array bi-static sodar the drift error is zero if t = r . But this is not generally the case. R 1.0 r x 0.8 y b Receiver 40m upwind Mr -0.02 -0.01 Height z Normalised spectral power 0.6 0.4 Cross-wind 0.2 0.0 0.00 df / fT Receiver 40m downwind 0.01 Transmit beam Receive beam 0.02 Bi-static sodar with z = 80 m and t = r . [email protected] 7 BI-STATIC SODAR DATA SARS 2014 Asymmetric transmitter and receiver. Doppler shift measurements when r= t = 5, compared with Doppler for r= 18 and t = 5 250 Apparent height of scattering volume u r uh/c [email protected] h/tanr b Height from timing [m] h 200 h+(h2+b2)1/2=ct h = h[1+M(1+sinr)/cosr] 150 100 Zero wind, zero gradients 50 Measurements 0 0 uh/(c sinr) 50 100 150 Height from steering [m] h=btan r 200 250 8 SARS 2014 MONOSTATIC 3D Normalised spectral power 1.0 1.0 0.8 0.6 t = 5, r = 2 0.8 0.6 t = 5, r = 5 0.4 0.4 0.2 0.0 0.00 df/fT -0.01 t = r Wind at 0 t = 2, r = 5 With vertical transmission, u = 10 m s-1 [email protected] Normalised spectral power 0.01 -0.03 -0.02 t = r 0.2 t = r Wind at 180 Wind at 90 0.0 -0.01 0.00 df / fT 0.01 0.02 0.03 Transmission in x-z plane at 18 zenith, M = 10 m s-1/ 340 m s-1 9 URBAN SODAR SARS 2014 A monostatic sodar with • Transmitter = single vertical beam • Receiver = N beams tilted at r • Wind speed error 1/(Nsinr) is similar to conventional monostatic sodars, but r is smaller and N is larger. Receiver zenith angle r • Advantages: 1. measurement volumes closer 2. single transmit beam means faster sampling 18 Equivalent to 3 beam, 18 deg 15 Urban design 12 Equivalent to 3 beam, 30 deg 9 6 3 0 0 [email protected] 8 16 24 Number N of receiver beams 32 10 SARS 2014 URBAN SODAR PROBLEM #1: GAIN Transmitted beam pattern Receiver beams No wind [email protected] Beam drift 11 SARS 2014 URBAN SODAR PROBLEM #2: t r 1.0 0.8 0.6 0.4 0.2 -0.02 -0.01 0.0 0.00 df / fT 0.01 0.02 Downwind receivers have smaller magnitude Doppler than upwind receivers. The cross-wind receivers show wind approaching them. However the shift is accurately described by (see the purple arrow). [email protected] 2 2r 2 M 1 2 t 2 r 1 2 t 1 12 SARS 2014 SUMMARY • We have developed an accurate, general, 3D theory for wind drift • The theory includes finite beam width for receiver and transmitter • A simple 2D thin-slice model (with wind in the plane of the transmitter and receiver beams) gives a simple analytic expression for wind drift in terms of beam widths • It is found that, for this 2D case, monostatic sodars with matched transmitter and receiver, there is no wind drift correction (but there is loss of gain!) • The 3D model does predict small drift effects on monostatic tilted beams • Measurements on a bi-static sodar are consistent with these models • Measurements on the urban sodar (vertical beam transmitted, narrow tilted receiver beams) show wind drift effects • It appears that the 2D thin-slice model can be used to correct wind drift error [email protected] 13
© Copyright 2024