Upper Air Wind Measurements by Weather Radar Iwan

Upper Air Wind Measurements by Weather Radar Iwan Holleman, Henk Benschop, and Jitze vd Meulen l l l l Contents: Introduction to Doppler Radar Velocity Azimuth Display (VAD) Volume Velocity Processing (VVP) Implementation and quality control Verification against radiosonde and Hirlam Conclusions

KNMI Weather Radar l l Only in presence of scatterers Measures radial velocity as function of range, azimuth, and elevation Unambiguous velocity is limited outliers due to dealiasing technique (dual-PRF) Assumption: wind speed = observed velocity

Weather Radar Wind Profiles l l l Volume scan of radial velocity Algorithm for analysis Wind profile above radar site

Velocity Azimuth Display (VAD) ff=20. 4 m/s dd=217 deg

Doppler Radar Volume Scan l Height layer l 10 Elevations between 0. 5 and 25 degrees Multiple VADs per height layer

Retrieval Techniques l l l VAD 1 -2: Fourier expansion of radial velocity as a function of azimuth : Vrad( ) = a 0/2 + b 1 sin + a 1 cos + b 2 sin 2 + a 2 cos 2 +. . . VVP 1 -3: Use all available volume scan data for a given height layer at once Assumption of linear wind model at radar site: Ø Ø Ø U(x, y, z) = u 0 + x u/ x + y u/ y + (z-z 0) u/ z V(x, y, z) = v 0 + x v/ x + y v/ y + (z-z 0) v/ z W(x, y, z) = w 0 + (z-z 0) w/ z

Weather Radar Wind Profiles l l 8 January 2005 Doppler Radar and Hirlam analyses

Implementation and Quality Control l l l Volume scan data are input to retrieval techniques Application of minimum and maximum range Removal of radial velocities close to zero Binning of scan data in 8 azimuth sectors: neighboring empty sectors are not allowed Linear fit using Singular Value Decomposition (SVD) VAD: parameters per height layer are determined using weighted average Removal of radial velocity outliers and re-run of wind model fit

Verification of Wind Profiles l l Verification of Weather Radar Wind Profiles against radiosonde and Hirlam 9 Months of volume scan data, radiosonde and Hirlam profiles Comparison of different wind profile retrieval algorithms Ø Availability Ø Quality Impact of quality control procedures

Availability of VVP profiles 40% 28% std. dev. 2 m/s 19%

VAD and VVP against radiosonde

VVP and radiosonde against Hirlam

Conclusions l l Weather radar provides wind profiles VVP algorithm performs better than VAD Availability of wind profiles is reasonable Weather radar winds are at least as good as radiosonde winds More info: I. Holleman, JAOT 2005, in press