Potential of multi-frequency Doppler spectra for rain, snow, and ice cloud studiesCurrent Limitations due to Doppler spectra quality Stefan Kneifel, PavlosKollias- McGill Frederic Tridon – Univ. Leicester Ed Luke - BNL
Whyusing Doppler spectrainsteadofmomentsandwhy multi-frequency? • Considering just microphysics, Doppler Spectradepend on: • Particlesizedistribution N(D) • Size – fall velocityrelation v(D) • Size – backscatteringrelation • N(D) and v(D) arefrequencyindependentandthus, thespectrashouldperfectlymatch in the„small“ particleregion (slowfalling Rayleigh scatterers) andincreasinglydeviatefromeachotherfor larger andfasterfallingparticles.
Case of rain with the Ka-W band combination • Scattering with the Ka-W band combination • Rayleigh conditions not satisfied as a whole • But, smallest drops scatter in the Rayleigh regime their contribution on the DWR depends on differential attenuation only • Doppler spectra ratio (DSR) • Drops sorted according to their fall velocity and size with Vt=f(D) (Atlas et al., 1973) • The DSR emphasizes the two scattering regimes • Rayleigh regime plateau • Mie region (with two peaks) Quasi universal pattern
Confirmation with KAZR and WSACR data Presentation of the DSR technique during MC3E session on Monday • 15/09/2011: homogeneous light rain • DSR shape agree well with theory possible to disentangle the Mie and attenuation effects (Tridonet al. (2013), Geophys. Res. Lett., 40) • But some spectra issues prevent this method with the KASACR data (while its volume better match the measurements of the WSACR) and on more inhomogeneous cases
‚Soft‘ spheres Additional challengeforice: Habit dependenceofscattering • (Back-)scatteringdepends on particlesize/mass, habitandfrequency Ku (13.4 GHz) Ka (36.5 GHz) W (89 GHz) ‚Soft‘ ellipsoids
Multi-frequencysignaturesofsnowfallseemtobesize AND habitdependent Observedtriplefrequencysignatures in Wakasa Bay aircraftdatabyKulie et al., JAMC, 2013 Aggregates?? Theoreticaltriplefrequencysignaturesindicatehabit-relatedsignatures… Graupel??
Light snowfallexample: Ka – W band Spectra Frequencyindependent Rayleigh scatteringregion (plateau in spectral DWR) Miescatteringregion NSA, 12.Feb.2012, 08:52 UTC KaZR-spectrum: black WSaCR-spectrum: yellow
Examplesof multi-wavelengthspectrafrom different siteswithfocus on problematicdataqualityissues
WSACR narrow Nyquist velocity • Rain spectra can extend over more than 8 m/s width a Nyquist velocity of ± 4 is insufficient • Need wider Nyquist velocity in rain but narrower in ice (cloud and snow) temperature dependant modes?
KASACR spectra artefact kazr-kasacr comparison Range and time kasacr spectrograms 2600m 1500m Spurious bulges appear at the sides of kasacr spectra where there should be only noise (kazr) 800m 15 Sep 2011 19:32 to 19:58
WSACR / KaSACRshouldbe THE perfecttoolforsnowfallstudies. • WSACR andKaSACR: Veryaccurate beam matching, similar beam widths, averagetimes, rangeresolution, … • NSA WSACR andKaSACR: Spectrafromicecloudsshowvery different spectralwidth. Time samplingandaveragingshouldbeexactlythe same; also same rangeresolution NSA - KaSACR NSA - WSACR NSA, 11.07.2013
Currentissueswith SACRs: Spectralsidelobes NSA, 14.01.2013 NSA - KaZR NSA - WSACR NSA - KaZR NSA - WSACR • KaZRshowsinterestingsecondarypeak – nospectralsidelobes! • The multiple artifacts in the WSACR makesitimpossibletodistinguishartifactsfrommicrophysicalsignals! • It also affectsestimationofradarmoments (e.g. skewness, kurtosis, etc) • Similarartifactsfound in the SGP SACRS (but different strengthand not all time periods) and PVC WACR. • Can thisbesolved/avoidedsomehow?
Importance of matching in range For best results, the volume should be exactly matched in range i.e. with the same volume centres and sizes (pulse width and range weighting function when pulse coding) DSR: KASACR-WSACR DSR: KAZR-WSACR Closest gate Only 5 m shift!
Importance of matching in range Otherwise some interpolation can alleviate the mismatch but not completely in case of inhomogeneous volume KASACR-WSACR KAZR-WSACR Weighted interpolation between the two closest gates
Requirementsfor Doppler spectraanalysis (rain/snow/icecloudspecific) • First spectrumanalysisclearlyindicatesthat multi-frequencyspectracontainwealthofinformationaboutcloudandprecipitationmicrophysics. • However, requirementsforspectralanalysisareverydemanding: • „Perfectly“ verticallymatchedradarvolumes-> same pulse width, same rangegates, same rangeweightingfunction • „Perfect“ beam alignment • Possibilityof variable Nyquistrangefor rain andice/snowclouds • Similaror same velocityresolutionforspectra(at least forboth SACR) • Same temporal averaging/sampling • Radar calibration Sanity check: Iceclouds -> Multi-frequencyspectra MUST match (smallice = Rayleigh scatterers)
IOP idea: Check radarsettings in icepart (Rayleigh) SGP, 20.04.2013 SGP - KaSACR SGP - KaZR SGP - WSACR KaZR: 0.3 m/s KaSACR: 0.7 m/s WSACR: 0.7-0.8 m/s
Collecting Off-zenith SACR Spectra During Scanning Motivation • Enhanced perspective on cloud process evolution • Increased opportunity for data QC • Reduction of large time-gaps lacking spectra (esp. at W band) • Availability of controlled off-zenith data will drive innovation • Offers a controlled development test-bed for moving platforms Approach • Start small (e.g. with IOP) • Collect occasional full RHI scans • Can radar be modified to always save spectra for theta < e?
Observing Microphysics with Off-zenith Spectra A 256-bin Doppler spectrum experiences shape distortion of less than 1 bin induced by pointing angles up to 7 degrees off zenith. dynr_pri, Vpeak_pri v_leftpeak_pri, dynr_leftpeak_pri v_rightpeak_pri, dynr_rightpeak_pri ref_sec mdv_sec sw_sec skew_sec, kurt_sec, snr_sec, lslope_sec, rslope_sec ref_pri mdv_pri sw_pri skew_pri, kurt_pri, snr_pri, lslope_pri, rslope_pri dynr_sec, Vpeak_sec noise npeaks_pri vmin_sec vmax_sec vmin_pri vmax_pri
Observing Microphysics with Off-zenith Spectra Non motion-compensated spectrum skewness during MAGIC
ARM – IOP withfocus on Doppler spectrafromice, andsnowclouds (rain mightfollow) • First IOP at NSA sitewithfocus on icecloudsandsnowfall: • Find out whichsettingsareoptimum(time, rangeresolution, etc.) in ordertoinvestigatethevariousprocesses (nucleation, aggregation, riming, etc.)? • Howgoodcanwegetthespectramatching in the Rayleigh partofthespectrumfromKaZR, KaSACR, WSACR ifwerunthemsimultaneouselyandwithsimilarsettings (range, time res., zenithlooking)? • Whatarebenefits/disadvantagesof pulse compressionregardingspectra? Additional snowfallspecificexperimentswithin IOP: • Performslow RHI scanswithKaSACR/WSACR and XSAPR in thesame vertical planetoobtaintriplefrequencysignatures • Comparetriplefrequencysignaturestonovel in-situ data (3D snowflakecamera MASC) -> Are thetrip.-freq. signaturesreally so stronglyrelatedtosnowfallhabit?
Let‘sstartdiscussingnow…More about Doppler Spectra in „Fingerprinting“ session on Wednesday…
Currentissueswith NSA SACRs: Spectralsidelobes NSA - KaZR NSA - WSACR Numberandstrengthof „sidelobes“ increasewithmagnitudeof real signal; firstartifacts at ca. -15 dBZforKaSACRand ca. -11 dBZfor WSACR
Comparison WSACR – KaZR: Higher icepart NSA - KaZR NSA - WSACR NSA - WSACR NSA - WSACR NSA - KaZR NSA - KaZR Despitethe different rangeresolution (25 m vs 30 m), thespectra in theicepartarematchingextremelywell.
Comparison WSACR – KaZR: Lowersnowpart NSA - KaZR NSA - WSACR NSA - WSACR NSA - WSACR NSA - KaZR NSA - KaZR In thesnowfallpart (900 m), the WSACR seemstobeshiftedby 0.1-0.2 m/s towards larger fall velocities (orKaZRistooslow…?) – Note theshiftisindependentofthevelocityregime -> noMiescatteringeffect!
Light rain observed by ARM radars at SGP • Need well-matched beams to avoid artefacts • Light stratiform rain with higher Z fall-streak zoom to avoid BB and low SNR due to wind shear • Dual wavelength ratio • increase with height because of rain and gas attenuation • except right above the fall-streak possible with important Mie effect in the fall-streak due to larger drops • Check by looking at spectra + + +
Confirmation with KAZR and WSACR data KAZR WSACR KAZR-WSACR