Snow Microphysical Retrieval Based on Ground Radar Measurements V. Chandrasekar Colorado State University June 27, 2007 1
Outline Role of inter comparing ground and space borne radar Class of measurements for Microphysical retrieval Spectral decomposition of polarimetric measurements Observations Summary June 27, 2007 2
Satellite instrument Ground Radar (GR) June 27, 2007 3
Fundamental class of measurements Covariance Measurements Polarization/Reflectivity -> Impact on microphysics Doppler spectrum -> Fall velocity -> Microphysics Combined Spectral decomposition of dual-polarization Elevation angle dependence June 27, 2007 4
Standard techniques v Converts Doppler power spectra to microphysical retrievals by using a velocity-size relation Utilizes covariance matrix observations to obtain microphysical retrievals June 27, 2007 5
Combine the two advantages Elevation angle 30 to 60 degrees Measurements taken at high elevation angle, 30-60 is optimum angle range This measurement setup allows for combining Doppler measurements with dualpolarization observations, and is useful for snow retrievals. June 27, 2007 6
Profile of polarimetric radar data at 10 GHz in snow from the Arctic region From Hudak et al ( 1999) June 27, 2007 7
Illustrating the approximate fall modes of plate- and column-like crystals. June 27, 2007 8
Radar measurements of slant LDR in a region of plate-like crystals Radar measurements of slant LDR in a region of column-like crystals Reinking et al ( 1999) Ka band observations June 27, 2007 9
Elliptic Depolarization ratio RHI scans from dendrites From Reinking et al. (1997) June 27, 2007 10
Winter storm 2006 Measurements taken on Dec 20, 2006. June 27, 2007 11
Dual-polarization spectral measurements in ice Dual-polarization spectrographs of the snow event ( with mean velocity shifted to zero) Spectral differential reflectivity allows for discrimination between particle types June 27, 2007 12
Comparison between classification schemes Graupel Crystals Aggr. Comparison between classification scheme results. The right figure shows output of (Lim et al 2005) fuzzy logic classifier. June 27, 2007 13
Ice particles distributions Spectral differential reflectivity in combination with the power spectral density provides a useful tool for discrimination between different ice particle types Can we use this information to estimate microphysical parameters? June 27, 2007 14
Interplay between Doppler and Polarimetric observations ( rain example ) June 27, 2007 15
Observation model Observed spectrum can be represented as Z σ 2 h( v) = Spr( v) * Sb ( v v0 ) + N Noise S pr Precipitation spectrum ( v) = σ( D) N( D) dd dv (m 3 / ms -1 ) S b Broadening kernel 2 1 v ( v) = exp( ) 2 2πσ 2σ b Ambient wind velocity To retrieve a PSD, N(D), one needs to estimate ambient mean wind velocity, v 0, and spectrum broadening kernel width, σ b b June 27, 2007 16
Spectral differential reflectivity Z dr ( v) = σ σ hh vv ( D) N( D) ( D) N( D) dd dv dd dv * * S S b b ( v ( v v v 0 0 ) ) In rain the spectral differential reflectivity can be used to retrieve the raindrop size shape relation as well as nonparametric DSD. June 27, 2007 17
Ice particle model Density for different particle types using Fabry and Szymer(1999) June 27, 2007 18
Ice observation model; Axis ratios and terminal velocity for ice particles Khvarostyanov et al (2002 ) June 27, 2007 19
Possible solution Methodologies The optimization/ model matching is divided into four stages max min D agg 0, D cry 0 v v= v min [ ] 2 obs mod agg cry Z ( v) Z ( v, D, D ) dr dr 0 0 max min σ b v v= v min Z obs dr ( v) Z mod dr ( v, σ, D b agg 0, D cry 0 ) D agg 0, D cry 0 2 Estimation of N w agg and N w cry based on Z dr (v) and Z h (v) Estimation of v 0 based on Z h (v) Contribution of Slant LDR / Multiple frequency spectral decomposition of polarization constraints this solution space. June 27, 2007 20
Resulting fit June 27, 2007 21
Commonly used Ice precipitation composition Ice particle type Z h (dbz) Z dr (db) ρ hv (0) K dp LDR Aggregates <35 0-1 >0.95 0-0.2 <-25 Crystals <35 0-6 >0.95 0-0.6 <-27 V. Crystals <35-0.5 to 0 >0.95-0.6 to 0 <-24 Graupel 20-50 -0.5 to 2 >0.95 0-0.5 <-20 It is difficult to discriminate between different types of ice particles. One of the reasons is that in a radar volume more than one type of particles can coexist. In addition the mapping space is not one-toone. June 27, 2007 22
Spectral Observations of super cooled water Doppler velocity power spectra at various heights over Barrow Alaska 6 October 1004. Each panel represents spectra at the stated height (indicated in the left upper corner) for the ~3 min period 1612 UTC to 1615 UTC. See text for discussion on the interpretation of the spectrographs From M. Rambukkange, J. Verlinde, E. Eloranta, E. Luke, P. Kollias and M. Shupe, 2006: Fine-scale Horizontal Structure of Arctic Mixed-Phase Clouds. AMS Cloud Physics Conference, Madison, WI June 27, 2007 23
Snow measurements on Jan 22, 2007 by King City Radar RHI plot is obtained from the recorded time-series data June 27, 2007 24
Doppler spectrum at vertical incidence June 27, 2007 25
Slant profile dual polarization spectral observations Mean velocities scaled to match ones at vertical incidence June 27, 2007 26
Comparison of in situ and spectral observations From (Petersen et al 2007) June 27, 2007 27
Architecture of hydrometeor classification scheme June 27, 2007 28
Result of classification and in situ data June 27, 2007 29
Summary Spectral decomposition of dual polarization observations show new opportunities for large scale snow remote sensing from ground with potential for microphysical retrievals Preliminary results show potential for ground based radars to get quantitative inferences The inference potential is enhanced with multiple frequencies. June 27, 2007 30
Different density diameter models for Ice ( Fabry and Szymer, 1999) June 27, 2007 31