Tampa Bay 2010
Transcription
Tampa Bay 2010
Polarimetric radar measurements of precipitation to interpret, retrieve, and validate X-band Synthetic Aperture Radar observations Misure radar polarimetriche della precipitazione per interpretare, ricostruire e validare le osservazioni da radar ad apertura sintetica in banda X Luca Baldini, Eugenio Gorgucci, Nicoletta Roberto CNR - Istituto di Scienze dell’Atmosfera e del Clima, Roma Dino Giuli, Luca Facheris, Fabrizio Cuccoli (CNIT) Università di Firenze, Lab. Radar e Radiocomunicazioni Roberto Deidda, M.G. Badas, G. Mascaro Gruppo di Idrologia – Dip. Ingegneria del Territorio, Università di Cagliari Chandrasekar V. Chandra, Jason Fritz Colorado State University, Radar & Communications Group, Ft Collins, USA Achieved results in response to the first COSMO‐SkyMed Announcement Of Opportunity Rome, March 27th, 28th and 29th 1 Project rationale and objective Several studies have analyzed and modeled the effect of precipitation on SAR at different frequencies (Ferrazzoli and Schiavon (1987) Danklmeyer et al (2009), Atlas and Moore, (1987), Marzano, Weinmann (2008, 2010). Interest has revived with SAR missions at X-band. At X-band precipitation attenuation is more pronounced than at longer wavelengths and can affect many applications of SAR measurements. The mechanism determining SAR NRCS in precipitation is not straightforward, combining scattering/attenuation from the volume distribution of hydrometeors and surface scattering which, in turn, is modified by precipitation. Interpretation of SAR data in precipitation can take advantage from ground instrumentations (Melsheimer at al. 2001), such as (Polarimetric) Doppler Weather Radar providing microphysical properties of precipitation along the SAR beam measurements to predict the attenuation/scattering properties along the SAR beam Effects of precipitation on SAR images Influence on surface backscattering Modification of surface NRCS Backscattering due to precipitation Attenuation due to precipitation Effects on polarimetric measurements due anisotropy of hydrometeors difference between HH-VV intensity of returns phase difference HH-VV Differential attenuation Azimuth resolution Precipitation has an inherent Doppler spectrum that degrades azimuth resolution (Atlas and Moore 1987). ra = 2σ v R / U ≈ 250 × σ v U: sat speed; R: slant range σV = precip Doppler spectrum width Shift along azimuth direction (Moore et. al 1997) Δr = ur R U For CSK and a ur of 20 m/s, shift is more than 4 km. Effects of precipitation on SAR images to Surface backscattering z R A S ΔR Near beam edge υf υj υn Q σ 0 (Q ) = σ 0s Far beam edge x Effects of precipitation on SAR images to Surface backscattering z R A S Precipitation backscattering Precipitation path attenuation ΔR Near beam edge r(x) σ att surf υn υf υn υj x x’ Q p( x ) σ0 (Q ) = σ0s ⋅ l [r ( x )] + sin υ ∫ η p ( x' ) ⋅ l [r ( x' )] dx' 0 Far beam edge att σ vol Effects of precipitation on SAR images to Surface backscattering z R A S Precipitation backscattering Precipitation path attenuation ΔR Near beam edge υn υf υn υj p( x ) σ0 (Q ) = σ0s ⋅ l [r ( x )] + sin υ ∫ η p ( x' ) ⋅ l [r ( x' )] dx' 0 Far beam edge x Effects of precipitation on SAR images to Surface backscattering z R A S Precipitation backscattering Precipitation path attenuation ΔR Near beam edge (1) r(x) σ eff (Q )HH σdr (Q ) = = σeff (Q )VV υn (σ ) υf υn x p( x ) 0 surf HH HH (σ ) υj Far beam edge l [r (x )] + sin υ ∫ ηHH (x' )lHH [r (x' )]dx' 0 p( x ) 0 surf VV VV l [r (x )] + sin υ ∫ ηVV (x' )lVV [r (x' )]dx' 0 Weather radar data to estimate SAR observation to Surface backscattering z R A S Precipitation backscattering Precipitation path attenuation ΔR Zh Z dr Near beam edge K dp υf υj υn αh p( x ) αh σ0 (Q ) = σ0s ⋅ l [r ( x )] + sin υ ∫ η p ( x' ) ⋅ l [r ( x' )] dx' 0 Far beam edge x Weather radar data to estimate SAR observation Geometric mapping 1. Define a grid on a cross track plane 2. Map the grid onto GR coordinate y ζ (j⋅-i)ΔR Oa Hmax Ψ0 ΟΡ Near beam edge i⋅ΔR υ ν Oa υ X Surface backscattering Precipitation backscattering Precipitation attenuation paths Q( y,s) x P Ψ Y’ Pi,j Z y′ CSK ground track Far beam edge υφ ξ ϕ ΔR/tanυn+ …+ ΔR/tanυϕ OS s θ X0 X’ X η x′ Weather radar data to estimate SAR observation Reflectivity and Specific attenuation estimation SAR image SAR metadata Z (CSKfreq, θ)=f1(GRmeas θ= 0°) Ground Radar (S, C band) 2 η = Zπ5 K p λ− 4 × 10 −18 Zh ⎛ Zdr ⎞ ⎜ ⎟ ⎜K ⎟ ⎝ dp ⎠ α (CSKfreq, θ)=f2(GRmeas θ= 0°) l (r ) = 10 Estimation Microphysical model −0.2 ∫ α ( s )ds r algorithms Non linear regression T-matrix simulation ηHH ,VV l HH ,VV Weather radar data to estimate SAR observation 0° elevation 42° elevation S-band Single polarization radar Gamma DSD variability 3 ≤ log10 N w ≤ 5 0.5 < D0 < 3.5 −1 < μ ≤ 5 Zh<55 dBZ; R<300 mm h-1 Canting angle:mean = 0°, width= 7.5° Temperature = 20° Shape-size model: Beard and Chuang 1987 Baldini et al. IGARSS 2012 Fritz and Chandrasekar, JTECH submitted 0° elevation 42° elevation Weather radar data to estimate SAR observation Datasets of simultaneous NEXRAD WRS 88D observation of precipitation CSK in its different modes Weather radars at ground Bric della Croce (TO) Owner: Regione Piemonte Site: Lat. 45°1’53’’, Lon. 7°43’59’’ Altitude: 720 m Manufacturer: ALENIA Doppler: YES - Band: C Dual polarization: NO Beamwidth: 1° - Radome: YES • Operational GR • ARPA Piemonte dual pol C-band radar • NEXRAD S-band single pol • Research GR • dual pol C-band ISAC-CNR Polar 55C research radar, Rome RU ISAC-CNR Polar 55C, Roma Weather radar data to estimate SAR observation Typical operational weather radar volume scanning T0 T0+ TR Weather radar data to estimate SAR observation Sampling effect φ=10 km Ideal rain cell: a cylinder with constant reflectivity. Z=40dBZ 2 cases: 1) The cell is stationary (vc=0) H=6 km 2) The cell is moving (vc=6m/s) Weather radar data to estimate SAR observation Sampling effect Optimized scanning SAR‐WR coincident sampling NEXRAD Volume Coverage Pattern VCP12 6 m/s Baldini et al. AMS Int Conf. on Radar Meteorology, 2011 sector including SAR footprint, Vertical resolution better than 1km within SAR footprint Case studies Case CSK Mode Weather radar Collection Tampa Bay 2010 Intense precipitation over land Ping Pong S-band single pol NEXRAD, operational Request Piemonte 2010 Observation over non constant background Ping Pong C-band dual pol ARPA Piemonte, operational Request IVREA 2011 Intense precipitation over mountains Stripmap C-band dual pol ARPA Piemonte, operational Archive Lousiana 2010 Intense precipitation over sea Scansar Huge S-band single pol NEXRAD, operational Archive Scansar Wide C-band dual pol ISAC-CNR, research Archive Roma 2012 Snow and mixed phase precipitation 17 Tampa Bay 2010 NEXRAD Tampa Bay (KTBW) – Melbourne (KMLB) S – band Doppler radar CSK Ping Pong HH-VV CSKS2 2010-06-18 CSKS3 2010-06-19 CSKS1 2010-06-26 CSKS2 2010-07-04 CSKS1 2010-07-12 CSKS2 2010-07-20 Tampa Bay 2010 18 CSK3 2010-06-19 KTBW VMI: CSKS1 2010-06-26 CSKS2 2010-07-04 CSKS2 2010-07-20 Blue:40dBZ Red:45dBZ Green:50dBZ 19 Tampa Bay 2010 Reconstruction of a SAR image from KTBW measurements: cross track CSK Reflectivity CSK Specific attenuation l [r ( x )] σ0 (no precip) σ0 (no precip) p( x ) Reconstructed σ0 profile sin υ ∫ η (x' )⋅ l[r (x' )] dx' p 0 Tampa Bay 2010 CSKS2 2010-07-04 CSKS1 2010-06-26 l HH (dB ) ηatt HH (dB ) l D (dB ) Tampa Bay 2010 CSKS1 2010-06-26 σ 0HH CSK σ 0HH CSK no rain + KTWB CSKS2 2010-07-04 Tampa Bay 2010 Scalloping effect σ 0HH (precip) (CSKS1 2010-06-26) σ 0D (precip) 2.1 km Δσ 0HH Δσ 0D precip − no precip precip − no precip 23 Tampa Bay 2010 -5 -10 -15 Δσ0hh (dB) 0 5 04 July 2010 (precip) and 20 July 2010 (coregistered, no precip) 25 30 35 40 45 Cumulative slant range S-band Z 50 [10 55 60 log10(mm6m-3 65 km)] Tampa Bay 2010 04 July 2010 (precip) Reconstruction from weather radar data [10 log10(mm6m-3 km)] Baldini et al. Atmospheric Research, 2012 submitted SAR measured Piemonte 2010 PING PONG HH-VV CSK images CSKS3 2010-05-13 ARPA Piemonte Bric della Croce C- band dual polarization radar CSKS1 2010-05-20 CSKS2 2010-05-28 CSKS1 2010-06-21 CSKS2 2010-06-29 CSKS1 2010-07-07 25 Piemonte 2010 red = |βHH|0.6 green = |βVV|0.6, blue = (2-1/2)|βHH - βVV|0.6 CSKS3 2010 May 13 05:09:06 UTC CSKS1 2010 May 20 05:09:01 UTC CSKS1 2010 Jun 21 05:08:37 UTC CSKS2 2010 Jun 29 05:08:32 UTC 26 Piemonte 2010 Dual polarization C-band weather radar data to estimate X-band SAR observation 0° elevation 65° elevation C-band dual polarization radar Gamma DSD variability 3 ≤ log10 N w ≤ 5 0.5 < D0 < 3.5 −1 < μ ≤ 5 Zh<55 dBZ; R<300 mm h-1 Canting angle:mean = 0°, width= 7.5° Temperature = 20° Shape-size model: Beard and Chuang 1987 0° elevation 65° elevation Piemonte 2010 Contours of Bric della Croce reflectivity (blue=29 dBZ; red=36 dBZ; green=43 dBZ) superimposed on CSK2 29 June image. Baldini et al. ERAD 2010 Reflectivity and specific attenuation reconstructed from Bric della Croce radar 28 Ivrea 2011 ARPA Piemonte Bric della Croce C- band dual polarization radar STRIPMAP HH CSK images CSK2 27 06 2011 (no rain 1) CSK2 13 07 2011 (rain) CSK2 29 07 2011 (no rain 2) 29 Ivrea 2011 Latitude Bric della Croce VMI: Blue:45dBZ Red:50dBZ Green:55dBZ Longitude Ivrea 2011 (NoRain1 –NoRain2) (Rain –NoRain1) (Rain –NoRain2) σ0(hh) (radar no class) σ0(hh) (precip) σ0(hh) (no precip) Ivrea 2011 σ0(hh) (simu class) σ0(hh) (precip) σ0(hh) (no precip) Ivrea 2011 Lousiana 2010 COSMO SkyMed monitoring of Deepwater Horizon oil spill NEXRAD KLIX (New Orleans) S – band Doppler radar scanning mode: 20 PPI in 5 min (20 elevation: 0.5-20deg; azimuth resolution 0.5 or 1deg; range resolution 250m, max. range = 250 km). CSK Huge Region image July 6th , 2010, 12:02 UTC Ascending, right Lousiana 2010 CSK2 7 July 2010 12:03 Louisiana (US) Precip Backscattering Attenuation Roberto et al. EGU 2012 Oil Spill KLIX reflectivity superimposed contour: Blue=35 dBZ, Red=40 dBZ Green=45 dBZ Lousiana 2010 backscattering and attenuation of radiation by hydrometeors in the rain cells; Backscattering of sea induced by the impact of raindrops and wind. to Surface backscattering z R A S Precipitation backscattering Precipitation path attenuation ΔR Near beam edge r(x) υn υf υn υj x x’ p( x) σ 0 (Q ) = σ s0 ⋅ l [r ( x )] + sin υ ∫ η p (x') ⋅ l [r (x')] dx' 0 Far beam edge Lousiana 2010 Sea surface NRCS behavior 1. Rain intensity variation Relationship h2(rms) and rainfall by experimental results1 Roughness of water surface is modeled through a Gaussian height distribution, with variance h2(rms). Bliven L F Sobieski PW,Elfouhaily T, 1995, “Ring Wave Frequency Spectra: Measurementes and Model”, Proc IGARSS ’95, p.830. Lousiana 2010 Sea surface NRCS behavior 2. Rain intensity variation ~15 dB Wind speed 4.3 m/s CSK incidence angles Bahar e and Fitzwater M A, 1984 “Scattering cross sections for composite rough surfaces using the unified full wave approach” IEEE Trans Ant. Propag.32, N7, pp730-734 Capolino F. Facheris L. Giuli D. Sottili F. , 1997, “ The determination of the sea surface NRCS when corrugated by blowing wind and raingall: an application to rainfall rate measurements over sea” Int. Conf. On Antennas and Propagation, 1997, N 436. Lousiana 2010 Radar rain rate PIA attenuation ηvv σ0 backscattering σ0 Vw=4.3 m/s Lousiana 2010 PIA ηvv Radar rain rate σ0 Vw=4.3 m/s Roma 2012 CSKS2 2012-02-02 CSK Scansar Wide Orbit: descending right Incident angles: 23.75- 36.78 ° CSKS3 2012-02-03 Polar 55C VMI: Blue:15dBZ, Red:20dBZ, Green:25dBZ Roma 2012 CSKS2 2012-02-02 CSKS3 2012-02-03 120 km Roma 2012 CSKS2 2012-02-02 120 km Roma 2012 …… Future acquisition of simultaneous observations using CSK stripmap and Polar 55C with scanning composed of: •PPI sectors •RHI on a SAR cross-track plane) Observation area 44 45 Conclusions COSMO SkyMed images collected using different modes in the presence of precipitation events have been analyzed using simultaneous observations of the same precipitation event collected by ground based radars Use of ground based weather radar to identify and quantify precipitation effects have been developed. Models were developed for S-band operational single polarization radar C-band operational dual polarization radar Better results are obtained with dual polarization radar that allow to achieve a classification of hydrometeors Concerning the Ping Pong HH mode, the relationship between attenuation and linear attenuation in CSK polarimetric images fairly agrees with that obtained from a simple propagation model. However, especially in cases of low S/N, the scalloping effect can alter significantly the ratio of copolar returns between adjacent strips. Better results are expected with the use of the quad pol scheme. (COSMO II) Interesting results have been obtained for precipitation over the sea surface where NRCS changes as result of wind and impinging raindrops Before August 2012, collection of CSK stripmaps and ISAC-CNR Polar 55C (using specifically designed scanning strategies) observations of precipitation will be pursued.