Atmospheric distortions of spaceborne SAR polarimetric signatures at X and Ka-band

Size: px

Start display at page:

Download "Atmospheric distortions of spaceborne SAR polarimetric signatures at X and Ka-band"

Muriel Butler
5 years ago
Views:

1 Atmospheric distortions of spaceborne SAR polarimetric signatures at X and Ka-band S. Mori 1,2, Frank. S. Marzano 1,2 and N. Pierdicca 1 1. Sapienza University of Rome, Italy 2. CETEMPS University of L Aquila, Italy

2 WHY RAINCLOUDS and SAR? POTENTIAL of X/Ku/Ka-band SAR for RAIN retrieval At higher frequency precipitating clouds may produce significant attenuation/scattering/depolarization effects The high spatial resolution of SAR sensors might provide new insights into the structure of precipitating clouds from space. Large availability of a new generation of X-SAR satellites near fully polarimetric Rain effects on SAR imaging and retrieval opportunity High-resolution microwave rain retrieval from SPACE at a scale below 1 km is only possible from SAR imaging data Rain effect can affect the interpretation of SAR data at X and higher frequencies, modifying the polarimetric signature of the ground Is there a minimum RainSAR retrieval sensitivity at a given spatial resolution (less than 1 km) nsidering the uncertainty of the surface background? Is there an upper limit to RainSAR retrieval? Are there geometrical effects to be solved?

3 X-SAR retrievals: Hurricane GUSTAV case South eastern Louisiana around 3.5 N x 89.5 W September 2, 28 at 12: UTC TerraSAR-X: HH pol data NEXRAD: S-band reflectivity PPI at.86 deg KMOB site

4 X-SAR retrievals: Hurricane GUSTAV case Precipitation rate [mm/h] Ground Weather Radar Spaceborne X-SAR Corr =.75 Bias = -.66 mm/h RMSE = mm/h FRMSE =.98

5 Outline Introduction Context and examples Parametric estimation of precipitation from X-SAR Case study Modeling of SAR observations due to precipitation SAR response model Polarimetric SAR observables High-resolution polarimetric simulated scenarios Polarimetric signature of precipitation Sensitivity analysis Conclusions

6 SAR cloud response and observables SAR polarimetric observables = + SAR SRF VOL SAR SAR SARhh SARvv S SAR S SARhh SARhh e j S 2 SAR * SARvv S SARvv SARpq: pq-polarized normalized radar cross section (NRCS) Z SAR : -polar ratio SAR : mplex rrelation efficient 2 Plane-wave incidence (avoid spherical wave front rrections) Radar swath

7 Altitude [km] Modelling SAR response: NRCS For a given pixel (x,y) the SAR NRCS can be formally expressed as follows: x, y x, y x y SARpq SRFpq VOLpq, SRFpq(x,y): surface backscatter, attenuated by the two-way path through the precipitating atmosphere VOLpq(x,y): volume backscattering due to hydrometeor reflectivit weighted by the two-way path through precipitating atmosphere SAR l Rain Frozen hydrometeors rt SRFpq VOLpq ground x, y x, yexp k l pq x, y t exp k l t x, y k pq pq pq 2 pq 2 Im F pq / S l x, y l x, y ground pq: surface target NRCS pq, k pq: : hydrometeors reflectivity and specific attenuation 4 S pq ( D, ) pp 2 pp dl dl l x, y l x, y N( D) p( ) ddsind k k qq qq l l dl dl dt Ground Range [Km] S pq, F pq : element of the mplex back or forward hydrometeor scattering matrix N(D): particle size distribution p(): particle orientation probability density function : wavelength

8 Modelling SAR response: Complex Correlation Coeff. For a ground point (x,y) the observable SAR mplex rrelation efficient is given by: C K SAR SAR VOL hh x, y arg SAR( x, y) vol t ( t) ( t) ( t e Re F vv K l( t) j hh vv arg( ) hh ( l) dl ground l vv e F vv hh )exp ( l) dl ( x ) hh SARhh kvvl dlexp khhl k l t e k ( l) dl l ( x ) vv SARvv e j2( x) l t sen dl e C VOL t( x) j2 ( t) ( t) dt SAR, ground, vol : mplex rrelation efficients of observed resolution cell, surface target and volume bin : backscatter differential phase S pq, F pq : elements of the hydrometeros mplex back or forward scattering matrix pq : hydrometeor reflectivity K : hydrometeor polar specific differential phase

parameterization teors polarimetric parameters can be modelled as function of water ntent by mean of power laws: Z k K epq pq x, z x, z a W x, z bkpq x, z akpqw x, z bk

9 Hydrometeor e.m. parameterization Hydrometeors polarimetric parameters can be modelled as function of water ntent by mean of power laws: Z k K epq pq x, z x, z a W x, z bkpq x, z akpqw x, z bk x, z akw x, z b ( x, z) aw x, z b ( x, z) a W x, z 5 K 4 2 pq Zpq bzpq W = water ntent [g/m 3 ] K = differential phase. [ /km] k pq = specific attenuation [db/km] Z epq = equivalent reflectivity [mm 6 /m 3 ] = mod. of the polar rr. Coefficient = arg. of the polar rr. efficient = wavelength [cm] K 2 =.93 for water and.19 for ice a Xpq, b Xpq efficients have been obtained by using APHESS T-Matrix radar scattering model, as described in [Marzano et al., 27] and [Marzano et al., 21].

10 Simulation case studies The proposed SAR response model requires as input a cloud structure (2D geometr hydrometeor types, hydrometeor water ntent distribution, hydrometeor e.m. response parameterization) and the polarimetric characterization of the ground target The ground target polarimetric variance matrix can be given by: Models (e.g., bare soil polarimetric response by [Oh et al., 22] Polarimetric signature of canonical targets (i.e., spheres with ground =1, dihedrals with ground =-1, others The cloud structure (i.e., hydrometeors distributions) can be derived by ad-hoc synthetic distributions, with simplified shape (e.g. parallelepiped), to assess main effects in a simple environment. Realistic fields simulated by a 3-D high-resolution mesoscale cloud-resolving models (CRMs)

Along Track [Km] Along Track [Km] Along Track [Km] Along Track [Km] Simulation of 3D realistic clouds SAR response The simulated hydrometeors distributions can be derived by simulated field through

11 Along Track [Km] Along Track [Km] Along Track [Km] Along Track [Km] Simulation of 3D realistic clouds SAR response The simulated hydrometeors distributions can be derived by simulated field through 3-D high-resolution mesoscale cloud-resolving models (CRMs) The System for Atmospheric Modeling (SAM) CRM [Blossey et al., 27] allows at simulating the distribution of Cloud, Rain, Ice, Snow, Graupel particles at 25 m resolution 2.5 DC section SARhh X band [db] X NRCS [db] Examples of horizontal and vertical distribution of densities of different hydrometeors predicted by model at a given epoch Ground plane SARhh for X, Ku and Ka band, hh polarization plus the total vertically-integrated lumnar ntent (VIWC). The ground response has been nsidered as nstant (about -7 db at X-band), and so the incident angle (4 ). The images of the storm are simulated in ground range (the x-y plane), 5 km width, spread all the cross range dimension (64 km) SARhh Ku band [db] SARhh Ka band [db] Ka Ku VIWC [kg/m 2 ] Ground Range [Km]

12 Example: Rectangular cumulonimbus at X-Band Parallelepiped structure Horizontal uniform 1 km wide hydrometeor density 2 vertical layers: raindrops.3 g/m 3 and snoflakes.5 g/m 3 water ntent, with 15 km overall height E. M. parameters from revised APHESS T-Matrix model [e.g., Marzano et al., 27] Homogenous background from bare soil model [Oh et al., 22]: Rms Height (k s ) 1.5 cm, Correlation Lenght (k l ) 5. cm, Volumetric soil moisture ntent (m v ).25 [cm 3 /cm 3 ] SARhh SAR Zero km ground -6 db f = 9.6 GHz =4 deg SRFhh VOLhh SAR

13 Example: Rectangular cumulonimbus at Ka-Band Parallelepiped structure Horizontal uniform 1 km wide hydrometeor density 2 vertical layers: raindrops.3 g/m 3 and snoflakes.5 g/m 3 water ntent, with 15 km overall height E. M. parameters from revised APHESS T-Matrix model [e.g., Marzano et al., 27] Homogenous background from bare soil model [Oh et al., 22]: Rms Height (k s ) 1.5 cm, Correlation Lenght (k l ) 5. cm, Volumetric soil moisture ntent (m v ).25 [cm 3 /cm 3 ] SARhh SAR Zero km ground -6 db f = 35 GHz SRFhh VOLhh SAR =4 deg

14 Outline Introduction Context and examples Parametric estimation of precipitation from X-SAR Case study Modeling of SAR observations due to precipitation SAR response model Polarimetric SAR observables High-resolution polarimetric simulated scenarios Polarimetric signature of precipitation Sensitivity analysis Conclusions

15 Ka-Band X-Band Case 1(A): Altostratus (liquid particles sensitivity) Non-precipitating cloud Ice crystals.5 g/m 3 2km height Cloud Liquid {.1,.1,.1, 1.} g/m 3 2km height Ground: dihedral rner reflectors.25 m 5 Detectable only at Ka-band, also.1 g/m 3 particles No phase rotation except Ka-band 1 g/m 3 Polarimetric effects at X- Band, also at.1 g/m 3 SARhh SAR SAR

16 Ka-Band X-Band Case 1(B): Altostratus (frozen particles sensitivity) Non-precipitating cloud Ice crystals {.1,.1,.1, 1.} g/m 3 2km height Cloud Liquid.2 g/m 3 2km height Ground: dihedral rner reflectors.25 m 5 X-Band sensitive only to high densities of ice Ka-band sensitive to ice, especially for W >.1 g/m 3 Ka loss of polarimetric herence for IC CL Ka phase rotations for W >=.1 g/m 3 SARhh SAR SAR

17 Ka-Band X-Band Case 2(A): Nimbostratus (liquid particles sensitivity) Precipitating cloud Ice crystals.5 g/m 3 2km height Light Raindrops {.1,.1,.1, 1.} g/m 3 4km height Ground: dihedral rner reflectors.25 m 5 Precipitation signature at Kaband, for every densitiy At X-Band only 1 g/m 3 causes volumetric effects (actually not-detectable NRCS) Loss of polarimetric herence for every band and density Phase rotation and increased herence for highest densities SARhh SAR SAR

18 Ka-Band X-Band Case 2(B): Nimbostratus (frozen particles sensitivity) Precipitating cloud Ice crystals {.1,.1,.1, 1.} g/m 3 2km height Light Raindrops.2 g/m 3 4km height Ground: dihedral rner reflectors.25 m 5 Loss of polarimetric herence at every band and density (at Ka only for volume mponent) Phase rotation for highest density only (@Ka >.1 g/m 3 ) Precipitation signature at Ka for every densit at X for W >=.1 g/m 3 SARhh SAR SAR

19 Conclusions A 2D/3D simulator of the polarimetric response of a SAR in presence of precipitating clouds has been developed Considering simple synthetic clouds as well as realistic clouds predicted by a mesoscale model we found that: The impact of clouds on X and Ka band SAR can be significant, even for clouds with relatively high occurrence The polarimetric signature of the ground target can be significantly modified or even mpletely masked X Band NRCS shows an appreciable sensitivity to only intense events Ka band nfirms a great sensitivity to frozen particle, but also X band is affected Complex rrelation efficient exhibits a good rrelation with lumnar ntents. SAR remote sensing of clouds has spatial resolution useful for water budget, water management and hydrological model initialization Acknowledgments This work has been partially funded by: European Union project Earth2Observe Global Earth Observation for Integrated Water Resource Assessment ( Weinman and Marzano, JAMC, 28 Marzano and Weinman, TGRS, 28 Marzano et al., TGRS, 21 Marzano et al., HESS, 211 Marzano et al., TGRS, 212 Pulvirenti et al., TGRS 214 Mori et al., PolInSAR 215 Mori et al., EOWCS 215 Mori et al., ESA LPS 216 Mori et al., SPIE RS 216

RainGain Project Meeting: High Resolution Rainfall Rate Estimation and Forecast

RainGain Project Meeting: High Resolution Rainfall Rate Estimation and Forecast RainGain Project eeting: High Resolution Rainfall Rate Estimation and Forecast Ricardo Reinoso Rondinel and Herman Russchenberg Faculty of Electrical Engineering, athematics and Computer cience, Remote