Faraday rotation estimation from unfocussed ALOS PALSAR raw data

Size: px

Start display at page:

Download "Faraday rotation estimation from unfocussed ALOS PALSAR raw data"

Bertram Stafford
5 years ago
Views:

iranda 3 1 DISP, Tor Vergata University, Rome, Italy 2 IETR UR CNRS 6164,

1 Faraday rotation estimation from unfocussed ALOS PALSAR raw data arco Lavalle 1 3, E. Pottier 2, D. Solimini 1, N. iranda 3 1 DISP, Tor Vergata University, Rome, Italy 2 IETR UR CNRS 6164, University of Rennes 1, Rennes, France 3 European Space Agency, ESA ESRIN, Frascati, Italy PolInSAR Workshop Jan 2008

2 Outline Introduction FR prediction and estimation methods Unfocussed raw data analysis Results on ALOS PALSAR Conclusions 2

3 Introduction What is Faraday rotation Cause The satellite target propagation path crosses the ionosphere Ionosphere is an anisotropic medium due to the charged particles in a persistent Earth magnetic field The polarization plane of the radio wave rotates Effects Reciprocity does not hold Error in the estimation of calibration distortion Polarimetric techniques (e.g. decompositions) can be affected 3

4 Faraday rotation Prediction from TEC data Prediction of FR: Ω K B cos sec 2 f ( ψ ) ( θ )TEC 0 Ω: one way FR angle K: constant f: frequency TEC: total electron content 4

5 Faraday rotation Estimation from SLC data FR model (assuming polarimetric calibration of the system) HH VH HV VV = Ae jφ cosω sin Ω sin Ω S cosω S HH VH S S HV VV cosω sin Ω sin Ω cosω 5

6 Faraday rotation Estimation from SLC data FR model (assuming polarimetric calibration of the system) HH VH HV VV = Ae jφ cosω sin Ω sin Ω S cosω S HH VH S S HV VV cosω sin Ω sin Ω cosω 1. From HV VH difference Ω = 1 2 tan 1 HV HH + VH VV 2. From circular basis change (Bickel and Bates, 1965) LL RL LR RR 1 = j j 1 HH VH HV VV 1 j j 1 Ω = 1 4 arg * ( ) LR RL 6

7 Faraday rotation Proposed approach: estimation from unfocussed raw data (A) L1.0 RAW data A SAR processor B L1.1 SLC data SLC formation involves several steps (example ESA PALSAR verification processor) Orthogonalisation of the signals Range focussing Doppler centroid estimation Azimuth focussing (Stolt interpolation) Polarization channel coregistration Some operations might be nonlinear in the polarization components The Faraday rotation estimation might be corrupted (B) 7

8 Faraday rotation Proposed approach: estimation from unfocussed raw data (A) L1.0 RAW data A SAR processor B L1.1 SLC data SLC formation integrates pulses coming from different portions of the ionosphere SAR ionosphere RAW image SLC image target 8

9 Faraday rotation Proposed approach: estimation from unfocussed raw data (A) L1.0 RAW data A SAR processor B L1.1 SLC data The physical relationship between transmitted and received pulses subject to FR is valid on each polarimetric sample of raw data W W HH VH W W HV VV = cos Ω sin Ω sin Ω R cos Ω R HH VH R R HV VV cos Ω sin Ω sin Ω cos Ω 9

10 Faraday rotation Proposed approach: estimation from unfocussed raw data Same estimation methods of SLC data Estimation of FR angle using the relationship of circular basis change applied to unfocussed raw data W W LL RL W W LR RR 1 = j j W 1 W HH VH W W HV VV 1 j j 1 Ω = 1 4 * ( ) arg W W LR RL 10

11 Results on ALOS PALSAR Pauli RGB of SLC product (PALSAR L1.1) PALSAR acquisition South Italy DESCENDING pass Local time: 10:15 am April

12 Comparison RAW vs. SLC RAW data preprocessing L1.0 RAW data gain and offset compensation interference removal Faraday rotation estimation 12

13 Comparison RAW vs. SLC Faraday rotation maps L1.1 SLC data L1.0 RAW data gain and offset compensation interference removal Faraday rotation estimation COPARISON 13

14 Comparison RAW vs. SLC Faraday rotation maps L1.1 SLC data L1.0 RAW data gain and offset compensation interference removal Faraday rotation estimation COPARISON Pauli FR SLC FR RAW 14

15 Comparison RAW vs. SLC Histograms of estimated FR SLC data (PALSAR L1.1) SLC o Ω = 8.10 o σ = 0.93 RAW data (PALSAR L1.0) o Ω = 7.94 o σ = 0.85 RAW 15

16 Comparison RAW vs. SLC Range profiles 16

17 Comparison RAW vs. SLC Azimuth profiles 17

18 Comparison RAW vs. SLC Extensive analysis 30+ polarimetric PALSAR products Good agreement between RAW data estimation and SLC estimation 18

19 Comparison with TEC data 30+ polarimetric PALSAR products TEC data from ftp.aiub.ch TEC prediction is biased with respect to FR estimation from data products 19

20 Effect of the polarimetric calibration Comparison between un calibrated and calibrated products using the distortion matrices written in the product header (Shimada, 2007) Good agreement between FR estimates over calibrated and un calibrated SLC data 20

21 Conclusions FR estimated from RAW data vs. SLC data FR model and estimation methods are the same for RAW and SLC estimation ean value and variance of FR angles are very close No particular range/azimuth trend has been observed Impact of the distortion matrices is negligible Prediction from TEC data is slightly biased Investigation of the spatial variation of TEC from RAW data is in progress Practical implications FR estimated from RAW data can be used to improve the focusing in low frequency SAR ean value of FR angle can be annotated in the product header of RAW data and copied into the header of SLC data 21

22 Introduction Dual Polarimetry VS Full Polarimetry A dual pol mode has the same transmitting characteristics of a single pol mode and the same receiving characteristics of a quad pol mode. 22

ALOS-Indonesia POLinSAR Experiment (AIPEX): First Result*

ALOS-Indonesia POLinSAR Experiment (AIPEX): First Result* Mahmud Raimadoya(1), Ludmila Zakharova(2), Bambang Trisasongko(1), Nurwadjedi(3) (1) Bogor Agricultural University (IPB), P.O. Box 2049, Bogor