Introduction to SAR Imaging Presenter: Chris Stewart, RSAC c/o ESA Authors: Michael Eineder, Richard Bamler Remote Sensing Technology (TUM/DLR)
Educational Objective Understand the SAR imaging process and SAR instrument operation modes Understand the relevant parameters of SAR systems Understand the basic SAR scattering process Understand the properties of SAR images Know some SAR satellites and their main properties
Further Reading Curlander, J. C., & McDonough, R. N. (1991). Synthetic Aperture Radar: Systems and Signal Processing. Wiley Series in Remote Sensing. New York: John Wiley & Sons, Inc. Cumming, I. G., & Wong, F. H. (2005). Digital Processing of Synthetic Aperture Radar Data: Algorithms and Implementation. Boston, London: Artech House.
Structure Introduction Radar imaging (range-component) Formation of synthetic aperture (azimuth-component) Advanced SAR imaging modes Characteristics of SAR images
Structure Introduction Radar imaging (range-component) Formation of synthetic aperture (azimuth-component) Advanced SAR imaging modes Characteristics of SAR images
Airborne (DLR E-SAR) X-band SAR image 1 m resolution The image shows some scattering phenomena: B/W, smooth runways, bright buildings (steel, corners), fields and forest with different roughness and dielectric constants © DLR
Synthetic Aperture Radar - SAR Active Þ independent of sun illumination microwave Þ penetrates clouds and (partially) soil, snow wavelengths: X-band: 3 cm C-band: 6 cm L-band: 24 cm coherent Þ interferometry, speckle polarization can be exploited spatial resolution: space-borne: 1 m - 100 m air-borne: > 0.2 m Synthetic Aperture Radar (SAR)
SAR Imaging Geometry radar Radar transmits pulses and receives echoes at the rate of the pulse repetition frequency: PRF @ 1000 - 4000 Hz range: radar principle = scanning at speed of light (slant) range azimuth: scanning in flight direction at plus aperture synthesis (holography) coherent imaging: complex-valued pixels contain amplitude (brightness) and phase information azimuth swath width for this lecture: straight flight path Fig.: © DLR ground range
SAR – A Two-Step Imaging Process SAR is a two-step imaging process: 1. Data acquisition Illumination of a scattering object and collection of received echoes Þ raw data Contribution of a single point is dispersed over 104 … 107 samples 2. Processing Raw data focusing Þ image of the object
SAR Image Examples Sensor: ERS-1 Mojave Desert CA, USA azimuth range Sensor: ERS-1 Mojave Desert CA, USA Size » 40 km x 40 km ERS-1 © ESA
SAR Raw Data (After Range Compression) azimuth ERS-1 © ESA
Focussed SAR Data azimuth ERS-1 © ESA range
Focussed SAR Data after azimuth pixel averaging by 4 ERS-1 © ESA after azimuth pixel averaging by 4 to achieve approximately square pixels
Structure Introduction Radar imaging (range-component) Formation of synthetic aperture (azimuth-component) Advanced SAR imaging modes Characteristics of SAR images
Advanced SAR Modes: e.g. TerraSAR-X ScanSAR (100 km swath, 15 m res.) Stripmap (30 km swath, 3 m res.) Azimut Range Spotlight (5 km swath, 1 m res.) Point target response Fig.: © DLR
ScanSAR (100 km swath, 15 m res.) ScanSAR Mode ScanSAR (100 km swath, 15 m res.) Periodic switching of antenna elevation look direction: Illuminate/receive only part (e.g. ¼) of synthetic aperture with bursts Use remaining time to look („Scan“) at other ranges by steering the antenna electronically + increased swath width (e.g. × 4) - reduced resolution (e.g. ÷ 4) Fig.: © DLR
Spotlight (5 km swath, 1 m res.) Spotlight Mode Spotlight (5 km swath, 1 m res.) Increase aperture time by steering the antenna electronically backward in azimuth longer illumination time, longer chirp Higher chirp frequencies & aliasing with PRF need special processing techniques + increased resolution (e.g. × 3) - continuous operation not possible, limited azimuth image size (e.g. 5 km) Fig.: © DLR
TOPS Mode (e.g. Sentinel-1) TOPS (5 km swath, 1 m res.) Reduce aperture time by steering the antenna electronically forward in azimuth More azimuth distance, less illumination time per target Saved time can be used to electronically steer the antenna to other elevation directions + increased swath width (e.g. S1: 3x = 250 km) - reduced resolution (e.g. S1: 17 m) Fig.: © DLR
Structure Introduction Radar imaging (range-component) Formation of synthetic aperture (azimuth-component) Advanced SAR imaging modes Characteristics of SAR images
SAR Image Characteristics azimuth range Issues: Resolution (shape of point scatterer response) Radiometry Geometry Moving objects Polarimetry Artifacts ERS-1 © ESA
Speckle “Noise” Fig.: © DLR ERS © ESA Random positive and negative interference of wave contributions from the many individual scatterers within one resolution cell Varying brightness from pixel to pixel even for constant σ0 Granular appearance even of homogenous surfaces
Example for Bayesian Speckle Reduction © AeroSensing GmbH © AeroSensing GmbH original SAR image SAR data © AeroSensing GmbH speckle filtered Bayesian algorithm
Speckle Reduction by Temporal Multilooking (ERS) +10dB Data: ERS ©ESA Data: ERS ©ESA/DLR -10dB 5 spatial looks 20 x 20 m ground resolution 2 dB radiometric resolution 320 spatio-temporal looks 20 x 20 m ground resolution 0.3 dB radiometric resolution Further Examples: Module 2106 - Speckle Filtering
Parameters Influencing Radar Brightness Sensor Parameters wavelength (e.g. penetration through canopy) polarization look angle resolution (texture) Scene Parameters surface roughness (e.g. Bragg scattering at ocean surfaces) local slope and orientation Ü geomorphology scatterer density, e.g. biomass, leaf density 3-D distribution of scatterers and scattering mechanism, e.g. surface, volume, or double bounce (canopy, trunks, buildings) dielectric constant e Ü scattering material soil moisture vegetation status
Calibration of SAR Systems Instrument parameters to be calibrated: transmit power receiver gain elevation antenna pattern (satellite roll angle !) Calibration objects: corner reflectors active radar calibrators (ARCs) rain forest
Corner Reflectors for SAR End-to-End Calibration Radar cross section of a trihedral corner reflector: L © DLR L © DLR
strip 12 strip 03 strip 11 Radiometric Product Accuracy Rainforest Stripmap VV =20°-41° strip 12 Radar brightness: b0 = k * pixel-amplitude2 (k=calibration constant) strip 03 Sigma nought from local incidence angle x,y σ0 = b0 * sin(x,y ) strip 11 Application specific calibration assuming isotropic volume scattering g0 = b0 * tan( x,y ) © DLR
Heavy Clouds and Rain Cells in X-Band SAR Images Only visible at short wavelengths and extreme conditions
Penetration of Microwaves X-Band λ=3 cm C-Band λ=6 cm L-Band λ=23 cm X C L vegetation dry soil glacier ice Fig.: © DLR
Scattering Mechanisms in Forests 2 4 1 3 5 1: direct single scatter 2: multiple bounce 3: direct ground reflection 4: double bounce trunk - ground 5: attentuation of ground scatter by canopy Fig.: © DLR
Geometry of SAR Images - Foreshortening range slant range foreshortening ground range Slopes oriented to the SAR appear compressed ERS-1 © ESA Fig.: © DLR
Geometry of SAR Images - Lay-over range lay-over Steep slopes oriented to the SAR lead to ghost images ERS-1 © ESA Fig.: © DLR
Lay-Over Mask Computed from DEM © DLR © DLR 100m DEM simulated ERS-Image white: lay-over
Geometry of SAR Images - Shadow radar shadow Steep slopes oriented away from the SAR return no signal Fig.: © DLR azimuth range SRTM/X-SAR SRTM© DLR
SAR-EDU – SAR Remote Sensing Educational Initiative https://saredu.dlr.de/