SAR Multi-Temporal Applications

Transcription

1 SAR Multi-Temporal Applications DOC-TAS-EN-001

2 Contents 2 Advantages of SAR Remote Sensing Technology All weather any time Frequencies and polarisations Interferometry and 3D mapping Change Detection applications - Key examples from COSMO-SkyMed Environmental monitoring Damage assessment

3 3 Advantages of SAR Remote Sensing Technology

4 All weather - Any time!!! 4 A PERFECT TECHNOLOGY FOR MULTITEMPORAL IMAGERY ACQUISITION

5 Wavelength vs. Penetration 5 Variety of backscatter from vegetation Wavelength and Rate of penetration to the canopy Apart from the technological aspects, the selection of the band is function of the applications. The interaction among the electromagnetic wave and a target can be considered at the wavelength scale. The higher frequencies view better the boundary of the target, and so typically is requested for detection application; the lower frequency can better penetrate consequently can give information also under the canopy.

6 Frequency vs. Resolution 6 One of the most important parameters in SAR system is the resolution: the azimuth resolution is related to the antenna pattern characteristic and the steering capability, the range resolution is related to the used bandwidth. ITU Allocation L C C (ITU primary*) X X (WARC 2013**) Frequency [MHz] Band [MHz] Slant resolution [m] Slant weighed resolution [m] On ground resolution [m] (25 incidence) On ground resolution [m] (45 incidence) * The presence in the same spectral range of mobile application, reduce, especially for urban site, the optimal portion of the allocated bandwidth. ** It expected an increase of the allocated bandwidth for earth exploration satellite (active).

7 Polarisations Polarization is the orientation of the electric vector of an electromagnetic wave. 7 Radar system antennas can be configured to transmit (Tx) and receive (Rx) either horizontally or vertically polarized electromagnetic radiation. When polarization of the Tx and Rx waves is the same, it is referred to as co-polarized. HH means horizontally Tx and Rx waves; VV refers to vertically Tx and Rx waves. When polarization of the Tx and rx waves is orthogonal, it is referred to as cross-polarized; HV refers to horizontal Tx and vertical Rx; VH for vertical Tx and horizontal Tx. Polarization can be modified when the radar wave interacts with a surface and is scattered from it, depending upon the properties of the surface. This modification affects the way the scene appears in polarimetric radar imagery, and the type of surface can often be deduced from the image. Multiple polarizations help to distinguish the physical structure of the scattering surfaces: the alignment with respect to the radar (HH vs. VV) the randomness of scattering (e.g. vegetation - HV) the corner structures (e.g. HH-VV phase angle) Bragg scattering (e.g. oceans - VV)

8 Polarimetry 8 HH+VV HV HH-VV

9 Frequency and Polarisation vs. Applications 9

10 10 Change Detection Applications Key Examples from Cosmo-SkyMed and others

11 Change Detection Applications 11 Some of the most important applications where Change Detection techniques may be usefully exploited by means of multi-temporal SAR images analysis and comparison are: Environment Monitoring Forestry (deforestation) Agronomic pattern detection and land change monitoring Maritime and coastal surveillance Damage assessment Earthquakes Floods Volcanoes Oil spills

12 12 Environment Monitoring

13 Environment Monitoring: Forestry SAR remote sensing can be very helpful especially in those regions where vegetation is extended and impervious. Application examples are: Cover Type Mapping Deforestation Mapping Forest Flood Mapping Fire Scars Mapping 13 microwave backscatter is highly dependent on: the orientation and size distribution of the scattering elements moisture content, so that individual components of forest canopies and other vegetative covers (e.g. leaves, branches, trunks) represent discrete scattering and absorbing elements to the microwave power transmitted by imaging radar. variations in the microwave dielectric constant of vegetation elements or ground surface. At longer radar wavelengths (P- and L-bands) microwave scattering and absorption results from interactions with the tree boles and larger branches found within forests, as well as the ground surface. At these wavelengths, the smaller woody stems and the foliage act mainly as attenuators. At shorter radar wavelengths (C- and X-bands) microwave scattering and absorption results from interactions from smaller branches and leaves and needles in the canopy. The presence of a water-saturated or flooded surface leads to increased double-bounce scattering that enhances the strength of the ground-vegetation interaction term. Finally, the polarisation combination of the received backscatter is dependent on the polarisation of the transmitted microwave power and on the horizontal and/or vertical orientation of the scattering elements present in the vegetation.

14 Deforestation: human-induced regular patterns Forested area in Turingia, Germany with clear cuts, crossing of power line, re-forestation CSM Spotlight 2 image, incidence angle 47, pol. VV, acquired 3/2/08 14

15 Agricultural Mapping 15 Agricultural remote sensing is traditionally done in the visible, near-infrared and thermal infrared portions of the electro-magnetic spectrum New applications in the microwave area are under development: complementary microwave information (e.g. different frequency bands or polarisations) allow to discriminate different agriculture classes. Fast, uniform and objective across-border agricultural statistics can be obtained. Traditional ground inspection methods are time consuming and expensive, and remote sensing is increasingly implemented as a cheap and effective control instrument. Remote sensing methods also allow time-extended monitoring for comparison and crops evolutions. High-repetition, low resolution SAR images also allow the state of vegetation in large areas to be assessed. Phenological conditions are observed and correlated with historical data or are input to agro-meteorological models.

16 Technique for Thematic Mapping and Change Detection 16 Interferometric coherence provides an additional information layer related to target change or stability: Collect 2-3 images within a short time range (few days) Use multi-temporal analysis to derive a thematic map using both amplitude variations and coherence Repeat the process and look for changes in time between the two thematic maps Image collection Multi-temporal analysis Thematic map Change detection years

17 Multi-Temporal Coherent Colour Composite COHERENCE 17 First SAR IMAGE Amplitude Second SAR IMAGE Amplitude COSMO-SkyMed - Courtesy e-geos

18 Interpreting a colour composite image 18 Vegetation with negligible growth during the time interval i.e. natural /wild vegetation Texture indicates trees = COSMO-SkyMed HIMAGE ifsar pair Sept 19 Oct 5, 2008 Sept. 19 October 5 Sept 19 + Oct 5 (red+green) Coherence

19 Fucino - central Italy: agronomic pattern 19 harvested after September November September Coherence autumn growing COSMO-SkyMed - Courtesy e-geos Cultivation growing sept to nov

20 Fucino - central Italy: agronomic, forest, quarry monitoring... November September Coherence COSMO-SkyMed - Courtesy e-geos natural grass: Nov. senescence 20 coniferous bare and coherent areas quarry monitoring: bluish means... no change broadleaves

21 Maritime and Coastal Surveillance 21 Marine environmental monitoring and surveillance sensors are required to observe environmental, human-induced and natural features. Ships, oil spills and mines are few examples of humaninduced features. The table identifies coastal activities that have socioeconomic value and a critical requirement for maritime surveillance. These activities are both military and civilian in nature and that the two communities have common requirements.

22 22 Damage Assessment

23 Flooding 23 SAR is able to observe in all weather conditions timely and fast observations of the flooded areas, difficult with other means due to the usually bad weather conditions during the period of flooding Capable to perform quantitative data analysis based on backscattering coefficient. backscattering intensity can be easily exploited for detecting flooded areas without the effects by season, time and atmospheric condition. Physical basis for flooded area detection SAR backscattering intensity decreases according to land cover change from non-water surface to water surface by flooding.

24 Flooding Ref.: 24 From JERS-1/SAR (L-Band)

25 Flooding over forest areas: double reflection effects 25 SAR backscattering intensity by surface scattering is also affected by incidence angle and wavelength. larger incidence angle Backscattering intensity lower Possible to detect non-flooded areas as flooded longer wavelength Specular reflection easier Possibly difficult to detect flooded areas in bare soil or concrete shorter wavelength Effect of surface wave larger Possibly difficult to detect flooded areas in windy condition Flooding may cause opposite effect in forest areas due to the effect of double reflection by water surface and tree trunks. On the right, a multi-temporal sequence of JERS-1 SAR images show that backscattering intensity in forest areas turn to be higher by flooding (SAR) Ref.: Analysis-Application-of-SAR-Data-for-Flood-and-landslide-monitoring-Eisuke-Koisumi.pdf

26 Australia Flooding (March 2010) 26 copyright COSMO-SkyMed ASI, processed and distributed by e-geos, courtesy of e-geos COSMO-SkyMed Wide Region Acquisition: Visible water extent Semi-automated detection from Cosmo-SkyMed by e-geos Cartographic scale: 1: to any third party without the prior written permission of Thales Alenia Space , Thales Alenia Space All rights reserved 2012, Telespazio/e-GEOS

27 Oil Spills 27 The oil can reliably be detected by the C-band SAR sensor only if the wind is not too low or too high. Low wind speed makes the smooth sea surface and the oil film covered sea areas appear as undistinguishable dark areas in a SAR image. High wind speed induces waves to break an oil spill and mix it into the ocean sub-surface, making it no longer detectable by SAR imagery Useful wind speeds for reliable oil speed detection ranges between 4 m/s and 10m/s. C-band VV pol. in the range appears the most suitable SAR configuration for oil slick detection. Cross-polarization (HV or VH) seems not to add benefits, since little multiple reflection of the signal occurs over the ocean In any case, the potential of polarimetric SAR for improved oil spill detection and classification is evaluated related to the extension of the validity ranges of wind speed and incidence angles. Strength of backscattering vs. incidence angle width of potential oil slick vs wind speed

28 Oil Spills 28 Courtesy of ASI, All rights reserved Detected Oil Spill Detected Oil Spill Detected Oil Spill

29 Earthquakes 29 The temporal evolution of the Earth surface can only be monitored by performing several acquisition in the time domain: to highlight possible surface deformation or subsidence phenomena it is required to analyze historical series on the area of interest. Differential SAR interferometry (DIFSAR) allows the investigation of deformations with an accuracy that is a fraction of the radar wavelength (in the range of centimetres to millimetres). Incoherent Change detection: when a highly radiometric accurate and stable SAR is available, change detection may also be achieved by way of incoherent analyses; this increases and extends the applicable dataset and possibly detects macro-scale variations (e.g. building collapses). In order to optimise the risk analysis and damages evaluation mixed/hybrid techniques can be exploited.

30 Earthquakes: L Aquila (2009) 30 High resolution map single points displacements close to the Paganica fault Permanent Scatterers technique using 26 COSMO- SkyMed StripMap images acquisition time range: 12 april september 2010

31 Damage: Volcanoes 31 volcano activity hazard may be synthesised and represented by multitemporal maps at a very detailed scale (1: :10.000), to cover the limited areas interested by volcano eruptive action. Remote sensing data should be acquired during the monitoring phase in order to assess swelling, measurable by means of SAR differential interferometry. Effort is actually spent onto detection of any possible precursor signal for volcanic events The hope is to support development of forecast models also required to prepare intervention plans and reliable simulations. In case of a volcano activity increase, the SO2 emissions (the main precursor for eruptions) and the other activity parameters has to be known very quickly by the risk management organization headquarters