Surface Displacement Measurement from Remote Sensing Images. Olivier Cavalie
Читать онлайн книгу.
as numerous added payloads, among which were the DORIS instrument for precise orbit determination and the MERIS spectrometer, giving simultaneous information on the composition of the atmosphere and overlapping most of the SAR swath. Envisat was put on the same 35-day orbit as the ERS series from April 2002 until October 2010, when a drifting orbit had to be set due to the lack of propellant until its end in May 2012.
ERS-2 had an overlap in the years 2007–2008 with Envisat, with a 30 min delay between the two satellites, but the central frequency was shifted from 5.3 GHz (ERS) to 5.33 GHz (Envisat) so that cross-intereferometry was restricted to common bandwidth in limited cases.
The geometry of ERS satellites was quite simple with a 23° fixed incidence angle, swath of about 100 km and one single polarization set to VV. The Envisat ASAR instrument was more complex, with modes divided in ScanSAR and its large swath of 405 km (composed of five sub-swaths), as well as seven stripmap modes (IS1 to IS7). A wave mode was operated over the ocean. Customers could order any kind of mode, so the archive was not homogeneous over the Earth and strongly depended on the area considered: it therefore became more difficult to get two images acquired in the same mode, suitable for interferometry. However, for some closely monitored volcanic islands, it was quite interesting to get different passes in the 35-day cycle with different incidence angles.
Note also that ASAR was not full polarimetric, but authorized different kinds of scheme, such as the alternating polarization (AP) mode, which could emit alternatively in H and V polarizations and receive H and V.
Some limitations of the missions with regard to interferometry: As these missions were decided before differential interferometry emerged, some limitations had to be tackled (Solaas and Coulson 1994). For ERS-1/2, orbital housekeeping looseness led to perpendicular baselines B⊥ (see section 4.2.2 in Chapter 4) higher than the limit of 1,100 m. A preferred value of 600 m was set for B⊥ (see definition in section 4.2), so that a tool to determine which orbits or track frames were compatible for interferometry had to be developed. In addition, the orbit was not known as precisely as the latest SAR missions: orbital fringes had to be removed before interpretation. PRARE, a dedicated instrument for precise orbit determination, failed after several weeks on ERS-1 but lasted several years on ERS-2 and was able to deliver precise ranging accuracy of less than 7 cm.
In contrast to future missions (e.g. ALOS-2 or Sentinel-1), Envisat ScanSAR bursts were not synchronized on board, meaning the Doppler overlaps had random values from one pass to another and were not always compatible. Some results were however published in ScanSAR mode; for instance, for the 2003 Bam earthquake in Iran.
In late October 2010, Envisat got a new orbit cycle of 30 days in a degraded configuration. At this point, interferometry could only work around latitudes of 38° (north and south). As a result, tracks could not be combined with the previous 35-day orbit cycle. However, along-track interferometry in the new geometry could produce interesting results, for example interferograms for the Sendai (Tohoku) earthquake that occurred on March 11, 2011, which led to the Fukushima disaster.
Data access: During the missions, European Space Agency (ESA) products were not free of charge and RAW data were much cheaper. Furthermore, some technical teams preferred to do the radar synthesis themselves to ensure a better signal shape in the range spectrum. Nowadays, as indicated by the ESA (2019), the (A)SAR on-the-fly (OTF) data processing and dissemination service allows end-users to gain direct access to ESA’s complete high-resolution (A)SAR archive from the ERS-1, ERS-2 and Envisat missions. The service provides (A)SAR Level 1 high-resolution data, which are processed from Level 0 on user request by the system. Users can register themselves and immediately download Level 1 (A)SAR data products. The service is available via the ESA Online Dissemination Service at https://esar-ds.eo.esa.int/oads/access.
1.2.2. Canadian C-band satellites: Radarsat-1, Radarsat-2 and RCM
Radarsat-1 was the first imaging radar developed for the Canadian Space Agency (CSA). It was launched in December 1995 and ended its operations in April 2013. The satellite had noticeable activity over the Arctic areas for ice monitoring. In the late 1990s, it was the only civilian radar capable of imaging part of the Earth, while ERS could only operate in the visibility of ground receiving stations due to the lack of onboard recording. The satellite offered a variety of imaging modes, including the large swath in ScanSAR (510 km) and narrow ScanSAR (305 km), seven stripmaps (100 km) and three others with 150 km swath, ranging from 20 to 47°, as well as 15 fine modes at incidences above 37° (45 km). The radar operated only in single co-polarization HH. The satellite native design was right-looking imaging, but the CSA conducted a dedicated campaign over Antarctica with a 180° rotation of the spacecraft, making it left-looking, and the first radar mosaic of the continent could thus be produced. Interferometry was demonstrated during the mission, although there were some strong limitations with this spacecraft. Radargrammetry for DEM was also demonstrated, prior to the SRTM mission.
There were several limitations of the mission with regard to interferometry:
– the satellite was not yaw-steered, meaning the Doppler centroid had strong values, exceeding several PRFs, as well as strong variations in the year, and for interferometry, we should take care when combining pairs, the minimum and maximum of Doppler shift being around the solstices;
– the orbit-keeping was quite loose, and baselines reached several kilometers;
– orbital knowledge was a special issue in computing correct interferograms, and the use of fine orbit interpolators, such as Hermite polynomials, was a necessity.
The Radarsat-2 spacecraft was launched in December 2007 while Radarsat-1 was still operating. The spacecraft added significant improvements in many areas: right-and left-looking, quad polarization, low σ0, ultrafine modes, ground track maintained within a 500 m wide range, yaw steering and onboard state recorders. Although on the same reference orbit as Radarsat-1, Radarsat-2 images were not pairable with Radarsat-1 for interferometry: the central frequency was shifted from 5.3 to 5.405 GHz and the Doppler centroid was not compatible, due to the difference in yaw steering. At the end of 2020, the spacecraft was still operating, making a transition to the next Canadian SAR mission.
The Radarsat Constellation Mission (RCM) aims to replace the Radarsat-2 mission, continuing the traditional Canadian C-band SAR imagery as well as adding automatic identification system (AIS) payloads to improve maritime surveillance. It is composed of three identical satellites that were launched together in June 2019 for an expected lifetime of seven years. The sun-synchronous constellation has a 18:00 ascending node and flies at an altitude range between 586 and 615 km. As the three satellites are equally spaced along the orbit, the repeat time is only four days, and the orbit is maintained within a 120 m radius orbital tube, which gives good advantages for several InSAR applications. Various schemes of polarization are implemented: single polarization, dual co-cross or compact polarimetry are available on all modes; dual HH–VV is available for specific modes, as well as a quad-polarization mode. Radar modes are numerous, from spotlight (1×3 m resolution, one look) to wide ScanSAR (100 m, eight looks)
Data access: Since April 2019, 36,000 images of archive data acquired by the Government of Canada all over the world have been freely available (see CSA 2019). Unfortunately, not all acquired data can be accessed at the time of editing (2021).
1.2.3. Japanese L-band satellites: JERS-1, ALOS and ALOS-2
The Japanese Earth Resources Satellite (JERS-1) was the first of a series of Japanese SAR sensors working in the L-band, followed later by the Advanced Land Observing Satellite (ALOS) and ALOS-2. All three projects were conducted under governmental institutions. Launched in February 1992, JERS-1 lost its onboard