The TerraSAR-L Interferometric Mission Objectives

Manfred Zink

TerraSAR Project, ESA-ESTEC Keplerlaan 1, 2200 AG, Noordwijk, The Netherlands

Tel: +31 71565 3038, Fax: +31 71565 3191, Email: Manfred.Zink@esa.int ABSTRACT TerraSAR-L is the new imaging radar mission of the European Space Agency. The platform, based on the novel Snapdragon concept, is built around the active phase array antenna of the L-band Synthetic Aperture Radar (SAR). Specification of the L-SAR has been guided by careful analysis of the product requirements resulting in a robust baseline design with considerable margins. Besides having a commercial role for the provision of geo-information products, TerraSAR-L will contribute to the Global Monitoring for Environment and Security (GMES) initiative and serve the scientific user community. Interferometry (INSAR) is a key element behind a number of mission objectives. A L-band SAR in a 14-day repeat orbit is an ideal sensor for solid earth applications (earth quake and volcano monitoring, landslides and subsidence) relying on differential interferometry. L-band penetration of vegetation cover facilitates these applications also over vegetated surfaces. Because of the high coherence, L-band is also the preferred frequency for monitoring ice sheet and glacier dynamics. Highly accurate orbit control (orbital tube <100m) and special wideband INSAR modes are required to support these applications globally and systematically. Precise burst synchronisation enables repeat-pass ScanSAR interferometry and global coverage within the short repeat cycle. A feasibility study into cartwheel constellations flying in close formation with TerraSAR-L revealed the potential for generating Digital Elevation Models (DEMs) of unprecedented quality (2m relative height accuracy @ 12m posting). The TerraSAR-L operations strategy is based on a long-term systematic and repetitive acquisition scenario to ensure consistent data archives and to maximise the exploitation of this very powerful SAR system. 1 INTRODUCTION ESA’s new imaging radar mission, TerraSAR-L, is currently being studied in a Phase B and the preliminary design review is planned for December 2004. The TerraSAR-L system will provide Europe with its most powerful SAR programme to date. Key features of the 5-year mission are a short (14-day) repeat cycle in a Sun-synchronous dawn-dusk orbit, global imaging coverage, tight orbit control and high precision orbit determination. The L-SAR is build around an active phased array antenna and provides full polarimetric capabilities, maximum bandwidth (more than 80Mhz) within the 85MHz allocation for Earth observation in L-band, and repeat-pass ScanSAR interferometry. Such a system can serve a number of applications and will be an important complement to current and future X- and C-band SAR sensors. Because of its penetration into vegetation canopies L-band SAR has strong capabilities in land cover classification. This feature is important for applications related to Climate Change like the Kyoto inventory and wetland monitoring and in combination with X-band data for commercial services in the area of agriculture, forestry and cartography. The capability to penetrate vegetation and to interact with the mechanically more stable lower parts of the canopy is also the main reason for increased coherence levels in L-band over vegetated surfaces and facilitates applications based on differential interferometry, which up to now have been limited to urban areas and bare surface, on global scale. Monitoring seismic and volcanic activities, landslides, subsidence and glacier ice motion are the main INSAR applications. This paper provides an overview of the TerraSAR-L system and its mission objectives including a detailed discussion of the interferometric capabilities. Important considerations for the operations strategy and a summary of the main mission features conclude the publication.

____________________________________________________________

Proc. of FRINGE 2003 Workshop, Frascati, Italy,1 – 5 December 2003 (ESA SP-550, June 2004) 123_zink

2 THE TERRASAR-L SYSTEM The TerraSAR-L system comprises a spacecraft carrying a large, fully polarimetric L-Band SAR, and a complementary ground segment architecture.

Fig. 1. Annotated View of the TerraSAR-L Spacecraft

2.1 Snapdragon Configuration The TerraSAR-L spacecraft (Fig. 1), based on the novel snapdragon configuration, is optimised for and build around the large L-SAR antenna. One single deployment (see Fig. 2) of the whole spacecraft deploys the 11m by 2.86m SAR antenna. The snapdragon architecture retains the modularity of conventional platforms, offers ample space for equipment accommodation and simplifies the payload design and AIT of a large spacecraft. It furthermore permits verification by testing at high levels of integration and overall reduces the risk, costs and schedule. The spacecraft has a high agility due to low roll inertia and a low aerodynamic coefficient (lower than GOCE). A simple solar array and a simplified thermal design are further advantages of the snapdragon concept. With a total launch mass of 2.8 tons (including contingencies and system margin), the Soyuz Fregat launcher has ample margins in volume to accommodate the stowed snapdragon and in mass to place TerraSAR-L into its ~630km orbit. The solar array provides more than 5kW power for an average L-SAR consumption of ~2kW during data acquisition. Consequently, the TerraSAR-L spacecraft has margins in all aspects of its design.

2.2 The TerraSAR-L Orbit TerraSAR-L will operate from a ~630km Sun-synchronous dawn-dusk orbit with a mean local solar time of 18:00 hours on the ascending and 06:00 on the descending orbits respectively. The short revisit time of only 14 days is optimised with respect to ensuring global coverage (no gaps at lower latitudes), effort and performance of the orbit maintenance manoeuvres and is a key feature of this mission, especially for INSAR applications.

Tight orbit cothe topographorbit control A dual-frequhighly accura

2.3 L-SAR C The L-SAR beyond 80MH

Table 1 sumswath width,and Noise Eqbe acquired iminutes, Wadownlink cap The ScanSAbandwidth topass ScanSA

Fig. 2. Deployment sequence of the TerraSAR-L spacecraft

ntrol (required to be within 100m tube) will be achieved by ground-generated manoeuvres and will reduce ic noise in the DINSAR applications to 1-2mm, if DEM data of the SRTM quality are available. The tight

is combined with a requirement on the pointing accuracy of 50mdeg (3σ) in all axes.

ency GNSS receiver allows for precise orbit determination to within 5cm, and a star tracker provides te spacecraft attitude (2mdeg).

haracteristics

operates at a centre frequency of 1257.5MHz. Observing the ITU regulations a maximum bandwidth z is feasible in the allocated frequency band between 1215 and 1300MHz. Nominal modes are:

• Stripmap mode in single, dual and quad polarization • ScanSAR mode in single and dual polarization

• Wave mode (sampled stripmap, vignettes of 20x20km acquired every 100km)

marizes the main performance requirements on the L-SAR modes including the incidence angle range, azimuth and ground range resolution, Distributed Target Ambiguity Ratio (DTAR), radiometric accuracy uivalent Sigma Zero (NESZ). The nominal look direction is right-looking, but for limited periods data can n left-looking geometry. In the high rate modes the maximum operations time per orbit is limited to 20 ve mode data can be acquired over full orbits. The 20 minutes limit is driven by the data volume and acity, instrument thermal and power constraints allow more than 30 minutes of operation per orbit.

R and Stripmap Single Pol modes are especially designed for INSAR applications requiring maximum allow multi-looking for phase noise reduction and precise (<5ms) burst synchronization enabling repeat-R interferometry.

Mode Inc. Angle Swath Width Resolution azi x rg DTAR

Rad. Accuracy

(3σσσσ) NESZ

Quad Pol 20-36 deg 40 km 5 x 9 m -20 dB 1 dB -30 dB

Dual Pol 20-45 deg 70 km 5 x 9 m -20 dB 1 dB -30 dB

Single Pol 20-45 deg 70 km 5 x 5 m -20 dB 1 dB -27 dB

ScanSAR Dual Pol 20-45 deg >200 km 50 x 50 m -20 dB 1.5 dB -30 dB

ScanSAR Single Pol 20-45 deg >200 km 20 x 5 m -20 dB 2 dB -27 dB

Wave 20-45 deg 20 km 5 x 9 m -20 dB 1 dB -30 dB

Tab. 1. Summary of L-SAR mode characteristics The above performance requirements have to meet under the end-of-life assumption of 6% random module failure. Primary design driver for the L-SAR instrument and the size of the antenna aperture is not sensitivity but ambiguity performance at high incidence angles. With dimensions of 11m x 2.86m the L-SAR active phase array antenna is more than twice as big as the ASAR antenna and radiates more than twice the power of ASAR at a similar weight of ~900 kg. It consists of 160 sub-arrays fed by 160 transmit/receive modules (TRMs) arrange in 16 rows and 10 columns (panels). The antenna front-end and the snapdragon deployment mechanism are the only critical new developments of the TerraSAR-L system. Both are covered under on-going pre-development (risk retirement) activities, where in case of the antenna front-end a complete antenna panel is being designed and build. In L-band propagation disturbances and especially ionospheric effects like Faraday rotation and phase delay have to be considered and if possible corrected. Quad-pol data allow estimating and compensating the Faraday rotation from the data itself. Dual-frequency (split-band) SAR operation permits Total Electron Content (TEC) estimation and is foreseen to support the ionospheric phase screening for interferometric applications. The principal constraint on the TerraSAR-L system is transferring data to the ground. Optimised sampling schemes and advanced encoding techniques are required to reduce overheads on the instrument data rate as much as possible. The data management subsystem provides SAR data formatting, a mass memory of more than 600 Gbit and downlink encryption. Via a single channel 300 Mbit/s X-band downlink the data are transferred to a network of 3 ground stations.

2.4 Ground Segment The ground segment is composed of three subsystems: the flight operations segment, the payload data segment and the instrument calibration segment. A modular architecture and re-use of existing facilities wherever beneficial is required. The ground segment must be compatible with the needs for interoperability with other missions, in particular TerraSAR-X, which is being developed as a German national programme. There is provision to accommodate direct access stations that could download data from the TerraSAR-L spacecraft directly using their own receiving station. Encryption of the X-band downlink and the S-band uplink is required, in order to prevent unauthorized access to the TerraSAR system. A data-driven approach is the baseline for the payload ground segment, removing the need for detailed scheduling of the ground segment facilities/elements beyond the contact plan for the X-band receiving stations.

3 OVERVIEW OF MISSION OBJECTIVES An L-band SAR system provides unique contributions in several application areas and will be an important complement to future X- and C-band mission. The TerraSAR-L system has the capabilities to serve a number of highly relevant applications, which can be categorised under the following areas.

3.1 Interferometry 3.1.1 Why L-band Interferometry Interferometry is one of the most important applications of SAR, which has been mainly developed using C-band data from the ERS missions. With the exception of the ice phase (3-day repeat) and the ERS-1/-2 tandem mission (1-day repeat) C-band interferometry is limited to urban areas and bare surfaces. Assuming an ensemble of scatterers in a vegetation canopy with a given motion dispersion, the coherence from L- to C-band scales with the power of the wavelength ratio squared [1], e.g. a coherence of 0.9 in L-band corresponds to a coherence of 0.15 in C-band. This assumption is underestimating the coherence at longer wavelengths due to the penetration of the L-band signals into the vegetation volume and interaction with mechanically more stable parts underneath. The increased levels of coherence over vegetation are a key advantage of L-band allowing to extend the successful C-band applications to global scales. Fig. 3 shows a comparison of JERS-1 and ERS-1 interferograms with comparable temporal baselines and height of ambiguity over the Fuji City area indicating similar coherence levels over the urban part and increased L-band coherence over the vegetated mountain slope in the left part of the images.

FcsJ Artibt Faiso

Fig. 3. Comparison of L-band (JERS-1 on the left) and C-band (ERS-1 on the right) coherence for a 10 month temporal baseline and similar height of ambiguity (courtesy of F. Rocca, POLIMI)

irst analyses of permanent scatterers (PS) in L-band using JERS-1 data indicate that the reduced sensitivity (factor of 4 ompared to C-band because of the wavelength ratio) will be compensated by the increase of coherence [1]. Advanced ensor characteristics (higher SNR, better orbit control and knowledge, higher resolution) of TerraSAR-L compared to ERS-1 will further improve the interferometric quality.

lso over snow and ice coherence is better preserved at L-band than at higher frequencies (see Fig. 4). This is of elevance for monitoring ice sheet and glacier dynamics with the DINSAR technique, as well as for speckle/feature racking. Steep phase gradients in shear zones of fast flowing glaciers and ice streams cause problems for nterferometric ice motion analysis. This can be resolved by selecting short time intervals (typically 1 to 3 days for C-and), higher bandwidth, and/or lower radar frequencies. The 14-day repeat cycle and increased bandwidth offered by he TerraSAR-L system are well suited for most glaciers and ice streams.

or a given surface motion velocity and Signal to Noise Ratio (SNR), C-band due to the shorter wavelength is pproximately four times more sensitive than L-band. As already mentioned above less temporal decorrelation and mproved sensor design (better SNR, higher resolution) compensates this loss in sensitivity. However, lower motion ensitivity is no general drawback and to the contrary is useful to unambiguously capture faster motion as being bserved over active landslides and fast moving ice streams and glaciers.

3 Abcgaam Amwlio3a HrbaTe FToto

L-band X-bandC-bandL-bandL-band X-bandX-bandC-bandC-band

Fig. 4. Comparison of L-, C-, X-band 1-day coherence (from October 1994 SIR-C/X-SAR mission) over Moreno Glacier (in the center left part of the images), Patagonia, indicating the increased coherence in L-band (courtesy of H.

Rott, University of Innsbruck)

.1.2 Interferometric Capabilities of TerraSAR-L

limitation of the current C-band missions is the long repeat cycle, which is a result of the limited swath width and the asic requirement for global coverage. TerraSAR-L will operate in a 14-day repeat orbit and offer global interferometric overage using its ScanSAR Single Pol mode. Because of the higher coherence in L-band the 14-day repeat cycle is a ood compromise between acceptable temporal decorrelation and global coverage. A bi-weekly cycle is also dvantageous from the operations point of view (phased with the lunar disturbances and the weekly operations cycle) nd helps to achieve the tight orbit tube requirement. Furthermore such a short revisit is expected to improve thematic apping applications.

n important feature of the ScanSAR Single Pol mode is repeat-pass burst synchronization to within a few illiseconds, which can be achieved by using the GNSS receiver to trigger the instrument command execution and hich will enable repeat-pass ScanSAR interferometry. With only 3 sub-swath and maximum burst duration (single-

ook ScanSAR) we achieve an azimuth resolution of 20m at a total swath width of more than 200km. Besides its mportance for global coverage, the increasing overlap of the ScanSAR swath with increasing latitude offers the pportunity to image one and the same target under different incidence angles. This improves the reconstruction of the D components of the motion vector [2] and is also helpful under difficult geometric conditions in mountainous regions nd in urban areas.

igh (more than 80MHz) bandwidth is another key feature for an interferometric system. We are not aiming at high-esolution images but do use the available bandwidth for multi-looking and phase noise reduction. High resolution ecomes important if we want to apply the speckle/feature tracking approach to monitor glacier motion. The large vailable system bandwidth allows operation in a split-band mode with frequency separation of up to 70MHz to provide EC information to support the ionospheric phase screen. If data at full bandwidth are acquired, the TEC can be stimated by comparing extreme range looks.

or DINSAR applications the interferometric baselines shall be as small as possible to reduce the topographic “noise”. his poses strong requirements on the orbit control and for TerraSAR-L we aim at keeping the spacecraft within a 100m rbit tube. The corresponding topographic noise would be 1-2mm, if DEM data of the SRTM quality are available. The ight orbit control is complemented by a highly precise knowledge of the orbit (5cm), which can be achieved via precise rbit determination from the GPS measurements.

3.1.3 Interferometric Mission Objectives TerraSAR-L is well suited to serve Solid Earth applications like monitoring of seismic and volcanic activities. These were the main objectives of EVINSAR, which has been proposed as an Earth Explorer mission. G. Wadge describes in a separate paper, that TerraSAR-L in fact can meet the EVINSAR requirements to a large extent with even slightly improved performance in the vertical and East-West components of the 3D motion vector, but some degradation on the North-South component due to the higher orbit inclination and systematically only right-looking acquisitions [3]. As explained above our system will also provide important contributions to subsidence and landslides, especially for vegetated surfaces and fast motions. Another important application is the monitoring of ice sheet and glacier dynamics, which are direct indicators of global warming and climate change. Up to now only case studies from the ERS ice phase and tandem mission are available and the urgent need for systematic monitoring of the major ice sheets and mountain glaciers is obvious. In addition to the above DINSAR applications an L-band SAR is also a favourite candidate for single-pass interferometry using a cartwheel constellation. In parallel to the main activities on TerraSAR-L a joint study team from CNES and DLR investigated the potential DEM generation performance of constellations of three receive-only micro-satellites flying in formation with TerraSAR-L [4]. The study reveals that such constellations are able to produce DEMs fulfilling the HRTI-3 (High Resolution Topographic Information) requirements: 2m/10m relative/absolute height accuracy at a 12m posting.

3.2 Climate Change Monitoring of the compliance to the Kyoto Protocol requires quantification of areas subject to land use change with respect to Afforestation, Reforestation and Deforestation (ARD). L-band SAR has strong capabilities in land cover classification in general, but especially in forest/non-forest area delineation. In addition, because of increased interaction depth in forest canopies, the L-band is also more sensitive to biomass changes than shorter wavelengths and reaches saturation at biomass levels of 50 t/ha, which enables the identification of afforestation and reforestation [5]. Beyond the detection of changes, biomass estimates as such are required for global carbon cycle science. Another important and unique application of L-band SAR is wetland monitoring [6]. Reversing the global trend of wetland degradation and destruction is the objective of the UN’s Ramsar Convention. Natural and anthropogenic wetlands (rice cultivation) are also sources of methane, one of the most effective greenhouse gases.

3.3 Land Cover Classification Highly accurate land cover classification into a number of individual classes requires a full-polarimetric L-band sensor. Due to the complementary properties of the L- and X-band backscattering joint products from TerraSAR-L and TerraSAR-X enable even higher levels of classification performance necessary for crop monitoring and forest inventory. Dual frequency combinations are also required for cartographic maps of different thematic content and scale. A full-polarimetric L-band SAR is the ideal sensor for soil moisture retrieval including surfaces with vegetation cover. Monitoring flood extent and supporting flood forecast in providing information on vegetation cover are further applications benefiting from TerraSAR-L products.

3.4 Marine Applications TerraSAR-L offers a Wave mode like the one on ENVISAT or ERS for retrieving ocean wave spectra as input to meteorological models. Shallow water bathymetry and monitoring of surface current fronts and internal waves are areas, where L-band is expected to provide unique contributions. Monitoring the extent and classification of sea ice is important for ship routing and climate change. L-band dual and quad pol data will provide improved sea ice classification capabilities.

4 SYSTEMATIC OPERATIONS STRATEGY Providing long-term systematic and repetitive observations over large areas is one of the major strengths of remote sensing technology, in particular for microwave sensors, which are not limited by low sun angle or persistent cloud cover. Our best example is the archive of data collected by ERS-1/-2 missions. For a system with basically one mode of operation a consistent and useful archive can be achieved without major planning efforts. For a more powerful system like TerraSAR-L, which can serve a number of quite different user needs, new strategies have to be implemented to optimise the mission exploitation. We plan to identify key driving applications early in the preparation phase and to establish a systematic observation plan. Key elements of such a systematic acquisition scenario are: adequate and consistent sensor modes, adequate repeat cycle, and adequate timing of the acquisition to account for seasonal changes. Long-term continuity is another important factor to guarantee consistent data archives. Pre-launch, this planning can be performed and potential conflicts identified and resolved, resulting in an optimised use of the system resources. 5 CONCLUSIONS TerraSAR-L is a very powerful SAR system based on a robust design with considerable margins. The spacecraft is based on the snapdragon architecture, which is optimised for large SAR antennas. Repeat-pass ScanSAR interferometry, more than 80MHz bandwidth and full polarimetric capabilities are the key characteristics of the L-SAR. The 14-day repeat cycle provides global coverage and enhanced performance for INSAR applications. Besides major contribution to applications in areas of climate change and oceanography, the TerraSAR-L design responds specifically to requirements coming from interferometric applications. Joint products from TerraSAR-L and TerraSAR-X are important for commercial services relying on detailed land cover classification products. A systematic operations strategy will ensure optimum use of the system resources, consistent data archives and maximised exploitation of the TerraSAR-L mission. 6 REFERENCES 1. K. Daito, et.al., L-band PS Analysis: JERS-1 Results and TerraSAR-L Predictions, Proceedings of Fringe 2003,

ESRIN, Frascati, 01-05 Dec 2003. 2. F. Rocca, 3D Motion Recovery from Multiangle and/or Left Right Interferometry, Proceedings of Fringe 2003,

ESRIN, Frascati, 01-05 Dec 2003.

3. G. Wadge & B. Parsons, Achieving the EVINSAR Objectives with TerraSAR-L, Proceedings of Fringe 2003, ESRIN, Frascati, 01-05 Dec 2003.

4. M. Zink, G. Krieger & T. Amiot, Interferometric Performance of a Cartwheel Constellation for TerraSAR-L, Proceedings of Fringe 2003, ESRIN, Frascati, 01-05 Dec 2003.

5. A. Rosenqvist, et.al., Support to Multi-national Environmental Conventions and Terrestrial Carbon Cycle Science

by ALOS and ADEOS-II – the Kyoto & Carbon Initiative, Proceedings IGARSS 2003, Toulouse, France, 21-25 July 2003.

6. A. Rosenqvist, et.al., The Use of Spaceborne Radar Data to Model Inundation Patterns and Trace Gas Emission in

the Central Amazon Floodplain, International Journal of Remote Sensing, Vol. 23, No. 7, pp. 1283-1302, 2002.