Remote Sensing of Environment 120 (2012) 9–24

Contents lists available at SciVerse ScienceDirect

Remote Sensing of Environment

journal homepage: www.elsevier.com/locate/rse

GMES Sentinel-1 mission

Ramon Torres a, Paul Snoeij a,⁎, Dirk Geudtner a, David Bibby a, Malcolm Davidson a, Evert Attema a,
Pierre Potin a, BjÖrn Rommen a, Nicolas Floury a, Mike Brown a, Ignacio Navas Traver a, Patrick Deghaye a,
Berthyl Duesmann a, Betlem Rosich a, Nuno Miranda a, Claudio Bruno b, Michelangelo L'Abbate b,
Renato Croci b, Andrea Pietropaolo b, Markus Huchler c, Friedhelm Rostan c
a
ESA/ESTEC, The Netherlands and ESA/ESRIN, Italy
b
Thales Alenia Space Italia, Italy
c
EADS Astrium GmbH, Germany

a r t i c l e i n f o a b s t r a c t

Article history: In the frame of the Global Monitoring for Environment and Security (GMES) Space Component programme,
Received 20 December 2010 the European Space Agency (ESA) undertook the development of a European Radar Observatory (Sentinel-1),
Received in revised form 14 April 2011 a polar orbiting two-satellite constellation for the continuation and improvement of SAR operational services
Accepted 7 May 2011
and applications. Satellite and payload are being built to provide routine, day-and-night, all-weather medium
Available online 18 February 2012
(typically 10 m) resolution observation capability. Ground infrastructure is provided for planning, mission
Keywords:
control, data processing, dissemination and archiving. Free and open data access is provided. Data quality
CSAR of the Sentinel-1 data products is shown along with uncertainty estimation of retrieved information products
Space borne confirming specified performance and indicating application growth potential. The unique data availability
C-Band performance of the Sentinel-1 routine operations makes the mission particularly suitable for emergency re-
Satellite sponse support, marine surveillance, ice monitoring and interferometric applications such as detection of
Remote Sensing subsidence and landslides.
GMES © 2012 Elsevier Inc. All rights reserved.
Constellation

1. Introduction for scientific discovery, revolutionised many Earth science disci-

The short-lived but successful US SEASAT satellite provided a
first glimpse of the potential of imaging radar from space in
1978. ESA's own programme to develop advanced microwave
radar instruments culminated with the launches of ERS-1 (17
July 1991) and ERS-2 (20 April 1995). ERS demonstrated for the
first time the feasibility of flying reliable, stable and powerful
radar imaging systems in space. The dependability and all-
weather capability of the instruments also provided a foundation
for developing and exploiting radar images for a wide variety of
applications. While the initial objectives for ERS-1 at launch
were predominantly oceanographic, other applications were con-
sidered during the project's preparation. The ESA Remote Sensing
Advisory Group in 1974, for example, emphasised commercial ap-
plications such as agriculture, land-use mapping, water resources,
overseas aid and mapping of mineral resources in its advice on
ERS objectives. The rigorous design of the ERS SAR hardware –
emphasising instrument stability in combination with accurate
and well-calibrated data products – created new opportunities
⁎ Corresponding author.
E-mail address: paul.snoeij@esa.int (P. Snoeij).
plines and laid the foundations for commercial applications.
For example, ‘SAR interferometry’, which can track land shifts of
only a few millimetres, was developed mainly using ERS data and is
now commonly used in Earth sciences and commercial applications.
The potential of space radars viewing the same scene only a short
time apart was demonstrated in 1995 and 1996 during the ERS ‘tan-
dem mission’, when the orbits of ERS-1 and ERS-2 were carefully
matched but with a 1-day gap.
An important milestone was the launch of the Advanced SAR
(ASAR) on Envisat on 28 February 2002. This ensured the contin-
uation of C-band data and added enhanced capabilities such as
wide swaths and dual polarisation, features that have since rapid-
ly been integrated into and exploited by many applications. The
archive of radar data since 1991 is extremely valuable for science
and applications, providing a consistent set of data spanning
16 years.
The ESA Sentinels constitute the first series of operational satel-
lites responding to the Earth Observation needs of the European
Union (EU) — ESA Global Monitoring for Environment and Security
initiative. The GMES Space Component (GSC) programme relies
strongly on new complementary developments by ESA in addition
to existing and planned space assets from different agencies. As part
of the GSC, ESA is currently undertaking the development of 3

0034-4257/$ – see front matter © 2012 Elsevier Inc. All rights reserved.
doi:10.1016/j.rse.2011.05.028
10 R. Torres et al. / Remote Sensing of Environment 120 (2012) 9–24

Table 1
Sentinel-1 mission requirements.
Data availability
1 In order to avoid service provision gaps Sentinel-1 operations shall start be-
fore the end of the operational lifetime of ERS/Envisat and last for a period of
at least 10 years.
2 Unless specified otherwise performance shall satisfy ERS/Envisat
requirements.
3 End-to-end product availability to the end user shall typically be better than
95%, based on a main mode of observation and on systematic operations with
on-demand observations in exceptional cases.
4 Coverage, revisit and exact repeat observation necessitate the use of non-ESA
satellites. Therefore similarity of data products and observation geometry is
required as well as ground segment interoperability.
5 All products shall be processed to Level-0 and Level-1 and archived.
Note: Level-0 is reformatted, time-ordered satellite data, in computer-
compatible format. Level 1-products are geo-located products in which data
has been converted to engineering units, auxiliary data has been separated
from measurements, and selected calibrations have been applied to the data.
6 Data shall be delivered to the user within the timeliness specification from the
archive or in near real time (NRT).
Coverage and revisit
7 For interferometry global exact repeat coverage shall be achieved bi-weekly
(e.g. with an interval of less or equal 14 days).
8 Fast global access on demand shall be provided within the timeliness
specification.
9 Daily full coverage shall be achieved north of +45° latitude north and south of
−45°.
Note: A constellation of 4 satellites would satisfy the corresponding
observation requirement. As the proposed concept initially includes only two
satellites an average revisit once every two days is achieved. Consequently
collaboration and interoperability with non-ESA satellites would be required.
Timeliness
10 Near real time data shall be delivered to users within 3 h of observation.
11 User request for archived data shall be completed within typically 24 h and in
no case within more than 3 days after issuing of a user request.
12 Emergency operations shall be completed typically within 24 h and in no case
within more than 48 h after issuing of a user request.
Characteristics of data products
13 The centre frequency shall be selected within 5250–5570 MHz (C-Band).
14 The system shall be designed to support interferometry.
15 A 240 km swath width follows from revisit requirements for the main mode of
observation.
16 A Wave Mode shall be provided with 20 × 20 km imagettes every 100 km
along track
17 The main mode of observation shall have both VV and VH polarisation.
18 Optional additional modes shall be provided including a mode with both HH
and HV polarisation.
Sentinel mission families. 1 Each Sentinel is based on a constellation of
2 satellites in the same orbital plane. This configuration allows to fulfil
the revisit and coverage requirements, and to provide a robust and af-
fordable operational service. The life time of the individual satellites is
specified as 7 years, with consumables on-board each satellite allow-
ing a mission extension up to 12 years. The life cycle of a satellite gen-
eration is planned to be in the order of 15–20 years.
and ENVISAT ASAR. The two-satellite constellation offers six day exact
repeat and conflict-free operations based on two main operational
modes allowing exploiting every single data take. In comparison with
its predecessors the Sentinel-1 mission represents an n-fold evolution
of capability as expressed by the figure-of-merit developed in
Section 9.5. In the framework of international interoperability agree-
ments the effective revisit and coverage performance may be further
improved by access to the planned Canadian C-band SAR Constellation
(RADARSAT Constellation Mission, see for example Flett et al. (2009)).
Observation requirements from GMES services are defined in terms
of data availability, coverage and revisit, timeliness and characteristics
of data products. The service requirements for C-band SAR observations
were translated into 18 Sentinel-1 mission requirements shown in
Table 1, providing the basis for the mission design, Attema (2005).
To deal with user requirements for both high and medium resolu-
tion data conventional SAR system designs include different opera-
tional modes that either optimise the spatial resolution (at the
expense of the swath, hence the coverage) or the swath width (at
the expense of the resolution). Taking account of data access through
GMES to complementary national very high-resolution SAR missions
(TerraSAR-X by DLR/Astrium GmbH, Cosmo-SkyMed by ASI)
Sentinel-1 has been designed to address primarily medium to high
resolution applications through a main mode of operation that fea-
tures both a wide swath (250 km) and high geometric (5 m × 20 m)
and radiometric resolution. Over sea–ice and polar zones or certain
maritime areas, an extra-wide swath mode may be used to satisfy
the observation requirements of certain service providers (e.g. sea–
ice monitoring), in order to ensure in particular wider coverage and
better revisit time by sacrificing geometric and radiometric
resolution.
As the operation follows a pre-programmed conflict-free scenario
there is no need to make data acquisition requests for these modes.
The imaging modes can be operated for a maximum of 25 min per
orbit. The remaining time the instrument operates over the open
ocean in the Wave Mode providing sampled images of 20 × 20 km at
100 km along the orbit with low data rate.
It is expected that Sentinel-1A will be launched mid 2013 and
Sentinel-1B about 18 to 24 months later.
3. Sentinel-1 satellite
Sentinel-1 (see Fig. 1) is being realised by an industrial consortium
lead by Thales Alenia Space Italy as Prime Contractor, with Astrium
Germany being responsible for the CSAR payload, incorporating the
central radar electronics sub-system developed by Astrium UK.

2. The Sentinel-1 mission

Sentinel-1 is designed to work in a pre-programmed conflict-free

operation mode, imaging global landmasses, coastal zones, sea–ice,
polar areas, and shipping routes at high resolution, and covering the
global ocean with imagettes. This ensures a reliability of service re-
quired by operational services and a consistent long-term data archive
built for applications based on long time series. Sentinel-1 revisit and
coverage are dramatically improved with respect to the ERS-1/2 SAR

1
In addition, two units of Sentinel-4 instruments to be embarked on Meteosat Third
Generation (MTG) meteorological satellites and a Sentinel-5 Precursor mission have
been adopted as part of the GSC Programme. Fig. 1. Artist impression of Sentinel-1 in orbit.
 R. Torres et al. / Remote Sensing of Environment 120 (2012) 9–24 11

The spacecraft is based on the PRIMA (Piattaforma Italiana Multi
Applicativa) Platform with a mission specific payload module, in
this case the CSAR Instrument. The design of the satellite leverages
experience gained from the Canadian RadarSat-2 and the Italian
Cosmo-Skymed programmes for which PRIMA was also used.
3.1. Platform
The spacecraft designed for the Sentinel-1 mission is a 3-axis stabi-
lised satellite, directly placed into the Earth orbit by the launcher. It is
characterised by sun, star, gyro and magnetic field sensors, and a set
of 4 reaction wheels, dedicated to orbit control and to attitude control
and 3 torque rods as actuators, to provide steering capabilities on each
axis. The spacecraft (S/C) is equipped with 2 solar array wings capable
of producing 5900 W (at End of Life) to be stored in a modular battery.
The Platform provides highly accurate pointing knowledge (better
than 0.004°) on each axis, high pointing accuracy (about 0.01° on
each axis), and real time orbit determination together with a dedicat-
ed propulsion system for precise orbit control. The reference orbit
will be maintained within an Earth-fixed orbital tube of a diameter
of 100 metre-rms during the nominal mode operation time. The sta-
tistical definition of the orbital tube implies the comparison of the
perturbed orbit with respect to the reference orbit at the same argu-
ment of latitude. Sentinel-1A and Sentinel-1B shall have the same ref-
erence orbit with a 180 degree orbital phasing difference.
Computer, avionics and the satellite communication backbone are
based on the MIL-1553 command bus with the ERC32-SC processor
providing high performance computing and processing capabilities
on-board.
The total mass of the spacecraft at launch is around 2300 kg.
The Sentinel-1 spacecraft main characteristics are shown in Table 2.
The Platform provides features for the management of the Atti-
tude and Orbit Control Systems (AOCS), Data Handling and Command
and Control functions of the Spacecraft and all the necessary support-
ing functions.
The main Platform functions are:
● AOCS including:
○ Precise attitude measurements
○ Stable and precise 3 axis pointing
○ Precise orbit determination
● Propulsion:
○ Orbit injection
○ Orbital tube maintenance
Table 2
Sentinel-1 spacecraft characteristics.
Lifetime 7 years (consumables for 12 years)
○ Attitude control during Ultimate Safe Mode
○ End of life disposal
● Spacecraft Data Handling including:
○ Telecommand reception, handling and distribution
○ Telemetry collection, packetisation and forwarding to ground
by TT&C communication equipment
○ Time and synchronisation signals management also in support
of payloads
● Spacecraft autonomy and failure detection identification and
recovery:
○ Payload and spacecraft status monitoring, management and
recovery
○ S/C Operations Planning support (Mission Time Line and Sub
schedule) housekeeping telemetry and command
● Power:
○ Power generation and storage
○ Power conditioning, protection, distribution and switching
○ Pyrotechnic devices actuation
● Thermal control:
○ Temperature monitoring
○ Heat generation and removal
● Structure:
○ Primary structural support and interface on the launch vehicle
○ Appendages tie-down points and release mechanisms
● Communication with Ground:
○ Spacecraft telemetry transmission and telecommand reception
via S-Band transponder
○ Instrument telemetry transmission via X-Band Payload Data
Handling Terminal
The Platform general architecture is based on three main structure
modules and functionally decoupled to allow the parallel modules in-
tegration and testing activities up to the satellite final integration.
The modules are:
● the Service Module carrying only Platform units apart from the
propulsion ones;
● the Propulsion Module carrying all the propulsion items con-
nected by propellant lines;
● the Payload Module carrying al the payload equipment including
the pertinent appendages.
Most of the Propulsion module is enclosed in the Service module,
being integrated into the cone section interfacing the S/C to the
launcher adaptor, while the Payload module is mounted onto the Ser-
vice module allowing the Payload (P/L) units/appendages allocation
through four lateral panels and the upper platform.

Orbit Near Polar Sun-Synchronous @ 693 km;

3.2. Payload Data Handling Terminal
12-day repeat cycle; 175 orbits per cycle
Mean local solar time 18:00 at Ascending Node The PDHT comprises all functions necessary for the acquisition,
Orbital period 98.6 min storage and handling of data generated by different sources:
Max eclipse duration 19 min
Attitude stabilisation 3 axis stabilised ● CSAR data
Attitude accuracy 0.01° (each axis)
● GPS raw data (or auxiliary data) and S/C telemetry data
Nominal flight attitude Right looking
Steering Zero Doppler yaw steering and roll steering In addition to data storage and handling functions the PDHT pro-
(− 0.8 to 0.8°)
vides data formatting and transmissions capability of the data to the
Attitude profile Geo-centric and geodetic
Orbit knowledge 10 m (each axis, 3 sigma) using GPS ground stations.
Operative autonomy 96 h Direct downlink transmission capability is available for all the ac-
Launch mass 2300 kg (incl. 130 kg monopropellant fuel) quisition modes of the CSAR in single or dual polarisation.
Dimensions (stowed) 3900 × 2600 × 2500 mm3
The PDHT is build up from the following subsystems: Data Storage
Solar array average power 5900 W (End-of-Life)
Battery capacity 324 Ah and Handling Assembly (DSHA), Transmission Assembly (TXA) and
Spacecraft availability 0.998 X-band Antenna Assembly (XBAA).
S-Band Telemetry, Tracking & 64 kbps Telecommand; 128 kpbs/2 Mbps The on-board memory has been sized to 1410 Gb. The memory
Command (TT&C) data rates Telemetry (programmable) has been designed to enable simultaneous storage of data from the
Launcher Soyuz from Kourou
two polarisation channels of the CSAR instrument and from
12 R. Torres et al. / Remote Sensing of Environment 120 (2012) 9–24
Housekeeping Telemetry (HKTM), and auxiliary data for data ac-
quired during at least two orbits.
The PDHT storage equipment is based on solid-state memory de-
vices. The number of packet stores has been defined in order to
allow prioritising of the stored data during downlink activities with
at least 4 levels of priority, and temporarily storing data to be trans-
mitted in direct downlink.
The main tasks of the TXA are:
● Provide the necessary performance to meet the communication
link requirements;
● Provide full redundancy and cross-strapping for each function;
● Receive discrete and serial telecommands to allow its configura-
tion in flight and for on-ground testing;
● Provide discrete and serial telemetry to allow its control in flight
and for on-ground testing;
● Interface the satellite power bus and generate the necessary
power supply interfaces for TXA equipments.
The X-Band Antenna Assembly provides transmission to ground of
the X-Band modulated signals received by the TXA. One of the main
tasks of the XBAA is that of limiting the radiation of RF energy outside
the nominal beam coverage, to avoid interference with the spacecraft
environment and especially with the CSAR antenna.
The PDHT is able to support the mission operations (data storage
and data downlink) by performing:
● data acquisitions from the CSAR potentially for the full extent of
the orbit duration and on all orbits;
● a continuous dump equal to 20 min;
● an overall dump duration within any 100 min window of up to
30 min.
4. Sentinel-1 CSAR instrument
4.1. Modes of operation
Following the Sentinel-1 mission requirements one main opera-
tional imaging mode the Interferometric Wide-swath mode (IW) is
Fig. 2. CSAR antenna folded around the platform.
provided that, in combination with the Wave mode (WV) satisfies
most currently known service requirements. Additionally the mutual-
ly exclusive Strip Map mode (SM) and the Extra Wide-swath mode
(EW) are provided for continuity reasons (with respect to ERS and
Envisat) as well as for accommodation of emerging user require-
ments. Two mutually exclusive dual polarisation modes are provided
for the imaging modes. A summary of the SAR instrument measure-
ment modes and their characteristics is given in Table 3.
Except for the Wave Mode, which is a single polarisation mode
(selectable between HH and VV), the CSAR instrument supports oper-
ation in dual polarisation (HH + HV, VV + VH) implemented through
one transmit chain (switchable to H or V) and two parallel receive
chains for H and V polarisation. The specific needs of the four different
measurement modes with respect to antenna pointing require the
implementation of an active phased array antenna.The CSAR antenna
folded around the platform is presented in Fig. 2.
To meet the demanding image quality and swath width require-
ments, the IW mode is implemented as a ScanSAR mode with pro-
gressive azimuth scanning. This requires a fast antenna beam
steering in elevation for ScanSAR operation, i.e. transmitting a burst
of pulses towards a sub swath. In addition, fast electronic azimuth
scanning has to be performed per sub swath using the Terrain Obser-
vation by Progressive Scan (TOPSAR) operation (De Zan and Monti
Guarnieri (2006)) in order to harmonise the performance in along
track direction and reduce scalloping. Hence, the IW mode will
allow combining a large swath width (250 km) with a high geometric
resolution (5 m × 20 m on ground). Interferometry is ensured
through sufficient overlap of the Doppler spectrum (in the azimuth
domain) and the wave number spectrum (in the elevation domain).
The Wave mode data product is composed of single strip map op-
erations with an alternating elevation beam (between 23 and 36.5
mid incidence angle) and a fixed on/off duty cycle, which results in
the generation of vignettes of 20 km × 20 km size with a 5 m × 5 m
on ground resolution in regular intervals of 100 km.
In Strip Map mode the instrument provides coverage with a high
geometric resolution (5 m × 5 m on ground) over a narrower swath
width of 80 km. Six overlapping swathes cover the access range of
375 km. For each swath the antenna has to be configured to generate
a beam with fixed azimuth and elevation pointing. Appropriate eleva-
tion beam forming has to be applied for range ambiguity suppression.
Finally the Extra Wide-swath mode covers an ultra-wide swath
width of 400 km at medium resolution (20 m × 40 m on ground),
which is achieved through the implementation of a ScanSAR mode
with a fast beam scanning capability in elevation. As in the case of
the IW mode the Extra Wide-swath mode is implemented with a pro-
gressive azimuth scanning (TOPSAR operation).
The three imaging modes are high bit rate (HBR) modes, while the
wave mode is a low bit rate (LBR) mode. The four operational modes
of Sentinel-1 are illustrated in Fig. 3.
4.2. Instrument architecture and key parameters
The CSAR instrument is a synthetic aperture radar operating in the
C-Band using horizontal and vertical polarisation for transmit and re-
ceive. A functional block diagram of the instrument is given in Fig. 4.
The CSAR instrument is composed of two major subsystems
● the SAR Electronics Subsystem (SES)
● the SAR Antenna Subsystem (SAS)
The radar signal is generated at base band by the chirp gener-
ator and up-converted to C-band within the SES. This signal is
distributed to the High Power Amplifiers inside the Electronic
Front End (EFE) Transmit/Receive Modules (TRMs) via the
beam-forming network of the SAS. Signal radiation and echo
reception is realised with the same antenna using slotted
 R. Torres et al. / Remote Sensing of Environment 120 (2012) 9–24 13

Table 3
Main characteristics of the Sentinel-1 nominal measurement modes.

Parameter Interferometric Wide-swath mode (IW) Wave mode (WV) Strip Map mode (SM) Extra Wide-swath mode (EW)

Polarisation Dual (HH + HV, VV + VH) Single (HH, VV) Dual (HH + HV, VV + VH) Dual (HH + HV, VV + VH)
Access (incidence angles) 31°–46° 23°–37° (mid incidence angle) 20°–47° 20°–47°
Azimuth resolution b20 m b5 m b5 m b40 m
Ground range resolution b5 m b5 m b5 m b20 m
Azimuth and range looks Single Single Single Single
Swath >250 km Vignette 20 × 20 km >80 km >410 km
Maximum NESZ −22 dB −22 dB −22 dB − 22 dB
Radiometric stability 0.5 dB (3σ) 0.5 dB (3σ) 0.5 dB (3σ) 0.5 dB (3σ)
Radiometric accuracy 1 dB (3σ) 1 dB (3σ) 1 dB (3σ) 1 dB (3σ)
Phase error 5° 5° 5° 5°

waveguide radiators. In receive, the echo signal is amplified by
the low noise amplifiers inside the EFE TRMs and summed up
using the same network as for transmit signal distribution. After
filtering and down conversion to base band inside the SES, the
echo signal is digitised and formatted for recording.
Table 4 provides a brief overview on the instrument key
parameters.
The key design aspects of the CSAR Payload can be summarised as
follows:
● Active phased array antenna providing fast scanning in elevation
(to cover the large range of incidence angles and to support Scan-
SAR operation) and in azimuth (to allow use of TOPSAR technique
and meet the image quality requirements)
● Dual channel Transmit and Receive Modules and H/V-polarised
pairs of slotted waveguides (to meet the polarisation
requirements)
● Internal Calibration scheme, where transmit signals are routed
into the receiver to allow monitoring of amplitude/phase varia-
tions for later correction in the processing, leading to high radio-
metric stability.
● Metallised Carbon Fibre Reinforced Plastic radiating waveguides
to ensure good radiometric stability outside the internal calibra-
tion scheme (Fig. 5).
● Digital Chirp Generator and selectable receive filter bandwidths to
allow efficient use of on board storage considering the ground
range resolution dependence on incidence angle
● Flexible Dynamic Block Adaptive Quantisation to allow efficient
use of on-board storage and minimise downlink times with negli-
gible impact on image noise.
The up-conversion and amplification module of Sentinel-1 is given
in Fig. 6, while a three dimensional view of the SAS Tile, which is the
elementary building block of the SAR Antenna Subsystem, is given in
Fig. 7.
Important to mention is also the roll steering mode of the space-
craft. This continuous roll manoeuvre around orbit (similar to yaw
steering in azimuth) compensates for the altitude variation such
that it allows usage of a minimal number of different pulse repetition
frequencies (PRFs) and sample window lengths (SWLs) around the
orbit. Only a single PRF and SWL are necessary per swath/sub-swath
around orbit, with the exception of swath 5 of the Strip Map mode
where two different PRFs have to be used. In addition, the update
rate of the sampling window position around orbit is minimised
(b1/2.5 min), leading to significant simplification of instrument oper-
ations. Since the instrument can work with a single fixed beam for
each swath/sub-swath over the complete orbit, the number of eleva-
tion beams is minimised. The roll steering rate is fixed at 1.6°/27 km

Fig. 3. Sentinel-1 operational modes.

14 R. Torres et al. / Remote Sensing of Environment 120 (2012) 9–24

Fig. 4. CSAR instrument functional block diagram.

altitude variation. The roll applied to the sensor attitude depends lin- CSAR instrument activities and X-Band downlink operations, on the
early on altitude and varies within the interval − 0.8° (minimum sen- basis of a systematic pre-defined observation scenario.
sor altitude) to 0.8° (maximum sensor altitude).
5.1.1. Data reception and ingestion
CSAR real-time sensed data as well as data played back from the
5. Sentinel-1 PDGS and CSAR processor on-board PDHT are downlinked directly to ground or via the Europe-
an Data Relay Satellite (EDRS), received, down-converted, de-
5.1. PDGS main functions modulated and transferred to the Processing facilities for systematic
generation and archiving of Level-0 and Level-1/2 data products.
The Sentinel-1 Payload Data Ground Segment (PDGS) provides re- This function also includes frame synchronisation and reconstruction
ception of CSAR instrument data, systematic processing, archiving of the CSAR data (Instrument Source Packets, ISP) in a Computer
and dissemination of operational SAR data for the GMES services, Compatible Format (CCF).
and carries out various planning and monitoring tasks including
5.1.2. Data processing
Payload Level-0 data are processed to produce Level-1 and Level-2
Table 4 products applying all the necessary processing algorithms and for-
Instrument key parameters. matting techniques. The desired products are produced systematical-
ly or on request either in near real time (i.e. within 3 h after sensing)
Parameter Value
for time-critical data and within 24 h from sensing for all acquired
Centre frequency 5.405 GHz
data. Processing management capabilities and re-processing capabil-
Bandwidth 0 … 100 MHz (depending on
the mode and sub swath) ities are provided as well as the possibility to re-assemble “partial”
Polarisation HH + HV, VV + VH
Antenna size 12.3 m × 0.821 m
RF Peak Power (sum of all 4368 W
TRMs, at TRM o/p)
Pulse width 5–100 μs (programmable)
Transmit duty cycle
Max 12%
Strip map 8.5%
Interferometric Wide-swath 9%
Extra Wide-swath 5%
Wave 0.8%
Receiver noise figure at TRM input 3.2 dB
Pulse repetition frequency 1000–3000 Hz (programmable)
ADC sampling frequency 300 MHz (real sampling) (Digital
down-sampling after A/D conversion)
Sampling 10 bits
Data compression Selectable according to FDBAQ, see
Attema et al. (2010).
Instrument operation Up to 25 min per orbit continuously
in any of the imaging modes and for
the rest of the orbit in Wave mode
Instrument mass 945 kg
CSAR DC power 3870 W (Interferometric Wide-swath Mode,
two polarisations)
CSAR data storage capacity 1410 Gb (End-of-Life)
Effective CSAR downlink data rate Two channels of 260 Mbps each
Fig. 5. SAS tile, front view with H/V-polarised pairs of slotted waveguides.
 R. Torres et al. / Remote Sensing of Environment 120 (2012) 9–24
Fig. 6. Up-conversion and amplification module of Sentinel-1.
Level-0 products (i.e. covering only part of an acquisition segment or
part of a dual-polarisation acquisition segment) resulting from frag-
mented downlinks over different stations (or over the same station
in different passes).
5.1.3. Archiving
Archiving and long-term preservation of data is ensured for all
Level-0 data and for a set of configurable systematic higher-level
products. In addition a configurable mid-term (months-years) stor-
age capability is provided for the products systematically generated
to fulfil GMES Service needs. This function includes all operations to
be put in place to store these data and ensure their integrity according
to the applicable requirements.
5.1.4. Data dissemination and exchange
GMES Sentinel-1 products are provided to the user typically by
electronic server access (e.g. on-line data provision) and data is ex-
changed within and amongst the PDGS physical Centre(s) and
Station(s). Near real time dissemination is allocated to the receiving
Station(s) and less time-critical data dissemination is allocated to
the Assembly, Processing and Archiving Centre(s).
5.1.5. Mission planning
Detailed payload sensing and downlink schedules are defined and
exchanged with the GMES Sentinel-1 Flight Operations Segment
(FOS). SAR instrument parameters tables required to configure the
instrument operations are also transferred to the FOS for uplink to
the spacecraft. Satellite contact schedules are transmitted to the re-
ceiving ground stations.
Fig. 7. Three-dimensional view of the SAS tile layout.
15
5.1.6. Mission access and user interface
Several different capabilities are provided for permitting and sup-
porting successful handling of submitted user queries. These include
management of user access, user information management, docu-
mentation and Help Desk.
5.1.7. Mission performance assessment and data quality control
Capabilities are provided for mission performance assessment and
data quality control, ensuring that data quality and instrument per-
formance meet the expected requirements. These include verifica-
tion, calibration and validation.
5.1.8. Monitoring and control
Tools are provided to ensure that all available resources (i.e. hard-
ware, software and network) required to operating the PDGS accord-
ing to its functional and performance requirements are operating
nominally.
5.2. Sentinel-1 CSAR Level-1 and Level-2 data products
The Sentinel-1 CSAR Instrument Processing Facility (IPF) can gen-
erate the following Level-1 (L1b) products from the four acquisition
modes:
● Slant Range, Single-Look Complex (SLC)
● Ground Range, Multi-Look, Detected (GRD)
● Browse (BRW).
The products are further classified according to their resolution
into:
● Full Resolution (FR) (SM mode)
● High Resolution (HR) (SM, IW and EW mode)
● Medium Resolution (MR) (SM, IW, EW and WV modes).
Table 5 gives an overview of the characteristics of the GRD
products.
The IPF supports the processing of dual-polarisation data. The fol-
lowing algorithms are implemented in the Sentinel-1 IPF are:
● Pre-Processing algorithms
● Doppler Centroid Estimation algorithms
● SLC Processing algorithms
● GRD Processing algorithms.
The Sentinel-1 CSAR IPF will also be able to generate Level-2 ocean
products, both from wave mode data and from high bit rate mode
data (SM, IW and EW modes). Level-2 ocean products include infor-
mation on ocean swell spectra, ocean wind and currents (in the
form of radial surface velocity). The exact content of the Level-2
ocean products depends on the specific Sentinel-1 operational mode.
16 R. Torres et al. / Remote Sensing of Environment 120 (2012) 9–24

Table 5
Sentinel-1 CSAR L1 ground range detected products characteristics.

SM IW EW

S1 S2 S3 S4 S5 S6

Swath width (km) 80 80 80 80 80 80 250 400

GRD FR Ground range resolution (m) 8.1 8.4 8.8 9.0 9.2 9.2
Azimuth resolution (m) 8.1 8.7 8.8 8.9 8.9 9.1
Total number of looks 4.0 4.0 4.0 4.0 4.0 4.0
Pixel spacing (m) 4.0 4.0 4.0 4.0 4.0 4.0
Size MByte/5 min of acquisition 19,495 19,495 19,495 19,495 19,495 19,495
GRD HR Ground range resolution (m) 21.4 22.2 23.1 23.7 24.2 24.4 20.3 49.3
Azimuth resolution (m) 21.3 23.0 23.2 23.6 23.4 24.0 21.8 50.4
Total number of looks 36.0 36.0 36.0 36.0 36.0 36.0 5.0 3.0
Pixel spacing (m) 10.0 10.0 10.0 10.0 10.0 10.0 10.0 25.0
Size MByte/5 min of acquisition 3143 3143 3143 3143 3143 3143 9744 2499
GRD MR Ground range resolution (m) 78.0 80.7 84.1 86.4 87.9 88.7 87.8 91.1
Azimuth resolution (m) 77.3 83.8 84.4 85.7 85.0 87.4 89.2 88.3
Total number of looks 484.0 484.0 484.0 484.0 484.0 484.0 110.0 12.0
pixel spacing (m) 40.0 40.0 40.0 40.0 40.0 40.0 40.0 40.0
Size MByte/5 min of acquisition 128 128 128 128 128 128 392 629

5.3. TOPSAR specific processing
The Sentinel-1 IW and EW modes acquire wide-swath data (com-
posed of 3 and 5 sub-swaths respectively) using the TOPSAR imaging
technique. This requires TOPSAR-specific algorithms to deal with the
antenna steering rate and the Doppler Centroid (DC) rate due to the
steering. The required azimuth pre and post-processing of the data
include de-ramping of the data prior to base band DC estimation, az-
imuth ambiguity estimation and GRD azimuth processing.
The Sentinel-1 IW and EW SLC products contain one image per
sub-swath and one per polarisation channel, for a total of 3 (single
polarisation) or 6 (dual polarisation) images for IW data and 5 (single
polarisation) or 10 (dual polarisation) images for EW data.
Each sub-swath image consists of a series of bursts, where each
burst has been processed as a separate SLC image. The individually fo-
cused complex burst images are included, in azimuth-time order, into
a single sub-swath image, with black-fill demarcation in between. Due
to the one natural azimuth look inherent in the data, the imaged ground
area of adjacent bursts will only marginally overlap in azimuth — just
enough to provide contiguous coverage of the ground (Fig. 8). The im-
ages for all bursts in all sub-swaths of an IW and EW SLC products are
re-sampled to a common pixel spacing grid in range and azimuth.
Burst synchronisation is ensured for both IW and EW products. The pro-
cessing is phase preserving.
5.4. Dual-polarisation processing
The processing steps required for the dual-polarisation data pro-
cessing are essentially the same as those required for the processing
of single-polarisation data, and generally speaking each polarisation
is processed in the same manner. However, there are a few additional
considerations when processing dual-polarisation data:
1. A separate image is generated for each polarisation, and
polarisation-specific correction and calibration factors will be ap-
plied during the processing of each polarisation (for example, cal-
ibration pulses, antenna patterns, etc.).
2. The same estimated Doppler Centroid (DC) must be used for pro-
cessing both polarisations, so that the two resulting images are
accurately co-registered. Therefore the DC estimated from the
co-polarisation channel will be used, for DC estimation.
5.5. The slicing concept
As a continuation of ENVISAT ASAR services Sentinel-1 is required to
give users systematic access to long product strips similar to the
currently disseminated ASAR Wide Swath Mode and Image Mode prod-
ucts today systematically. The product size for L1 imaging modes would
however be difficult to manage at the user side (Table 6).
Therefore the level-1 image mode products are segmented in
“slices” of defined length along track, optimised per mode and product
type. The level-1 slices cover a sub-set of the data take in along-track di-
rection and the complete data take area in the across track direction.
● Slices are referred to the start of each acquisition segment.
● Slices are in the nominal product type projection (slant-range for
SLC, ground range for GRD).
● Slices are stand-alone products and can be handled separately in
terms of archiving and dissemination.
● Slices are seamlessly “concatenable” into a continuous product or
“strip” covering up to the complete data take.
● Slice concatenation may be performed before dissemination by
the PDGS or after dissemination by the user.
6. Sentinel-1 operation and observation capabilities
6.1. Mission operation concept
The Sentinel-1 mission is designed to provide ‘guaranteed data
services’ for which liability can be accepted commensurate with the
user needs following the GMES public institutional user model. This
is common practice for meteorological data provision. Most acquisi-
tions can be pre-planned and the routine operations are normally
not interrupted. Therefore no specific acquisition requests are re-
quired and data access is mainly by subscription supporting multi-
temporal observations and near real time product generation. This
is particularly important for responding to emergency requests to
support disaster management in crisis situations as no time is wasted
for satellite commanding and conflict resolution. However, if special
operational modes are required the system is designed to respond
at the expense of slightly increased response time. All data is system-
atically processed and available within 24 h. Data can be also re-
trieved off-line from the archive.
6.2. Spacecraft operations concept
The following list summarises the main characteristics of the
Sentinel-1 mission that determine the Operations Concept:
● The Sentinel-1 spacecraft is designed such that its on-board re-
sources allow storing the complete instrument schedule covering
a maximum time interval of 4 days.
 R. Torres et al. / Remote Sensing of Environment 120 (2012) 9–24
IW SLC
Fig. 8. IW SLC de-burst azimuth overlap.
● In terms of on-board autonomy, the satellite can operate nominal-
ly for at least 96 h without any ground intervention.
● The resulting routine Operations Concept is characterised by the
following main aspects:
○ With the exception of emergency response acquisitions the
spacecraft will operate according to a pre-planned schedule of
commands (a predefined mission timeline similar for every
cycle, with few updates during the year to cope with seasonal
variations e.g. for sea–ice and iceberg monitoring) uploaded
from ground.
○ The ground operator will acquire housekeeping data via the
TT&C ground station network. The downlinked housekeeping
data from the platform and the payload will be processed to
verify the spacecraft health;
○ Real time transmitted SAR data will be received by X-band sta-
tions and via the European Data Relay Satellite System (EDRS);
○ The recorded SAR data will be downlinked every orbit using a
network of X-band stations and via EDRS. The data dump will
be managed via a predefined sequence of scheduled telecom-
mands stored on-board.
○ Updates to the mission timeline can be generated regularly to
incorporate for instance changes to instrument settings as
well as non-nominal specific requests for instrument sensing
or data download.
6.3. Flight operations segment concept
The FOS provides the capability to monitor and control the space-
craft during all mission phases. This includes the facilities for the gen-
eration and uplink of the mission schedules, and for the systematic
reception, processing, archiving and analysis of the acquired satellite
Table 6
Sentinel-1 data volume examples.
Data volume in GByte for dual polarisation
Segment length [min]
SM IW
Length 2 10 2 10
SLC 36 182 30 150
GRD HR 3 13 8 39
17
IW SLC de-burst
housekeeping telemetry. Moreover, the FOS includes a Flight Dynam-
ics System facility responsible for orbit determination and prediction,
and generation of attitude and orbit control telecommands. Other FOS
functions comprise the monitoring and control facilities for the Track-
ing, Telemetry and Telecommand ground station network.
The FOS provides spacecraft monitoring and control services to
support the mission exploitation according to specifications and re-
quirements previously introduced. More specifically, the FOS imple-
mentation shall:
● offer flexibility and scalability capabilities to integrate comple-
mentary functions for evolving requirements (e.g. extension to
emergency services)
● offer efficient data exchange between FOS and PDGS
● for each Sentinel mission, support the progressive build-up and
operations management of the satellite constellation in particular
by provision of orbit determination and maintenance services
● guarantee a high degree of reliability and availability in order to
ensure the spacecraft safety under nominal and failure scenarios.
6.4. Sentinel-1 observation scenario
Following Sentinel-1 mission requirements, as outlined in
Section 2, into account a Sentinel-1 observation scenario has been de-
veloped based on systematic global coverage to secure the reference
image archive and with sufficient resources to accommodate specific
emergency acquisitions. The main output of the Sentinel-1 mission is
a dependable SAR image data flow and following a systematic, pre-
programmed and conflict-free operational scenario with time-
critical data dissemination in near real time and all data available
within 24 h. Special efforts are made to support rapid mapping for
emergency response services. Global average response time is largely
defined by orbit configuration and swath width and amounts in the
worst case to about 5 days for a single satellite and 2.5 days for the 2-
satellite constellation. The actual performance depends on the location
of the satellite(s) with respect to the emergency site and on latitude and
Slice length
is worst at the equator and best at high latitude.
[sec]
In its main operational IW mode operating up to 25 min per orbit,
SM IW
it potentially provides complete global coverage of all relevant land
25 25 surfaces, sea ice, coastal zones and North Atlantic shipping routes at
8 6 least once per 12 days for each of the two satellites. For high priority
1 2
areas within Europe, Canada and the Northern Atlantic coverage will
18 R. Torres et al. / Remote Sensing of Environment 120 (2012) 9–24

Fig. 9. Sentinel-1 coverage after 12 days for one satellite with the IW mode.

be more frequent ranging from 4 to 2 days per satellite depending on
latitude (see Fig. 9).
Over the open ocean, except the North Atlantic, data are normally
collected in the wave mode operating all the time. This provides im-
ages of 20 by 20 km separated by 100 km for use in data assimilation
in global wave models.
In order to satisfy the service requirements for Sentinel-1 geo-
graphic coverage and temporal re-visit, 2 satellites (Sentinel-1A and
1B) are required.
The role of GMES Contributing Missions (GCMs) is essential, par-
ticularly in the time frame until all B-units are deployed. The GCMs
are important in completing the observations provided by the Senti-
nels for GMES in time and space. As such, the GCMs and the Sentinels
provide the full observational capabilities, necessary, to sustain the
GMES services.
A long-term continuity of observations from the GMES Space
Component beyond the period of the Sentinel A- and B-units is fore-
seen and will be necessary to sustain these services in the future.

6.5. Long-term continuity of observations Table 8

Sentinel-1 user products — processing concept and timeliness.

The capacity of the GMES Space Component (GSC) is being built Processing Product Timeliness Comments
up progressively, starting with the launch of the A-units of Sentinels concept type (availability
after sensing)
1–3 in the 2013–2014 timeframe. This will be augmented by the
launch of B-units (2015–2017). Systematic SAR L0 and b3 h For regional areas defined through
Global SAR L1 HLOP.
GRD b24 h All acquired data.
SAR L2 b3 h For WV mode data acquired over the
OCN ocean and SM, IW and EW mode data
Table 7 over regional areas defined through
Sentinel-1 User Products from Core and Collaborative PDGS. HLOP.
Systematic SAR L1 SLC b3 h Data available from the core PDGS for
Core PDGS (committed Collaborative PDGS
regional regional areas defined through HLOP.
quality and timeliness) (non-exhaustive list)
Faster availability from sensing relies on
L0 SAR L0 see Table 8 collaborative PDGS
L1 SAR L1 SLC see Table 8 b24 h Data available from the core PDGS for
SAR L1 GRD SAR L1 Orthorectified regional areas defined through HLOP.
L2 and higher SAR L2 OCN (wind, wave and currents) Soil Moisture Systematic SAR L0 b10 min Only for data over ocean acquired in
Swell propagation local direct downlink over a S-1 Core Ground
PS Cal products (*) Station. For areas outside the visibility of
Sigma0 mosaic (*) S-1 core stations, 10 min availability
from sensing relies on the collaborative
Products with a (*) are generated by the Core PDGS for calibration purposes only. See
PDGS.
also Table 8 for information about the processing concept and timeliness.
 R. Torres et al. / Remote Sensing of Environment 120 (2012) 9–24 19

Radiometric resolution in dB of a 7.1. Processing concepts and timeliness

150 x 150m spatial resolution product
3.0 The processing concept for the Sentinel-1 Core PDGS User prod-
ucts is summarised in Table 8. All Sentinel-1 SAR data acquired is sys-
2.5 tematically processed to a predefined product type and available
within a defined timeliness. Systematically generated products can
2.0 be classified in:

● Global products
1.5 Global products are those systematically generated for all acquired
data. They include Level-0, detected Level-1 and Level-2 ocean
1.0 products. These products are made available within 3 h from ob-
servation over near real time areas defined in the High Level Oper-
0.5 ations Plan (HLOP) and in any case, within 24 h from observation.
These products are made available on-line, accessed through sub-
scription and archived for long-term access.
0.0
● Regional products
-30 -25 -20 -15 -10 -5 0
Regional products are those systematically generated for a sub-set
Radar Backscatter Coefficient σ0 [dB]
of the total acquired data, over well-defined regional areas in the
Sentinel-1 Sentinel-1 Stripmap 5*5
Interferometric Wide 1 look 80km
HLOP. These products are made available within 3 h from observa-
Swath 20*5 1 look tion over specific near real time areas in line with the HLOP and in
250 km
any case, within 24 h from observation. Regional products include
Sentinel-1 Extra-wide ASAR Wide Swath 11.5 Level-1 SLC products. These products are made available on-line,
Swath 40*20 1 look looks 400km
400 km accessed through subscription and archived for long-term access
like the global products
Fig. 10. Radiometric resolution in dB as a function of the radar backscatter coefficient. ● Local products
For critical GMES services, notably maritime surveillance, requir-
ing fast delivery, Level-0 products will be made available within
10 min from observation over local areas defined in the HLOP.
7. Sentinel-1 user products
This timeliness is foreseen for data acquired over European waters
and only possible when Sentinel-1 is inside the coverage zone of
Following the GMES Space Component data, Sentinel-1 data ac-
one of the Core ground stations. Local products are systematically
cess is based on systematic and free on-line data dissemination at
generated, made available on-line, accessed through subscription
no charge and without the need for an ordering process except in
and archived for long-term access like the global products.
cases of support to emergency situations.
The Sentinel-1 mission will be able to provide a variety of image All systematically generated products (as shown in Table 8) are
and/or information products to the users. The Sentinel-1 Core PDGS archived and available for download from archive within 24 h from
will generate some of these products operationally, the Sentinel-1 observation. Products different from those available in the archive
Core PDGS or the Sentinel-1 Collaborative PDGS may support some over a given area can be generated from the archived Level-0 data.
of them and the GMES Core and Downstream services are expected On-request generation of new products, different from those sys-
to generate the final information products (e.g. ice charts, forest tematically generated, is supported but restricted to a limited number
maps, ground-surface deformation maps). The Sentinel-1 products of authorised users (currently restricted to GMES emergency and se-
operationally generated by the Core PDGS and available to users in- curity services).
clude SAR “raw” Level-0 data and “SAR processed” Level-1/2 data.
Level-1 data include two main categories: complex imagery (SLC) 8. Calibration and validation of Sentinel-1 products
for interferometric applications and Ground Range Detected Geo-
referenced imagery (GRD) for most intensity based applications. For The Synthetic Aperture Radar measures a series of radar echoes
the SLC and GDR categories image stacks (time series of repeat cover- that are processed within the ground segment to generate SAR data
age) can be provided. products. Main contents of these data products are the normalised
As a baseline Sentinel-1 provides ocean Level-2 products (wind, radar cross-section and the phase of the individual pixels. Both quan-
wave and currents). tities have to be provided with specified stability and accuracy. This is
The Sentinel-1 user products generated by the Core PDGS are ensured through calibration and validation tasks executed through-
summarised in Table 7. The table introduces as well the concept of out the mission lifetime.
products generated by the Collaborative PDGS, with some examples The SAR images are generated on the basis of the individual radar
of candidate collaborative products. echoes that will contain instrument noise. Transmit power, receiver

Table 9
C-band SAR comparison.

Mission/mode Azimuth resolution (m) Ground range resolution (m) No. of looks (N) NESZ (dB) Swath width (km) PI Abs. PI relative to ERS

Sentinel-1/Interferometric 20 5 1 − 22 250 2300 4.4

Wide-swath
Sentinel-1/Strip Map 5 5 1 − 22 80 1472 2.8
Sentinel-1/Extra Wide-swath 40 20 1 − 22 400 1301 2.5
Envisat/ASAR Wide Swath 150 150 11.5 − 23.5 400 937 1.8
Envisat/Image 5 20 1 − 22 80 736 1.4
ERS-2/Image 5 20 1 − 20 80 526 1.0
20 R. Torres et al. / Remote Sensing of Environment 120 (2012) 9–24

Table 10
Intensity contrasts between major ice classes in Sentinel-1 IW images and classification accuracy.

Location Pairs of dominant ice types IC [dB] and A [%] for types 1 and 2, IC [dB] and A [%] for types 1 and 2, IC [dB] and A [%] for types 1 and 2,
VV-polarisation HH-polarisation cross-polarisation

Storfjord 1: young level ice 1.46 (62.9) 1.67 (64.7) 1.02 (59.1)
2: broken + brash ice
1: new ice 3.58 (78.9) 7.93 (96.0) 0.21 (51.9)
2: young level ice
Barents Sea 1: young/first-year level ice 2.96 (74.7) 3.26 (76.8) 1.16 (60.3)
2: brash + rubble ice
1: new ice 2.92 (74.4) 2.40 (70.5) 0.02 (50.2)
2: young/first-year level ice
Fram Strait 1: first-year level ice 7.28 (94.7) 6.42 (92.3) 11.47 (99.3)
2: ridges + brash ice
1: new + young ice 5.62 (89.5) 6.37 (92.2) 0.57 (55.1)
2: first-year level ice

IC = intensity contrast, A = classification accuracy (in parentheses).

gain and to a lesser extent antenna gain are subject to variations in
time due mainly to temperature changes but also due to ageing or
other effects. One main task of the internal calibration is to provide
corrections for changes of the transmit power and the electronics
gain. The Sentinel-1 Antenna Model is a software tool that can accu-
rately estimate the Sentinel-1 SAR antenna radiation patterns based
on the instrument configuration, pre-launch measurements and re-
sults derived from in-orbit measurements. The antenna model nomi-
nally covers the antenna gain and it is a task of the internal calibration
to ensure the validity of the antenna model. Internal calibration also
covers the signal phase. The overall phase of the echo signal depends
on two major elements, the measurement geometry and on instru-
ment internal phase stability. As the hardware can in general not pro-
vide the required phase stability, it is a task of the internal calibration
scheme to cover the internal phase variations by adequate measure-
ments. All internal calibration measurements, either for gain or for
phase, are used within the ground processing to correct the data
products in order to achieve the required stability.
Main task of external calibration is to derive the calibration con-
stant by measurements of targets with exactly known backscatter
coefficients. This is necessary as in general it will not be possible to
know all parameters with sufficient accuracy prior to the in flight
measurements.
The key outputs of the calibration and validation activities are
● Deriving and maintaining the system calibration parameters;
● Quality Assessment of the Level-1, Level-2, and Level-3 products,
in the form of reports, diagrams, statistics;
● Feedback into the Sentinel-1 Space and/or Ground Segment by
tuning the instrument, the Level-1 or the Level-2 and Level-3 pro-
cessors e.g. through calibration parameter updates.
9. Sentinel-1 product and application performance
9.1. Overview
Data products from Sentinel-1 consist of imagery representing the
radar echo from the Earth surface. Basic calibrated data are provided
in the form of both complex (In-phase and Quadrature components)
and intensity images. These are referred to as Level-1. Level-1

Fig. 11. Example of a sea–ice product.

 R. Torres et al. / Remote Sensing of Environment 120 (2012) 9–24 21

Table 11
Requirements for Snow Covered Area (SCA) products.

Level Measurement Spatial Temporal Driver

accuracy resolution resolution

Threshold 10% 0.5 km 1 daya Hydromet

Target 5% 0.1 km 12 ha
a
Temporal requirements are relaxed for climate research (5–7 days) and for hydrology
of mountain areas (3–5 days).

performance is defined by the parameter values listed in Table 3.

Level-2 data products include extracted information from the
Level-1 imagery in the form of estimated values of the geophysical vari-
ables listed in Davidson et al. (2009). This type of Level-2 product Fig. 12. Snow Cover Algorithm (SCA) classification result. White colour represents wet
requires a retrieval algorithm and the Level-2 performance depends snow, grey represents dry snow, dark red represents bare-soil and red represents areas
therefore on both the Sentinel-1 system performance and the retrieval covered by radar shadow. Further, masked out lakes and forested areas are cyan and
green, respectively.
algorithm. In the following subsections the link between the instrument
specifications in Table 3 and Level-1 and Level-2 image quality are
briefly discussed.
expression applies if the noise bias is derived from a relatively large num-
9.2. System noise ber of system noise estimates.

The Sentinel-1 specification of − 22 dB for noise equivalent σ st 1 þ SNR

1
¼ pffiffiffiffi ð1Þ
sigma zero (NESZ) encompasses all sources of system noise includ- I0 L
ing thermal noise and quantisation noise sources. The impact of sys-
tem noise on Level-1 quality will depend on the actual signal where,
strength of the backscatter, and will generally broaden the distribu-
tions of amplitude and phase, decreasing the accuracy of the Level-2 σst is the standard deviation of the image intensity in linear
product. power
Phase noise effects interferometric performance and in particular the I0 is the mean image intensity in linear power
error budget associated with subsidence monitoring using Permanent SNR is the signal-to-thermal noise ratio
Scatterers. L is the effective number of looks and is equal to (mean in-
tensity) 2/variance
9.3. Radiometric accuracy and stability
The radiometric resolution has been calculated using the equa-
Radiometric accuracy and stability are specified for Sentinel-1 for tions given above for a radar backscatter coefficient σ0 varying from
all measurement modes to be within 1 dB (3σ) and 0.5 dB (3σ) re- −26 dB m 2/m 2 up to 0 dB m 2/m 2 for the different imaging modes
spectively. The 3σ specification is based on the confidence intervals of Sentinel-1 with a noise equivalent sigma zero (NESZ) of −22 dB.
of a Gaussian (normal) distribution and refers to the probability of For comparison the radiometric resolution of the Envisat ASAR Wide
the instrument exceeding the specified interval. Swath is also given. The results for a 150 × 150 m spatial resolution
The accuracy and stability numbers refer to measurement uncer- product are shown in Fig. 10.
tainty of the radar backscattering coefficient using the image intensity
excluding the effect of speckle and noise. These effects are defined by
the radiometric resolution that is only implicitly part of the system
specification and can be derived from the noise equivalent sigma
zero and the spatial resolution.

9.4. Radiometric resolution

Radiometric resolution refers to measurement uncertainty of the

radar backscattering coefficient using the image intensity arising
from speckle and noise for an otherwise perfect system. Radiometric
resolution is expressed as the standard deviation of the measurement
uncertainty normalised with respect to the mean.
For an imaging system with noise bias subtraction Eq. (1) (Ulaby et al.,
1982) gives the expression for radiometric resolution or accuracy. The

Table 12
Specifications for Snow Cover products derived from Sentinel-1 IW mode data.

Mode Pixel size of Azimuth Range Total number of looks

SCA product looks looks for SCA product
Fig. 13. Minimum detectable ship length for Sentinel-1 HH (blue) and Envisat ASAR HH
IW 100 × 100 m2 5 20 100
(green).
22 R. Torres et al. / Remote Sensing of Environment 120 (2012) 9–24
From Fig. 10 it is clear that the IW mode of Sentinel-1 combines its
excellent coverage with a high radiometric resolution capability.
9.5. ERS, ENVISAT and SENTINEL-1 performance indicator
In order to put the performance of the Sentinel-1 Level-1 data
products in perspective with a view to demonstrate the expected ap-
plication potential with reference to its predecessors a performance
indicator (PI) has been defined that takes not only account of sensi-
tivity, spatial resolution and radiometric resolution but also coverage
(and revisit) in terms of swath width. PI is derived by dividing the
swath width (in km) by the radiometric resolution of a 150 × 150 m
product with a backscatter level of −20 dB. A large swath width in
combination with a good radiometric resolution will yield a high
value for the PI. The results for ERS, ENVISAT ASAR and Sentinel-1
CSAR are summarised in Table 9 and underline the emphasis during
the Sentinel-1 design put on high quality multi-temporal data products.
9.6. System errors related to Level-2 retrieval performance
Retrieval algorithms for e.g. soil moisture and wind speed estima-
tion rely on the absolute level of the received radar signal while re-
trieval algorithms for land cover classification are in general based
on simple generic algorithms:
● application of a threshold to a single intensity channel;
● maximum likelihood as a discriminator between two surfaces;
● estimation of the ratio between two intensity channels (prior to
comparison with a threshold).
For a number of algorithms currently in use and expected to be
used for Sentinel-1 Level-2 product generation, the performances re-
lated to the measurement uncertainties of the Level-1 products have
been studied.
9.6.1. Sentinel-1 subsidence monitoring performance
The subsidence rate error budget is calculated by inserting the
total phase noise coming from the individual terms in the model as
derived in De Zan and Monti Guarnieri (2006).
The calculated subsidence rate (mm/year), taking into account the
system geometry, parameters and performances of Sentinel-1, as well
as other items affecting the estimate, e.g. temporal changes, atmo-
spheric disturbances, orbital tubes, S/C attitude and assuming linear
motion, and the standard deviation of the overall phase noise for
Sentinel-1. The resulting subsidence rate error is in the order of
1.3 mm/year.
9.6.2. Sentinel-1 sea–ice classification accuracy
Amongst potential users of Sentinel-1 SAR image products are
operational ice services that provide information on ice conditions
(extent, concentration, thickness, deformation) to ships and marine
platforms in ice-covered ocean regions. Sea–ice mapping includes
the visual or automatic separation of different types and structures
of ice. In Dierking (2009) it was shown that the radar intensity con-
trast between different ice types depends on the ice regime and on
environmental conditions. A list of mean intensity contrasts between
dominant ice classes has been derived for three different ice regimes
in the region of Svalbard. The notation “dominant class” indicates that
an operator who analyses a SAR image can clearly recognise the re-
spective ice type or feature and distinguish it from other classes.
Table 10 is based on Fig. 8 of Dierking (2009). The data are valid for
an incidence angle range of 30 to 45°. They are based on the simula-
tions of images with the characteristics of the Sentinel-1 Interfero-
metric Wide-swath mode, using a pixel size of 20 × 20 m, which
implies approximately four looks.
At like-polarisation, the classification accuracy using an image
with number of looks = 4 (20 m × 20 m) is moderate for the Storfjord
and the Barents Sea ice regime. With pixels of double size and a cor-
responding increase of the number of looks, the accuracy improves to
values >80%. The use of even larger pixel sizes, however, will have a
considerable impact on the perceptibility of narrow deformation
structures such as ice ridges and brash ice between ice floes because
of the related decrease of spatial resolution. It has to be taken into ac-
count that the effective radar intensity contrast of a structure relative
to the adjacent ice is decreased if the structure does not fill the entire
area of a given pixel. For the Fram Strait ice regime, the classification
accuracy is already ≥90% for 20 × 20 m pixels.
At cross-polarisation, the intensity contrast between the dominant
ice classes is low, with the exception of the pair level — deformed ice
in the Fram Strait ice regime. Hence, the classification accuracy for an
automated separation of ice types is not sufficient. Only if the neigh-
bouring pixels are averaged such that the number of looks is in-
creased to values close to L = 100, the accuracy would be >80%. But
in this case, the theoretical gain may be counter-balanced by the
loss of spatial resolution as explained above.
Increasing the coverage by using the EW mode would reduce the
classification accuracy to unacceptable low levels due to the decrease
in radiometric resolution of the EW mode compared to the IW mode.
Fig. 11 shows an example of a sea–ice product.
9.6.3. Sentinel-1 snow cover monitoring performance
C-band SAR can be applied for mapping the extent of snow area
(melting snow only). The requirements on spatial resolution and re-
peat observations of the Snow Covered Area (SCA) product are spec-
ified in the Integrated Global Observing Strategy (IGOS) Cryosphere
Theme Report (Table 11).
Spatial resolution of the SCA product is not a primary issue with
Sentinel-1, allowing for considerable multi-looking. The temporal re-
quirement of 1 day is challenging. This number refers to hydrometeorol-
ogy of lowlands, where snow cover may be quite volatile. Obviously, the
contemporaneous operation of 2 Sentinel-1 satellites will enable opera-
tional products with a temporal resolution of less than 5 days. For practi-
cal applications, the integration of SCA products in hydro-meteorological
models enables to bridge gaps in the temporal sequence.
It has to be pointed out that C-band SAR is not able to supply the
full SCA product, because it is not sensitive to dry snow. There are var-
ious possibilities to infer the extent of dry snow from other sources,
briefly addressed below. For hydrological applications, the timely
and accurate observation of melting snow areas, as possible by
means of Sentinel-1, is very important.
For the spatial resolution of the snow product the target requirement
of 5% measurement accuracy from IGOS for 100 m × 100 m products is
assumed (a 10% figure is assumed for the threshold requirement).
Only the speckle-related uncertainty for classification of wet snow is
taken into account. The Interferometric Wide-swath Mode of Sentinel-
1 has been used as the baseline for the SCA retrieval (Table 12).
The probability of error for the discrimination of two classes is 5%
for a 2 dB difference between the classes using the Sentinel-1 IW
mode. Using the average of several images for the reference and by
reducing the spatial resolution of the product the error can be re-
duced. An example of the Snow Cover Algorithm (SCA) classification
result is given in Fig. 12.
9.6.4. Sentinel-1 rice field mapping performance
Rice is a vital world food crop. In many countries, it forms the basis
of the economy and provides the staple food of the people. To control
and maintain a close balance between rice production and food
needs, an effective rice-monitoring programme is required at region-
al, national, and international scale.
For rice field mapping C-band SAR intensity data are particularly well
suited to rice monitoring, because of a) the all weather capability of the
SAR systems to acquire data over rice growing regions which are under
frequent cloud cover, b) the remarkable increase in backscatter intensity
 R. Torres et al. / Remote Sensing of Environment 120 (2012) 9–24 23

Table 14
Sentinel-1 Interferometric Wide-swath Mode Level-2 product performance.

S-1 Level-2 product Resolution Performance Units

Subsidence rate 5 × 20 m2 1.3 Rate uncertainty in

mm/year
Land cover (2 dB 30 × 30 m2 90.0% % correct classification
contrast)
2
Forest non-forest 30 × 30 m 74.3% % correct classification
classification 100 × 100 m2 100%
2
Soil moisture 100 × 100 m 3.9 (3σ) Moisture content
uncertainty in volume%
Flood mapping 30 × 30 m2 79.4% % correct classification
Snow cover 30 × 30 m2 75.0% % correct classification
classification
(3 dB contrast)
Ship detection HH 5 × 20 m2 25–34 (90%) Minimum ship length in
+ HV polarisation (Swath m, (probability of
dependent) detection)
Sea surface wind 100 × 100 m2 2.4 (3σ) Speed uncertainty in m/
speed s
Sea surface currents 5 Hz 30 Speed uncertainty in
cm/s

Fig. 14. Minimum detectable ship length for Sentinel-1 HV (blue) and Envisat ASAR HV
(green).
● The increased number of acquisitions by S-1 during the growing sea-
son enables to properly sample the growth cycle and to limit the im-
pact of geophysical noise (for example by identifying a date with no
throughout the growth cycle as the rice plants grow above the water wind where the water surface is very smooth). With one Sentinel-1
surface and interact with the incident radar signal. acquisition every six days, it is expected that the performance of the
Quantitative performance requirements in terms of the threshold multi-temporal algorithm drastically improves compared to what
or target accuracies required by individual applications are generally was obtained with ASAR. In addition, in regions where multiple agri-
not available. cultural practices are common (e.g. different sowing dates throughout
The traditional method for cultivating rice is flooding the fields the region), a higher temporal sampling will be very beneficial.
while, or after, setting the young seedlings. From a previous assess-
ment of the use of ENVISAT ASAR data for rice mapping in China, it The better range resolution of S-1 will allow for more efficient spa-
is shown that for Sentinel-1 the temporal behaviour of co-polar sig- tial multi-looking, which is important for this application that deals in
nals can be used as a classifier over areas where flooding practices some regions with very small paddy fields.
are used for rice crops.
Using Sentinel-1 for rice crops mapping the following conclusions
can be drawn: 9.6.5. Sentinel-1 ship detection performance
● There is a limited impact of the NESZ on the quality of the seg- The minimum detectable ship lengths for Sentinel-1 are shown in
mented maps based on the multi-temporal co-polar images. This Figs. 13 and 14. The fixed parameters used in each case are shown
was to be expected as the values of σ 0 for these natural surfaces across the title of each plot. The plots use the actual ground range res-
are relatively high, at least for co-polar. A sensitivity analysis olution and NESZ specific to each position in the swath. The plots are
shows that there is no much gain in performance when the NESZ colour coded by sensor and include Sentinel-1 (blue), and Envisat
is lowered below −22 dB/−23 dB. ASAR (green). The beams are plotted alternately with solid and
dashed lines for each sensor and the beam mode labels are placed
at the centre of each swath (some of the plots are quite busy). The
Table 13 curves for the HH case tend to be smooth and co-linear due to the
Sentinel-1 Strip-Map mode Level-2 product performance. clutter-limited imaging. The curves for the HV case follow the shape
of the noise floor due to the noise limited imaging.
S-1 Level-2 Resolution Performance Units
product By inspection the following can be noted:
Subsidence rate 10 × 10 m2 1.3 Rate uncertainty in
mm/year
1. The minimum detectable ship length for the groups of single beam
Land cover (2 dB 30 × 30 m 2
90.0% % correct classification modes decreases with increasing incidence angle, which illustrates
contrast) the importance of incidence angle in reducing the background
Forest Non-Forest 30 × 30 m2 98.3% % correct classification ocean clutter level.
Classification 100 × 100 m2 100%
2. The minimum detectable ship length decreases for modes with
Soil Moisture 100 × 100 m2 3.2 (3σ) Moisture content
uncertainty in volume% increasing (i.e., smaller) resolution.
2
Flood Mapping 30 × 30 m 93.0% % correct classification
Snow Cover 30 × 30 m2 80.0% % correct classification
Classification Table 15
(3 dB contrast) Sentinel-1 Wave Mode Level-2 product performance.
Ship Detection HH 5 × 5 m2 12–23 (90%) Minimum ship length in
S-1 Level-2 product Resolution Performance Units
+ HV (Swath dependent) m, (probability of
2
polarisation detection) Significant wave 20 × 20 km 0.3 Height uncertainty in m
Sea Surface Wind 100 × 100 m2 1.3 (3σ) Speed uncertainty in height
2
Speed m/s Dominant wavelength 20 × 20 km 40 Wavelength uncertainty in m
Sea Surface 5 Hz 30 Speed uncertainty in Dominant direction of 20 × 20 km2 12 Direction uncertainty in
Currents cm/s propagation degrees
24 R. Torres et al. / Remote Sensing of Environment 120 (2012) 9–24
Fig. 15. Multi-temporal ASAR image for land use.
3. For co-polarisation, the ship RCS is larger, but the clutter level is
higher, especially for smaller incidence angles. The co-polarisation
case is generally clutter limited.
4. For cross-polarisation the ship RCS is smaller, but the clutter is
below the noise floor; the cross-polarisation case is noise limited.
There are significant benefits to using cross-polarisation, especially
for acquisitions at smaller incidence angles. Furthermore, cross-
polarisation provides more uniform ship detection performance
across the image swath since the noise floor is somewhat indepen-
dent of incidence angle.
Considering Sentinel-1, there is a penalty in using EWS for its
broader swath due to its coarser resolution than IWS.
9.7. Sentinel-1 Level-2 product performance summary
A key element in the design of the Sentinel-1 mission is the rela-
tion between the technical implementation of the mission and the
accuracy of the geophysical information products that the mission is
required to generate and support for operational applications and
service provision. Analysis is required to demonstrate the compliance
between the technical choices of the mission and system and user
requirements. It also provides data users with quantitative estimates
of the system contribution to the overall performance (i.e. accuracy)
of geophysical products along with the methodological framework
used to estimate such system contributions. The complete analysis
is given in Davidson et al. (2009). The expected Sentinel-1 Level-2
product performances are summarised in Tables 13 to 15 for a given
spatial resolution adapted to a product type. It is noted that the
results given are based on ideal retrieval cases, with no other errors
than the ones related to the Sentinel-1 performance. The retrieval
algorithm is assumed to be error free.
10. Conclusion
The Sentinel-1 constellation is a completely new approach to SAR
mission design by ESA in direct response to the operational needs for
SAR data expressed under the EU-ESA GMES programme. The mission
ensures continuity of C-band SAR data and builds on ESA's heritage
and experience with the ERS and Envisat instruments (Fig. 15),
notably maintaining key characteristics such as stability and accurate
well-calibrated data products. At the same time, mission parameters
have been vastly improved to meet major user requirements collect-
ed and analysed through EU Fast Track and ESA GMES Service Ele-
ment activities, especially in areas such as reliability, revisit time,
geographical coverage and rapid data dissemination. As a result, the
Sentinel-1 pair is expected to provide near daily coverage over Europe
and Canada, independent of weather with delivery of radar data within
an hour of acquisition — vast improvements over the existing SAR
systems.
In addition to responding directly to the current needs of GMES,
the design of the Sentinel-1 mission is expected to enable the devel-
opment of new applications and meet the evolving needs of GMES,
such as in the area of climate change and associated monitoring
Acknowledgement
The content and figures in this paper are based on the inputs from
the ESA Sentinel-1 team members, the industrial consortium lead by
Thales Alenia Space Italy as Prime Contractor and the members of
former The SAR Advisory Group (TSAG).
References
Attema, E. (2005). GMES Sentinel-1 mission requirements document ES-RS-ESA-SY-0007,
issue 4, revision 4, 29/9/2005.
Attema, E., Cafforio, C., Gottwald, M., Guccione, P., Monti Guarnieri, A., Rocca, F., et al.
(2010, Oct.). Flexible dynamic block adaptive quantization for Sentinel-1 SAR missions.
IEEE Geoscience and Remote Sensing Letters, 7(4), 766–770.
Davidson, M., Attema, E., Snoeij, P., Rommen, B., & Floury, N. (2009). Sentinel-1 system
error budget. S1-TN-ESA-SY-0046.
De Zan, F., & Monti Guarnieri, A. M. (2006, Sept.). TOPSAR: Terrain observation by
progressive scans. IEEE Transactions on Geoscience and Remote Sensing, 44(9),
2352–2360.
Dierking, W. (2009). Mapping of different sea ice regimes using images from Sentinel-1
and ALOS synthetic aperture radar. IEEE Transactions on Geoscience and Remote
Sensing, 48(3), 1045–1058 March 2010.
Flett, D., Crevier, Y., & Girard, R. (2009, July 12–17). The RADARSAT constellation mission:
Meeting the government of Canada's needs and requirements. Proceedings of IGARSS
(IEEE International Geoscience and Remote Sensing Symposium) 2009 Cape Town,
South Africa.
Ulaby, F. T., Moore, R. K., & Fung, A. K. (1982). Microwave remote sensing, Vol. 2, Reading,
MA: Addison-Wesley.