Methods Altimetry 1

Since the launch of ESA’s CryoSat mission in April 2010 we have been involved in the production and distribution of ocean altimeter data acquired by the SIRAL altimeter system. We reprocessed all level-1 LRM and SAR data and currently distribute level-2 ocean geophysical data records via our Radar Altimeter Database System, RADS. In this paper we will compare our products with different versions of the ESA baseline products as well as the recently established intermediate (IOP) and final ocean products (GOP) available from ESA. The quality of the geophysical data records is similar to other altimeters such as Jason-2 and AltiKa. The design of the SIRAL altimeter is fundamentally different compared to existing ocean altimeters with benefits to observing waveform data. SIRAL is the precursor for a new generation of altimeter systems that are implemented on Sentinel-3 and will be on Sentinel-6.

This paper addresses the impact of sea state condition on the radar altimeter waveforms, for both LRM and SAR modes. To carry out this study, several months of Cryosat-2 data obtained under different wave conditions are analysed and compared to reliable co-located data making possible the identification of potential altimetric errors.
To our knowledge, no additional sea state signatures in LRM altimeter waveforms (with respect to the model) or sea state-induced biases on LRM data have been reported in the past. The antenna beamwidth of conventional altimeters is relatively wide so that the footprint is large enough to filter out the effects of waves on the sea surface. This would be no longer the case for SAR-mode altimeters since their resolution (~300m in Cryosat-2 configuration) is now smaller or of comparable size to the wavelength scales of swells, which makes the Doppler-processed echoes potentially sensitive to them. For any deviation from the Brown assumptions (due to non-Gaussianity of the wave height and slope distributions), altimeter echoes may have shapes differing significantly from the Brown-based model. By using a standard oceanic retracker that is not appropriate with non-traditional waveforms, non-negligible errors in the parameter retrievals would be expected.
This issue has been first addressed through numerical simulation, pointing out the possible effect of swell on SAR-mode waveforms that is all the more important as the wave heights and wavelength are high. Simulations also showed evidence of the effect of swell direction on altimetric measurements (for both along-track and across-track directions). At a lesser extent this effect has also been observed on simulated LRM mode waveforms, introducing substantial impact on the estimates.
No studies have so far investigated the sensitivity of the SAR-mode altimeter data to the swell with Cryosat-2 data, which is of high importance for the Sentinel-3 topography mission and the further Sentinel-6 mission, embarking both a SAR-mode radar altimeter. Similar questions are likely to come up with the LRM mode data.
CLS, and Ifremer and Meteo France partners, have been conducting a study, under CNES funding, to characterize these effects through both Cryosat-2 flight data analysis and simulation. An important component of this study is the joint use of collocated wave data from independent sources: wave buoy, satellite (such as altimeter derived significant wave height or SAR wavenumber spectra) and wave model, to enable the comparison of altimeter wave measurements with trustworthy data and report errors of the wave altimeter data when inconsistencies are found with respect to the sea state condition.
First, the paper presents the main results from this quantitative collocated-data analysis. Second, this paper deals with a numerical study based on the use of an end-to-end simulation tool including a SAR simulation capability with a generator of realistic models of long ocean surface waves and a SAR retracking algorithm. In particular, we will examine by simulation the effects of different swell spectrum characteristics (in direction and frequency) on the surface geophysical estimated parameters and assess the consistency of these results with those obtained on real data.
These first results will be useful in the near future to develop specific retracking algorithms or improve the current ones for the specific cases where SAR mode altimeter data are affected by swell.

This work presents and compares the latest altimeter calibration results for the Sentinel-3, Jason series, as well as the SARAL/AltiKa and the Chinese HY-2 missions, conducted at the Gavdos/Crete calibration/validation facilities. At first, the Jason altimeter calibration values will be given for the ascending Pass No.109 and the descending Pass No.18, based on the the GDR-E (Jason-1) and GDR-D (Jason-2) products. Secondly, these values will be cross-examined against the altimeter bias for the SARAL/AltiKa (GDR-T) satellite at Gavdos Cal/Val using its reference ascending orbit No. 571. The Chinese HY-2 satellite altimeter bias will be presented using the CRS1 permanent site in south west Crete for the descending HY-2 Pass No. 280, at 20 Hz based on SGDR data products. Finally, values will be compared against the Sentinel-3 altimeter. Additionally, altimeter biases as determined by locally developed Mean Sea Surface model will be presented and compared with the conventional sea-surface calibration methodology.

Delay/Doppler altimetry is a new technique that was used for the first time on the CryoSat-2 mission. This technique, inherited from the SAR imaging instruments and adapted to nadir altimetry by Raney 1998, aims to improve the along-track resolution by exploiting the altimetry signal Doppler frequency due to the instrument movement. In addition, this processing is expected to reduce the estimation noise by increasing the number of looks available for each surface Doppler resolution cell with respect to what is currently possible with conventional altimeters. On CryoSat-2 mission this processing is dedicated to sea ice areas (and ice caps margins with an additional interferometry processing).

While the contribution of the delay/Doppler processing can be considered obvious over ice areas (reduction of the along track resolution directly improves the distinction between sea ice and leads) it is not easy to extrapolate to the ocean surface. The first demonstration of the high capabilities of this processing to observe finer ocean scales than conventional altimetry has been conducted through CNES/CLS studies using CryoSat-2 real data over ocean in SAR mode and ESA CP4O project.

On Sentinel-3, the same technique (called the SAR-mode) is implemented and will be operated full time over the world. CLS was in charge of the analysis, the specification and the development of this processing for Sentinel-3 SRAL instrument. During this activity, several questions have been raised about the appropriateness of this processing for the ocean surface and whether this processing is optimal to reach the performances expected from Raney work.

We therefore proposed to perform a further analysis, revisiting these techniques to improve the SAR-mode performances in order to maximize the scientific returns from the Sentinel-3 mission and the upcoming Sentinel-6 mission. This analysis was supported by CNES and ESA. Several aspects have been analyzed to check the impact of each of the processing steps on the performance and to check if any improvement is still possible to enhance the performance. Among them:

The impact on the performance of a wrong alignment of the beams due to a bias in the knowledge of the surface location

The impact of a non perfect focusing of the beams on the surface due as it is applied over ocean

Is the accuracy of the orbit velocity vector enough for the expected performance

The impact of a wrong burst phase and power variations correction

…

In this study, we also performed an assessment of the behavior of the speckle noise affecting the SAR mode stacked waveforms because the performance observed using CryoSat-2 data, even highly improved versus the LRM mode, are not in accordance with the theory announcing a noise reduction directly linked to the number of beams in the stack. It has been demonstrated that the number of effective looks is lower than the number of stacked beams because the averaged beams have different mean shapes depending on the looking angle of measurement. This number, dependent on the significant wave height and on waveform range sample, has been computed theoretically and is consistent with the actual SAR retracking estimation noise.

To maximize the number of effective looks, an alternative stacking method has been studied by using smaller stacks optimized so that they contain beams with similar shapes. This method has been tested using the CNES CPP processing that has been applied on simulated and CryoSat-2 real data. The first results are promising as showed a behavior consistent with the theoretical analysis.

In the future, this work will continue by the analysis of a higher amount of CryoSat-2 data and hopefully with Sentinel-3 real data for a finalized version of the method.

In this presentation, we will provide the main conclusions of the studies performed and the results obtained with the new proposed stacking method. The perspectives of the work will be also presented.

One of the main benefits of delay/Doppler (D/D) altimetry is the improved resolution of the system along the satellite track. By means of an unfocused Synthetic Aperture Radar (SAR) processing technique, the altimeter footprint along the flight direction can be reduced by an order of magnitude with respect to conventional altimeters. This has allowed to resolve small-scale features on the ocean, and to provide altimetry data up to several hundreds of meters off the coast. The CryoSat-2 and Sentinel-3, operated in a closed burst form, make use of a coherent processing the 64 echoes within each burst (about 3.5 milliseconds of flight), which allows narrowing the footprint in the direction along the track to about 300 m.

We are experimenting with techniques and applications for fully focused coherent processing of radar altimeter echoes. By accounting for the phase evolution of the targets in the scene, it is possible to counter-rotate the phase of the FBR I/Q complex echoes along the aperture, and perform inter-burst coherent averaging potentially as long as the target illumination time. This process, similar to SAR imaging systems, reduces the along-track resolution down to the theoretical limit equal to L/2, where L is the antenna length. We call this the fully focused SAR Altimetry processing. For the development of the technique we have used the CryoSat-2 SAR Mode data, but our methods could also be used with similar data from Sentinel-3 or Sentinel-6/Jason-CS.

The footprint of a fully focused SAR altimeter measurement is an elongated strip on the surface, which is pulse-limited across-track and SAR focused along-track. The technique has been demonstrated using transponder data, showing an achievable along-track resolution of 0.5 meters.

Despite the asymmetry of the altimeter footprint, the fully focused technique may be useful for applications in which one needs to separate specific targets within highly heterogeneous scenes, such as in the case of sea-ice leads detection, hydrology, and coastal altimetry applications. Applying this technique on CryoSat-2 data over land and sea-ice, we can correctly measure the along-track extent of water bodies and ice-leads only a few meters long in the along-track dimension.

On a random rough surface like the open ocean, the fully focused altimeter waveform is a random realization of speckle noise. However, these single looks will be inherently uncorrelated between each other, so they can be incoherently averaged to obtain a multi-looked waveform that could potentially have significant improvements not only with respect to conventional and D/D (unfocused SAR) altimeters. Firstly, the fully focused SAR processing technique could potentially achieve an increase on the equivalent number of looks by a factor of 3 with respect to D/D. This could represent a significant speckle noise reduction in the final multi-looked waveform, and therefore an improvement in the estimation of geophysical parameters. And secondly, the multi-looked unfocused D/D altimeter waveform is a combination of several beams at different look angles. For each single beam, the impulse response widens as the square of the beam angle, so multi-looking has the effect of widening the final waveform. In contrast, the fully focused SAR altimeter achieves the limiting case of an infinitesimally narrow beam at nadir, resulting in a narrower waveform, which improves the estimation of both range and significant wave height.

Our initial results verify the previous points, and demonstrate the great potential of the fully focused SAR altimeter technique for a wide variety of applications.