Menu Close

Uncertainties and GMSL

Homogenization/mitigation of differences and/or discrepancies captured in ocean products available from S6-MF, Jason-3, Sentinel-3 and other satellite data

Sentinel-6 MF will play a key role to extend the Topex/Jason series that monitors the Mean Sea Level variations for more than 27 years. Such a long time series helps to characterise the pace of the sea level rise all over the globe which is now part of the Essential Climate Variables (ECVs) distributed by the Copernicus Climate Change Service.

The long-term quality and stability of such variables is ensured thanks to precise inter-calibration of the successive altimetry missions that have flown on the historical Topex’s orbit. Such in-depth inter-calibrations have been performed during tandem phases that helped to mitigate discrepancies between Topex/Jason-1 (Ablain et al., 2012) as well as Jason-1/-2 and -3, at the sensor parameters level (range, sigma-0, SWH) as well as for the geophysical corrections such as the Sea State Bias (Tran et al. 2010) that contribute to the SSH computation and thus to the MSL.

In addition to the conventional LRM, S6-MF will use a SAR mode which has never been operated on previous reference missions and consequently will require specific attention to ensure a homogeneous transition. The bias estimation between S6-MF and Jason-3 will directly impact the global and regional MSL trends and their uncertainties. Zawadzki et al. (2016) showed that the Jason-2/Jason-3 GMSL bias uncertainty is of 1 mm (90%) and the resulting GMSL trend uncertainty (due only to the bias) is of 0.14 mm/yr over the last 10 years of these missions’ operational time. In case of loss of the reference mission, they showed that switching to Sentinel-3A increases the GMSL trend uncertainty by a factor 3, mainly due to the change of orbit. It is important to notice that these figures are based on simulations of Jason-3 and Sentinel-3A data (real data were not yet available in 2016) and based on a method that assumed strong properties of the Jason-2/-3 and Jason-2/Sentinel-3A inter-comparisons (noise levels and inter-correlations).

Recent studies performed in the S3TC ESA project proposed an update of their method based on actual Sentinel-3A/B data and showed that special care needs to be taken to connect two missions to obtain adequate GMSL trend uncertainties (Clerc et al. 2020). The S6-MF tandem phase with Jason-3, combined with the Sentinel-3 missions will give the opportunity to consolidate the current method of bias estimation based on a reference mission, as well as to explore new methods such as a multi-mission approach and/or the use of FRM data. This will allow quantifying how accurate these new methods are as compared to the one used so far.

This activity aims at:

  • Estimating the regional and global MSL biases between S6-MF and Jason-3 and their associated uncertainties
    • For all available modes (LRM/LRM, LRM/SAR, etc.)
    • Use a statistical approach as done in Clerc et al. (2020) to estimate the GMSL bias uncertainties 
    • Propagate these onto the GMSL trend uncertainty and characterize the long-term stability of the record. 
    • For the regional MSL, we will perform similar analyses at different regional scales, with a focus on the spatial correlation determination of the residual signals. We will be able to determine at which spatial scale the MSL data records is accurate enough for any scientific needs.
  • Comparing the proposed approach to alternative methods (multi-mission approach, tide-gauge comparison) 
    • Quantify by how much the bias uncertainties are increased if one uses alternative approaches:
      • Multi-mission approaches with different orbits (Sentinel-3’s)
      • Multi-mission approaches with and w/o tandem phases
      • Tide-gauge data 
    • Determine what is the optimal approach to improve the uncertainty?”

REFERENCES

  • Ablain, M., Meyssignac, B., Zawadzki, L., Jugier, R., Ribes, A., Cazenave, A., et al. (2019). Uncertainty in satellite estimate of global mean sea level changes, trend and acceleration. Earth Syst. Sci. Data Discuss. 1–26. https://doi.org/10.5194/essd-2019-10 
  • Ablain, M., Philipps, S., Ollivier, A., Picot, N., Mitchum, G., Scharroo, R., Lillibridge, J., Haines, B., Beckley, B., Desai, S. (2012) Reciprocal Benefits of Multi-mission Satellite Altimetry Comparison. 20 years of Progress in Radar Altimetry Symposium. 
  • Clerc, S., Donlon, C., Borde, F., Lamquin, N., Hunt, S.E., Smith, D., McMillan, M., Mittaz, J., Woolliams, E., Hammond, M., Banks, C., Moreau, T., Picard, B., Raynal, M., Rieu, P., Guérou, A. Benefits and Lessons Learned from the Sentinel-3 Tandem Phase. Remote Sens. 2020, 12, 2668. https://doi.org/10.3390/rs12172668 
  • Guérou, A. et al., S3VT 2020, Investigating the Sentinel-3 SARM range drift 
  • Prandi, P., Meyssignac, B., Ablain, M. et al. Local sea level trends, accelerations and uncertainties over 1993–2019. Sci Data 8, 1 (2021). https://doi.org/10.1038/s41597-020-00786-7 
  • N. Tran, S. Labroue, S. Philipps, E. Bronner & N. Picot, 2010. Overview and Update of the Sea State Bias Corrections for the Jason-2, Jason-1 and TOPEX Missions, Marine Geodesy, 33:sup1, 348-362. https://doi.org/10.1080/01490419.2010.487788. 
  • Zawadzki, L. and Ablain, M., 2016. Accuracy of the mean sea level continuous record with future altimetric missions: Jason-3 vs. Sentinel-3a, Ocean Sci., 12, 9–18, https://doi.org/10.5194/os-12-9-2016. 

Study of alternative approaches to inter-calibration of S6-MF and Sentinel-3 SRAL while in tandem

Intercalibration studies allow differences between altimetry missions to be detected, enabling in-depth investigations to be carried out to understand their origin and then correct them. Several types of error signals such as biases, drifts or signals correlated in time and space can be detected for the main altimetry parameters (e.g. SWH, Sigma0), but also geophysical corrections (e.g. wet tropospheric correction) and, most importantly, sea surface height. This is possible in different satellite configurations: 1) when the satellites are on two different orbits, which is the most common situation; 2) during tandem phases, when two satellites are on the same orbit with a few seconds time lag. Thus, in the past, significant geographically correlated SSH biases have been detected between TOPEX and Jason-1 (Ablain et al., 2012) during the tandem phase of these missions. It was mainly due to errors in the orbit solutions but also to heterogeneous processing between missions (e.g. sea state bias correction). The correction of these errors made it possible not only to homogenise the two missions, but also to improve each of the two timeseries. Many other examples can be provided: for instance, the detection of significant GMSL drifts in the ENVISAT GMSL in 2003/2005 by comparison with Jason-1 (Olivier et al., 2012) and more recently in the S3A/S3B GMSL by comparison with Jason-3 and SARAL-Altika (Jugier et al., 2020; Guerou et al., 2020). By helping to detect and correct many important altimetry errors, intercalibration studies are therefore essential to ensure a good accuracy of all altimetry products, and especially the GMSL indicator. It is also very likely that the current tandem phase between Jason-3 (JA3) and Sentinel-6 Michael Freilich (S6-MF) will provide relevant information in terms of differences and discrepancies between the two missions as these potential differences will be very precisely estimated and certainly statistically significant. However, if such discrepancies are found, and despite their very good accuracy, the errors cannot be attributed to either of the two missions. Therefore, one of the most relevant ways to attribute the errors detected is to use other altimetry missions such as Sentinel-3 (S3A and S3B). Cross-comparisons between S3A and S3B with S6-MF and JA3 while in tandem will provide us with another source of information to analyse the discrepancies between all these missions.

This activity aims at:

  • Specifying the alternative intercalibration method between Sentinel-3 SRAL and S6-MF.
    • To find the best balance between, on the one hand, having accurate but few measurements and, on the other hand, having a lot of measurements but little accuracy due to ocean variability.
    • To carry out a sensitivity study on the uncertainties of the method, by progressively varying the spatial and temporal collocation criteria of the S6-MF and S3-A (or S3-B) measurements.  
  • Analysing in depth the level of uncertainty obtained
    • By calculating its evolution over time for both bias and drift estimates at global and regional scales, and for the main physical variables (e.g. ssh, swh, sigma0, …). 
    • By comparing these uncertainties with those obtained during a tandem phase, which is the optimal case (maximum number of measurements and very accurate). 
    • By applying the approach to some concrete discrepancies found between J3 and S6 .

REFERENCES

  • Ablain, M., Meyssignac, B., Zawadzki, L., Jugier, R., Ribes, A., Cazenave, A., et al. (2019). Uncertainty in satellite estimate of global mean sea level changes, trend and acceleration. Earth Syst. Sci. Data Discuss. 1–26. doi: 10.5194/essd-2019-10. 
  • Ablain, M, Philipps, S, Ollivier, A, Picot, N, Mitchum, G, Scharroo, R, Lillibridge, J, Haines, B, Beckley, B, Desai, S (2012) Reciprocal Benefits of Multi-mission Satellite Altimetry Comparison. 20 years of Progress in Radar Altimetry Symposium.  
  • Donlon C., OSTST, 2020: Altimetric Reference Transfer Standard (ARTS) 
  • Jugier R. et al., OSTST, 2020: What sea-level drifts can be detected at global and regional scales by comparing recent altimetry missions together: S3A, Jason-3 and Saral-Altika? Available here. 
  • Guerou et al., S3VT 2020, Investigating the Sentinel-3 SARM range drift 
  • A. Ollivier, Y. Faugere, N. Picot, M. Ablain, P. Femenias & J. Benveniste (2012) Envisat Ocean Altimeter Becoming Relevant for Mean Sea Level Trend Studies, Marine Geodesy, 35:sup1, 118-136, DOI: 10.1080/01490419.2012.721632.