Abstract's details
Sea Level Monitoring in the coastal zone: impact of retracking and correction choices
CoAuthors
Event: 2017 Ocean Surface Topography Science Team Meeting
Session: Science I: Climate data records for understanding the causes of global and regional sea level variability and change
Presentation type: Type Oral
Contribution: PDF file
Abstract:
During the second phase (2014-2017) of the ESA Sea Level CCI Project (SL_cci) particular efforts have been put on improving the retrieval of sea level from altimeters in the coastal zone. These efforts aim at filling the observational gap between the rates of sea level rise (SLR) from tide gauges (at the coast) and the rates over the open ocean (from altimetry). Progress in coastal altimetry since the early 2000s has been steady, providing scope for filling this gap, and showing some encouraging results.
Within a dedicate workpackage of the SL_cci, NOC and SkyMAT Ltd have been looking at the impact of specialized retracking (ALES) and correction choices for the Wet Tropospheric correction (WTC) and Dry tropospheric correction (DTC) on rates of SLR retrieved in the coastal zone from along-track Jason-1 and Jason-2 measurements over 2002-2015. The study has focused on two wide coastal regions (A: Western European coast; and B: SE Australian coast) with different oceanographic conditions and tidal regimes. As terms of comparison we have taken: SLR rates from TGs; the gridded SL_cci v2.0 SLR rates, re-interpolated over the altimetric tracks; and the SLR rates from along-track altimetric datasets processed with conventional retracking (SGDR and RADS).
Our results show that a local comparison of altimeter-derived SLR rates with those from tide gauges remains problematic, due to the significant variability of SLR at short (tens of km) spatial scales. Over Region A the SLR rates from along-track altimetry do not match well the mean rates from the TGs, but they are closer to the median SLR rate from the TGs; the likely explanation for this is the extreme variability of the SLR rates over this very wide region, especially over the North Sea. Vertical land motion at the tide gauges is also likely to be a contributing factor to this discrepancy. The agreement improves substantially once the North Sea is excluded from the analysis. Over Region B, where SLR is more homogeneous and TG-derived mean and median trends coincide, the SLR rates computed from along-track altimetry for different datasets and combination of corrections are all very close to those from the TGs.
The choice of a particular retracker does not seem to strongly affect the results: the SLR rate curves as a function of distance from coast for the various data sets are all within 0.5 mm/yr from each other, which is the expected level of accuracy in the estimated trends. However the curves with the ALES dataset are the closest to the SL_cci curve in both regions, and approach the TG-derived rates in the last few km from the coastline in both regions, which is encouraging for the use of this specialized coastal retracker. Another important point is that ALES recovers a significantly higher percentage of valid range measurements in the last few km from the coast than the other datasets, as we illustrate in detail.
The choice of WTC has some noticeable impact on the SLR rates, but the differences are still at the level of the expected noise (~0.5 mm/yr). The GPD+ WTC by University of Porto, which is now recommended by the SL_cci project, yields the highest rates amongst all the considered WTC. The combination of GPD+ and ALES gives the best match to the TG-derived rates (and the closest to the reference SL_cci interpolated rates) over Region B.
The sensitivity of the SLR rates to the choice of DTC is very small, at the level of 0.1 mm/yr, so that no firm recommendation can be made on the choice of DTC in the coastal zone.
Interestingly, in region B we find a mild spatial trend of the rate of SLR, increasing towards the coast at approximately 0.01 mm/yr/km in the last 50 km. While not very large this is important as it would augment the impact of sea level rise on society, which happens predominantly at the coast. This coastal trend increase is clearest for the combination ALES and GPD+ in Region B.
In summary we have found over two separate regions that the combination of coastal retracking and ad hoc correction choices (in particular the GPD+) yields rates of SRL that compare to those from tide gauges almost as well (and in some cases better) than the rates from conventionally retracked datasets and standard correction models.
Within a dedicate workpackage of the SL_cci, NOC and SkyMAT Ltd have been looking at the impact of specialized retracking (ALES) and correction choices for the Wet Tropospheric correction (WTC) and Dry tropospheric correction (DTC) on rates of SLR retrieved in the coastal zone from along-track Jason-1 and Jason-2 measurements over 2002-2015. The study has focused on two wide coastal regions (A: Western European coast; and B: SE Australian coast) with different oceanographic conditions and tidal regimes. As terms of comparison we have taken: SLR rates from TGs; the gridded SL_cci v2.0 SLR rates, re-interpolated over the altimetric tracks; and the SLR rates from along-track altimetric datasets processed with conventional retracking (SGDR and RADS).
Our results show that a local comparison of altimeter-derived SLR rates with those from tide gauges remains problematic, due to the significant variability of SLR at short (tens of km) spatial scales. Over Region A the SLR rates from along-track altimetry do not match well the mean rates from the TGs, but they are closer to the median SLR rate from the TGs; the likely explanation for this is the extreme variability of the SLR rates over this very wide region, especially over the North Sea. Vertical land motion at the tide gauges is also likely to be a contributing factor to this discrepancy. The agreement improves substantially once the North Sea is excluded from the analysis. Over Region B, where SLR is more homogeneous and TG-derived mean and median trends coincide, the SLR rates computed from along-track altimetry for different datasets and combination of corrections are all very close to those from the TGs.
The choice of a particular retracker does not seem to strongly affect the results: the SLR rate curves as a function of distance from coast for the various data sets are all within 0.5 mm/yr from each other, which is the expected level of accuracy in the estimated trends. However the curves with the ALES dataset are the closest to the SL_cci curve in both regions, and approach the TG-derived rates in the last few km from the coastline in both regions, which is encouraging for the use of this specialized coastal retracker. Another important point is that ALES recovers a significantly higher percentage of valid range measurements in the last few km from the coast than the other datasets, as we illustrate in detail.
The choice of WTC has some noticeable impact on the SLR rates, but the differences are still at the level of the expected noise (~0.5 mm/yr). The GPD+ WTC by University of Porto, which is now recommended by the SL_cci project, yields the highest rates amongst all the considered WTC. The combination of GPD+ and ALES gives the best match to the TG-derived rates (and the closest to the reference SL_cci interpolated rates) over Region B.
The sensitivity of the SLR rates to the choice of DTC is very small, at the level of 0.1 mm/yr, so that no firm recommendation can be made on the choice of DTC in the coastal zone.
Interestingly, in region B we find a mild spatial trend of the rate of SLR, increasing towards the coast at approximately 0.01 mm/yr/km in the last 50 km. While not very large this is important as it would augment the impact of sea level rise on society, which happens predominantly at the coast. This coastal trend increase is clearest for the combination ALES and GPD+ in Region B.
In summary we have found over two separate regions that the combination of coastal retracking and ad hoc correction choices (in particular the GPD+) yields rates of SRL that compare to those from tide gauges almost as well (and in some cases better) than the rates from conventionally retracked datasets and standard correction models.