| Popis: |
By enhancing the comparability of coastal satellite altimetry (SAT) and tide gauge (TG) measurements, we improve the determination of vertical land motion (VLM) at the coast, which is a substantial contributor to relative sea level trends. Coastal GNSS measurements of VLM are accurate, however, the station distribution is not homogeneous worldwide and GNSS data are not available at large areas. An alternative approach is to subtract absolute satellite-based sea level (SL) measurements from relative SL changes recorded by tide-gauges. This application is a promising complement to GNSS VLM-measurements, given the over 20-year long records from altimetry and the larger availability of TGs. In this study, we focus on the improvement of this approach, with respect to the accuracy of resulting trends and the uncertainties of the estimates. We investigate the impact of innovations in two fundamental components that contribute to the performance of trend estimates by SAT-TG difference: (1) Quality and resolution of the data and (2) the method for SAT-TG combination. First (1), we analyze performance of the gridded altimetry product AVISO against multi-mission along-track altimetry, which features latest improvements of coastal retracking (based on ALES) and geophysical corrections. VLM trends are derived by combining sea level anomalies (SLAs) within a 250km radius around TGs from the monthly PSMSL database. Comparing the RMS of differences of SAT-TG and GPS (ULR6a) trends, application of monthly averaged along-track altimetry instead of gridded products reveals no significant improvements neither by the RMS nor the trend uncertainties. In a second step (2) we introduce refined selection criteria of SLAs to reduce residuals of the differenced time-series and thus the associated trend uncertainties. Therefore, we match high-frequency TG data from the “Global Extreme Sea Level Analysis” (GESLA) data set with the along-track data, to determine zones of coherent sea level variability at the highest possible spatial and temporal scale. Using correlation, RMS or the residual annual cycle, we isolate spatial coherent SLA pattern in the coastal regions, representing the best match of SL variability measured at the tide gauge. Such ‘Zones of Influence (ZOI)’ capture predominant dynamics of coastal variability e.g. coastal currents and often project onto bathymetric gradients. High frequent along-track altimetry in combination with GESLA better detects small scale features of those ZOIs than the gridded, monthly-sampled product. We compute trends from SLAs, which are averaged within the ZOI. Variation of relative thresholds of the aforementioned statistical criteria provides sub-sets of SLAs of different comparability. An optimal relative threshold is identified by minimization of the uncertainties of the estimated trends as well as their deviations from the GPS trends. This novel method of selection reduces the RMS by 20% and uncertainties by 25% with respect to the 250km-radius-selection of along-track data. We show that the application of advanced along-track multi-mission altimetry data fosters the adjustment to the fine structures of coastal sea level variability, which leads to significant improvements in long-term coastal sea level and VLM trend determination. |
