Possible datum errors at tide gauges detected by satellite altimetry: some case studies
Event: 2022 Ocean Surface Topography Science Team Meeting
Session: Regional and Global CAL/VAL for Assembling a Climate Data Record
Presentation type: Type Oral
Contribution: PDF file
The OSTST is accustomed to using tide gauge data for validation of satellite altimetry, and even, when the need arises, for calibration of altimetry (e.g., when determining the TOPEX Side-A to Side-B bias). Here we turn the problem around -- we use altimetry to find possible cases of datum errors in about a dozen tide gauges, all of them high-quality stations, most located on small islands. We estimate datum errors by solving simultaneously for implied linear vertical land motion (VLM) at the tide gauge, plus one or more discontinuous jumps. This is done using daily mean tide-gauge sea levels and two complementary altimeter gridded datasets: Aviso/CMEMS data and MEaSUREs data. The gauge data must be adjusted for long-period tides, the pole tide, and dynamic ocean loading (consistent with that used in altimetry processing). The estimated errors/jumps in all cases are fairly small, usually only a few cm; the largest found is about 10 cm. Analysis of implied VLM provides perhaps the most convincing evidence that the detected tide-gauge errors are real: the VLM for the periods before and after detected jumps, as well as the VLM for the whole (adjusted) time series, should show improved consistency. The adjusted VLM should also show better consistency with VLM measured with nearby GNSS instruments, although this is sometimes ambiguous owing to differences between multiple GNSS stations, their distances from the tide gauge, and uncertainties in GNSS rates due to equipment changes or errors. For the dozen or so cases we have discovered, we are revisiting tide-gauge instrumentation and associated documentation for evidence of the suggested errors. As of this writing, we have resolved one case, at Rodrigues, where a jump during October 2013 has been traced to a site visit during which cables to two water-level calibration sensors were inadvertently reversed. The sensors are separated by 43 mm, and reversing them adjusts measured sea level by that amount. The altimeter-based tide-gauge offset had previously been estimated at 32 ± 6 mm (one sigma), which is consistent within a 95% confidence interval of the offset from the cable reversal. This datum error at Rodrigues has now been corrected in the public data archives of the UH Sea Level Center. This use of altimetry to identify tide gauge errors, combined with independent information to correct the errors, obviously then benefits altimetry thanks to the improved validation datasets.