The research question
In coastal zones and estuaries, ocean tides play a key role for a wide range of processes and communities. They are fundamental contributors to coastal flooding, pollutant dispersal, coastal ecosystems, and navigation, even influencing the beachgoing experience. Tides vary considerably in coastal regions, making it challenging to predict and monitor them using remote sensing, numerical models, and in situ observations. Satellite altimetry has been a vital source of data for expanding our understanding of tides across the global oceans. However, due to their design, they traditionally struggle to retrieve reliable sea-level measurements near the coast, with state-of-the-art methods typically used only up to 3 km offshore. Yet, it is in these final kilometers also that the greatest changes occur in ocean tides (see Hart-Davis et al 2024).
In recent years, advances and increased availability of both methods and data have led to a significant rise in the use of satellite imagery, particularly in coastal research. The over forty-year observation record provided by a combination of Landsat and the Sentinel satellites has resulted in significant research investigating shoreline variations and how this is influenced by major changes in the climate system, e.g., the work of Vos et al (2023) that looked at the impact caused by the El Niño/Southern Oscillation.
At the scales of this research in the coastal zone, tides play a significant role and are 'an "annoying" source of noise' that needs to be addressed and removed to study other phenomena. But the abundance of research and the need to 'correct' for ocean tides opened the question in our heads, "If they need to remove tides, surely we can recover tides?"
The findings
Given the importance of shoreline research and the significant efforts of the broader community, time series of shoreline measurements have been made publicly available. This presented us with the opportunity to answer the above question. Particularly, the 40 + years of shoreline measurements presented in Vos et al (2023) across the entire Pacific Rim, where tides vary considerably.
An important consideration before proceeding is the importance of beach slopes for this research. Knowledge of the beach slope enables us to convert shoreline measurements from a horizontal to a vertical reference, thereby extracting height variations. A current limitation of space-borne slope estimations is that no study can capture a variable beach slope from space, at least not variable enough to match each measurement of the shoreline position. However, several efforts have been made to extract a single beach slope that represents the mean over the entire observation period (e.g., 40+ years). Thanks to this, we can convert horizontal measurements to vertical measurements and extract tidal information; however, if local knowledge of beach slopes or new methods are presented in the future, this will only improve the estimability of tides from shoreline measurements, a key potential future research topic.
Once all the shoreline images are collected and the slope applied (e.g., the top half of Figure 1), 40+ years of previously unused sea-level observations are available for every 100 meters of beach across the Pacific Rim (e.g., of a time-series in Figure 1 bottom). From this, we run a harmonic analysis (HA) and response method (RM) to extract key tidal components. Focusing purely on the M2 here (in the manuscript, we expand this), Figure 2 provides an example of the extracted tides from these methods across New Zealand. Qualitatively, we observe that our estimated coastal tides match the spatial features of the ocean model FES2022 and tide-gauge observations in capturing the variability along the coastline. Overall, we derive a root-mean-square error of ~7 cm for both observations and models throughout the Pacific Rim.
Looking ahead
We clearly see that tides can be predicted from satellite imagery-based shoreline measurements, and these reveal fine-scale variability not captured in in-situ observations. We provide insights that can be used not only to inform tidal variability but also to validate our models in regions not covered by in-situ measurements. The 40+ year time series can also be used to assess how tides vary over time, which we tentatively examine in the manuscript. As methods continue to advance and new avenues of research are explored within this community, especially regarding time-varying beach slopes, the use of these data for sea-level and tidal research will be incredibly important to a wide range of scientific communities.
Manuscript written by: Michael Hart-Davis, Thomas Monahan, Kilian Vos, Ole Andersen