This paper, published in Ocean Science, investigates how tides and storm surges "talk" to each other along the coasts of Singapore and Peninsular Malaysia. By combining a clever new statistical test with a physics-based model, the researchers show that storm surges actually change the timing of the tides, a phenomenon that varies predictably as you move along the coastline.
TL;DR
Researchers developed a new statistical framework and a semi-empirical model to prove that tide-surge interaction is a major factor in Southeast Asian coastal waters. They discovered a distinct spatial gradient where extreme water levels peak before high tide in the south but after high tide in the north, primarily driven by surges altering the speed of tidal waves.
What Problem Does This Paper Address?
In coastal engineering and flood risk assessment, we often treat "tides" (predictable) and "surges" (weather-driven) as independent additions. However, in shallow waters like the Sunda Shelf, they are non-linearly coupled. This is "tide-surge interaction" (TSI).
The challenge is that standard statistical methods for detecting TSI often assume tidal cycles are perfectly symmetric and uniform. In regions with "mixed" tides (like Singapore), this assumption fails, leading to "false positive" signals—where researchers think they see an interaction that is actually just a quirk of the local tidal shape. This paper aims to fix that statistical bias and explain the physical mechanism behind the observed interactions.
Methods
The authors employed a two-pronged approach:
- The "no-TSI" Null Distribution: Instead of comparing observed surges against a flat, uniform distribution, they created a site-specific "null" distribution. This accounts for the actual duration and asymmetry of local tidal cycles, ensuring that any detected interaction is physically real rather than a statistical artifact.
- Semi-Empirical Modeling: They built a model using ERA5 atmospheric data (wind and pressure) to estimate surges. Crucially, they included a tidal-phase alteration component. This uses the shallow-water wave speed formula ($c = \sqrt{gD}$) to calculate how a surge-induced increase in water depth ($D$) speeds up the incoming tidal wave, shifting the timing of high water.
Datasets
- Tide Gauge Records: High-quality hourly data from 7 stations (Tanjong Pagar, Johor Baharu, Sedili, Tioman, Kuantan, Cendering, and Geting) provided by the University of Hawaii Sea Level Center, spanning roughly 30 years.
- Atmospheric Forcing: ERA5 reanalysis data for 10m winds and mean sea-level pressure.
- Bathymetry: GEBCO (General Bathymetric Chart of the Oceans) for depth data.
Key Insights and Ideas
- The Timing Shift: Extreme residuals (the "extra" water from a surge) don't happen randomly. In the south (Singapore), they cluster ~5 hours before high tide. In the north (Geting), they cluster ~4 hours after high tide.
- The $M_2/K_1$ Connection: The transition from "pre-high-tide" to "post-high-tide" peaks correlates with the ratio of semi-diurnal ($M_2$) to diurnal ($K_1$) tidal constituents. As the $M_2$ influence weakens moving north, the timing of the interaction shifts.
- Surges Change the "Speed Limit": The "aha moment" is that a storm surge doesn't just sit on top of the tide; it increases the total water depth, which allows the tidal wave to travel faster. This "phase alteration" is a dominant mechanism for TSI in this region.
- Statistical Rigor: The paper demonstrates that using a standard uniform distribution would have resulted in false-positive signals at several stations, proving the necessity of their "no-TSI" framework.
Key Results
- Confirmed Interaction: TSI was statistically significant at all seven locations.
- Bimodal Distributions: Extreme residuals are least likely to occur exactly at high tide, creating a "bimodal" frequency distribution. This is actually good news for flood risk, as the largest surges rarely coincide perfectly with the highest astronomical tides.
- Model Performance: The semi-empirical model's surge estimates had a correlation of 0.7–0.8 with actual residuals. While simple, the model successfully reproduced the bimodal timing patterns observed in the real-world data.
- Phase Alteration Impact: Even though phase alteration accounts for less than 1% of the total residual variance, it is the primary driver behind the timing of when those residuals peak.
Limitations and Future Work
- Simplified Physics: The semi-empirical model assumes tides travel in a straight line toward the coast and ignores complex factors like bottom friction and momentum advection.
- Geographic Specificity: The model was tailored for the Sunda Shelf. While the statistical framework is globally applicable, the physical model would need recalibration for regions with different shelf geometries.
- Data Length: While 30 years is good, the authors note that the relatively short records in Southeast Asia still pose a challenge for long-term "return level" (1-in-100-year flood) estimations.
- Future Direction: The authors suggest that "skew surge" (the difference between the maximum observed water level and the predicted high tide) is a more robust metric for this region to mitigate the complexities of phase alteration.
Summary updated