Typhoon-induced megarips as triggers of turbidity currents offshore tropical river deltas

Tropical cyclones regularly generate extreme waves, surges and rainfall that are known to devastate coastal regions and, less well appreciated, to trigger sediment gravity flows posing risks to seafloor infrastructure. Multiple mechanisms can initiate turbidity currents, including submarine failures, river floods, downwelling and edge waves, but only recently have nearshore circulations during extreme events been directly linked to turbidity currents. Along energetic coasts, rip currents arise from alongshore variability in wave breaking controlled by morphology; during extreme conditions, large-scale megarip circulations with high velocities can extend well beyond the surf zone and export significant sediment offshore. This study addresses the initiation mechanism responsible for a turbidity current event inferred from a lateral displacement of a submarine pipeline offshore the Magasawang-Tubig (MT) River delta, northeastern Oriental Mindoro, Philippines. The MT delta debouches onto a very narrow shelf that transitions into a steep slope incised by canyons. A displacement detected in early 2016 where the pipeline crosses the largest canyon was attributed to turbidity currents, with Typhoon Melor (December 2015) suspected as the trigger. To resolve the mechanism, the authors combine high-resolution bathymetry, sub-bottom profiles, sediment cores, rainfall records, and a dedicated tropical cyclone hindcast, and couple a depth-averaged coastal circulation model with a 3D turbidity current model. The central hypothesis is that typhoon-induced megarips focused flow and sediment into a submarine canyon, igniting a turbidity current that impacted the pipeline.

Prior work documents that turbidity currents can be initiated by river floods, slope failures, downwelling, internal and infragravity waves, and storm-driven processes. Observations have linked pipeline damage to turbidity currents associated with typhoons, implicating intense alongshore currents directed toward canyons as triggers. The inner-shelf circulation under storms arises from interacting forces such as river plumes, tides, waves, and winds, modulated by nearshore heterogeneity and shoreline morphology. Rip currents form from alongshore gradients in wave setup and breaking; during extreme wave climates, megarips develop, extend beyond breakers, and can erode and export large sediment volumes. Studies in other regions (e.g., Monterey Canyon, Gulf of Lions, Taiwan) show storm and dense-water cascading influences on canyon processes, while modelling and field observations describe nearshore circulation, rip structures, and surf-zone dynamics. However, uncertainty persists in realistic boundary conditions for field-scale turbidity current simulations in coastal settings. This paper builds on and extends this literature by proposing a cyclone-induced megarip initiation mechanism tied to the rotation of wind and wave directions relative to embayed coastlines and adjacent deltas.

Data and site characterization: The study area offshore the MT delta includes a very narrow shelf and a canyon system with a prominent longest southeast canyon (LSEC). Two historical bathymetric datasets and a 2017 geophysical seabed survey (multibeam bathymetry and shallow reflection sub-bottom profiles) were merged into a 2 m × 2 m grid after tidal corrections and offset adjustments. EOMAP shallow-water satellite data and a 30 m DTM refined nearshore morphology and delineated the MT catchment and river network. Sediment cores (push and piston) collected inside and outside canyons were analysed: canyon cores contain 10–30 cm fine sand beds interpreted as turbidites, with outside cores showing thinner or absent sand layers; one proximal canyon core contained older stiff mud (radiocarbon ages >2000 years) indicating erosion. Hemipelagic mud sedimentation is ~0.3 cm/yr outside canyons vs ~0.02 cm/yr inside canyons; turbidite recurrence is as high as ~1 event per 43 years (3 in 130 years). Typhoon tracks from NOAA IBTrACS were compiled for Mindoro and Verde Island Passage. A reanalysis hindcast by Oceanweather Inc. (ADCIRC+SWAN, 1951–2016) provided significant wave height, period, direction, surge, wind, and depth-averaged currents at ~1 km resolution near Naujan embayment. Rainfall and hydrology: NOAA Calapan station provided 2000–2016 rainfall and meteorological records. Six-hour and daily values were downscaled to 3-hourly series for hydrological modelling, assuming spatial uniformity over the basin due to data limitations; typhoon-period rainfall was conservatively doubled to offset potential under-catch in high winds. A physics-based, spatially distributed rainfall–runoff model (unit hydrograph framework) was used to simulate MT catchment discharges, incorporating landscape morphology, land use, and climatic forcing. The model reproduced highly dynamic discharge responses with large floods when antecedent soil moisture was high. Sediment transport and hyperpycnal assessment: A sediment transport model estimated suspended sediment concentrations (SSC) at the MT delta outlets. River hydraulics were approximated as uniform flow with bankfull width ~25 m, depth ~1.8 m, bed slope ~2%. Grain-size proxies were inferred from offshore cores: mean sand 100–200 μm; an average 150 μm was adopted, with sensitivity spanning 20–150 μm to represent mud to sand mixtures. Transport predictor followed a generalized Engelund–Hansen formulation modified for fine sediments; total flow conductance used van Rijn’s approach. The MT river bifurcates into two outlets; SSC at each was computed from modelled sediment load divided by half the discharge. Even during Typhoon Melor’s peak flood (return period ~35 years), predicted SSCs remained below hyperpycnal thresholds for plunging underflows, including an extreme 100% mud scenario, indicating hyperpycnal flows were unlikely since pipeline installation. Coastal circulation modelling: High-resolution, depth-averaged (2DH) Delft3D was applied with online coupling of Delft3D-Flow and SWAN. Coupling timestep was 15 min; Flow solved depth-averaged shallow-water equations with a computational timestep ~10 s. SWAN used JONSWAP spectra; bottom friction coefficient 0.067 m^2 s^-3; wave breaking by Battjes and Janssen with γ=0.55; whitecapping and nonlinear interactions included. Flow bottom friction used Chézy with C=65 m^1/2 s^-1; wave–current interaction used Fredsøe’s model; horizontal eddy viscosity 1 m^2 s^-1. Wave computations ran in stationary mode after tests showed negligible differences from non-stationary for this application. The domain encompassed Naujan embayment with two nested grids: an outer curvilinear SWAN grid refined from ~100×50 m offshore to ~50×25 m nearshore; an inner Flow grid refined to 15×15 m near river mouths. Offshore SWAN boundary conditions were linearly interpolated from ADCIRC+SWAN nodes; Flow offshore boundary used water levels from ADCIRC+SWAN; lateral boundaries were Neumann (water level gradients). Spatially and temporally varying pressure (NOAA CFSR, 0.5°) and temporally varying, spatially uniform winds (from hindcast nodes) were applied. Total MT river discharge time series (from hydrology) were imposed just upstream of the delta bifurcation; bifurcation partitioning and branch geometry were estimated from satellite imagery of past floods. Mesh-independence tests and 24 h warm-up were performed. Three storm phases (rising, peak, falling) were simulated for each typhoon (Durian 2006, Melor 2015, Nock-Ten 2016), including river flood peaks with ~12 h lag after storm passage. Model outputs included depth-averaged currents and nearshore SSC along the 10 m isobath for subsequent turbidity current modelling. Turbidity current modelling: Three-dimensional simulations used TCsolver (validated multiphase RANS CFD for turbidity currents). Governing equations included incompressible bulk flow, mass conservation for each sediment class, and k–ε turbulence transport. Inflow boundary conditions were applied along the southwestern shallow boundary near the 10 m depth contour, specified from Delft3D outputs (velocity vectors, sediment concentrations) and sediment transport model results. Seabed composition used two classes (mud and sand) consistent with surveys; initial seabed was coarser on the shelf and well-mixed along the slope. Bed sediment entrainment followed García and Parker; mud-to-sand ratio in canyons was <5% (found to have limited effect on flow evolution). The simulations tracked current development along the slope and within canyons, especially the LSEC, and interactions with the pipeline alignment. Sensitivity tests indicated turbulence closure had modest impact for weaker flows but more influence for strong events. Malaylay–Baco comparative application: The Delft3D framework was also applied to Malaylay and Baco deltas (Verde Island Passage) to interpret prior typhoon-triggered turbidity current events impacting the same pipeline in 2006 and 2016. Circulation snapshots at peak conditions were analysed for megarip formation and deflection relative to canyon heads.

• Geophysical evidence indicates active turbidity currents in the MT canyon system: widespread deep-water dunes interpreted as antidunes (wavelengths 50–100 m, heights ~5 m) with upslope migration and orientations indicating flow convergence into the LSEC. A sea-floor headwall at ~120 m water depth was incised by younger channels. Only the longest southeast canyon (LSEC) extends beyond 250 m depth where a pipeline displacement was detected.
• Eleven typhoons passed near North Mindoro since 2001; three key events (Durian 2006, Melor 2015, Nock-Ten 2016) had eyes within ~25 km and category ≥3 intensity. Proximity governed wave and river discharge severity; waves >4 m were predicted when category ≥3 storms passed within 50 km.
• Hydrological and sediment transport modelling showed Typhoon Melor produced the largest flood (return period ~35 years), yet predicted suspended sediment concentrations at the MT outlets remained below hyperpycnal thresholds even under a 100% mud assumption, making hyperpycnal triggering unlikely during the pipeline’s lifespan.
• Delft3D simulations revealed contrasting coastal circulations: • Durian: persistent southeastward alongshore current with velocities >2 m s^-1 near the MT delta at peak, no flow convergence into the LSEC; no megarip formation directed offshore toward the canyon. • Nock-Ten: circulation similar to Durian, lacking megarip convergence toward the canyon. • Melor: clockwise rotation of wind and wave directions to near shore-normal (~45°N) at peak generated opposing alongshore currents within the Naujan embayment. Their convergence produced a strong megarip jet in the embayment center, directed offshore toward the LSEC, flushing water and sediment beyond the shelf-break.
• TCsolver predicted a severe turbidity current triggered within the LSEC only for Melor, with strong near-bottom velocities concentrated where the megarip intersected the shelf break; Durian and Nock-Ten did not generate intense canyon-confined currents. The modelled confined current impacted the pipeline nearly perpendicularly in the LSEC, matching the observed displacement location.
• Comparative modelling at Malaylay–Baco showed megarip circulations across a western embayment in all scenarios; deflection depended on storm approach: Durian and Nock-Ten (northwest-directed forcing) deflected megarips eastward toward Malaylay canyon (consistent with observed 2006 and 2016 events), while Melor (northeast forcing) deflected megarips westward away from those canyons, preventing intense flows there.
• The analysis identifies a new initiation mechanism: typhoon-induced megarips formed by the rotation of wind/wave directions relative to embayed shorelines between protruding deltas can focus sediment-laden jets into submarine canyons, igniting turbidity currents capable of displacing subsea pipelines.

The study resolves the previously uncertain initiation mechanism of a turbidity current that displaced a pipeline offshore the MT delta by demonstrating that Typhoon Melor generated a megarip current due to convergence of opposing alongshore flows in the Naujan embayment. The megarip advected water and suspended sediment beyond the shelf-break and into the head of the LSEC, where canyon confinement and self-acceleration ignited a strong turbidity current. Hyperpycnal river discharge was ruled out as the primary trigger, despite an extreme flood, because modelled SSCs were below plunging thresholds. The timing and localization of the simulated canyon-confined flow align with the observed pipeline displacement and with deep-water bedform evidence of high-energy underflows, supporting the causality of the megarip-trigger mechanism. The findings generalize to a conceptual model whereby the approach latitude of a tropical cyclone relative to an embayed coastline determines the rotation of incident waves and winds, modulating the direction of alongshore currents and whether convergent cells form. When convergence occurs, a transient megarip can flush the surf zone and inner shelf and drive focused sediment export into nearby canyon heads, initiating turbidity currents. The comparative application to Malaylay–Baco demonstrates the mechanism’s sensitivity to embayment geometry and storm approach: megarips may need to be deflected alongshore to feed canyons at embayment edges, whereas at MT the canyon lies centrally and requires an offshore-directed jet. This mechanism has broad implications for sediment transfer across continental margins and for risk assessments of seafloor infrastructure in cyclone-prone, morphologically complex coasts.

The paper identifies and substantiates a new initiation mechanism for turbidity currents in submarine canyons: typhoon-induced megarip currents formed by convergence of opposing alongshore flows in embayed shorelines. Using coupled depth-averaged coastal circulation modelling (Delft3D–SWAN) informed by a hindcast, and 3D turbidity current simulations (TCsolver) constrained by detailed seabed mapping and sediment data, the study shows that Typhoon Melor uniquely generated a megarip that focused flow and sediment into the LSEC off the MT delta, igniting a canyon-confined turbidity current that plausibly displaced a subsea pipeline. Hyperpycnal river discharge is unlikely to have triggered the event. Comparative modelling at Malaylay–Baco corroborates the role of storm approach and wave direction rotation in governing megarip formation and deflection toward canyon heads. The results highlight the importance of extreme-event coastal circulation in cross-shelf sediment transport and deep-water hazard generation. Future work should incorporate fully 3D coastal circulation to capture vertical structure and the transition from rip currents to underflows, improve field constraints on sediment supply and nearshore dynamics during cyclones, and extend hazard assessments for seafloor infrastructure in similar settings worldwide.

Key limitations include the absence of direct in situ measurements of river discharge and suspended sediment at the MT delta, necessitating rainfall–runoff and transport modelling with parameter estimates (e.g., bankfull geometry, grain size proxies from offshore cores). Rainfall forcing used a single station (Calapan) assumed spatially uniform over the basin, with typhoon-period values doubled to mitigate under-catch, introducing uncertainty. The exact timing of the turbidity current that displaced the pipeline is unknown; inference is based on circumstantial alignment with Typhoon Melor and model predictions. The coastal circulation model is depth-averaged (2DH), which cannot resolve vertical shear and stratification; nevertheless, literature suggests near depth-uniform rip currents within surf-zone channels, but offshore transitions remain approximated. Hindcast boundary conditions (ADCIRC+SWAN) have ~1 km resolution near the embayment, and winds were applied spatially uniform, potentially smoothing spatial variability. River bifurcation partitioning at the delta was estimated from imagery rather than measured. The turbidity current model relies on selected entrainment and turbulence closures and simplified two-class sediment mixtures; while validated and sensitivity-tested, parameter choices affect quantitative outcomes. Data ownership restrictions limit public availability of raw datasets for independent replication.