Tidal Wave Transformation in the Mahakam Delta: Integrating Observations and Machine Learning

Abstract

Tidal dynamics in tropical estuaries have made it difficult to predict water level variations and to understand the interaction between river discharge and ocean forcing. This study investigated tidal wave propagation, attenuation, and asymmetry in the Mahakam Delta, East Kalimantan, Indonesia. The objective was to quantify tidal wave transformation along the estuary-upstream transect and to evaluate the capability of machine learning to reproduce key hydrodynamic parameters. Water level time series with a 10-minute interval, collected from 12 July to 4 August 2018 at an estuary station and an upstream station, were analysed. A Random Forest regression model with temporal lag features predicted upstream water level with a coefficient of determination (R-squared) of 0.883 and a mean absolute error of 0.150 m; the most informative lag was 60 minutes. Analysis of 34 tidal cycles showed a mean attenuation coefficient of 0.76, indicating hyposynchronous conditions dominated by friction. A machine learning model for attenuation coefficient prediction reached an R-squared of 0.968 and a mean absolute error of 0.084, with estuary peak water level as the most influential predictor. The delta exhibited flood-dominant tidal asymmetry and negligible phase lag between estuary and upstream peaks. These findings demonstrated that the integration of data-driven modelling and hydrodynamic analysis can improve the prediction of water levels and support flood risk assessment in tropical deltaic systems.

Keywords

• Deltaic Hydrodynamics
• Hyposynchronous Estuary
• Mahakam Delta
• Random Forest Regression
• Tidal Wave Propagation