Evaluation of coastal land subsidence and inundation risk in coastal Kelantan, Malaysia

The coastal regions of Kelantan, Malaysia, are predominantly low-lying and vulnerable to inundation owing to rising sea levels and strong monsoonal waves. Excessive groundwater extraction, which accelerates land subsidence, may further increase these risk factors. A Persistent Scatterer Interferometric Synthetic Aperture Radar (PS-InSAR) analysis was conducted using 162 Sentinel-1 ascending images from January 2020 to July 2025 to assess land subsidence rates and trends, particularly along the coast of Kelantan. The present coastal inundation risk map was generated by integrating Sentinel-2 imagery and Digital Elevation Model (DEM) with a deep learning U-Net model (B0 backbone). To produce an accurate coastal land cover map for the Kota Bharu and Tumpat districts, Sentinel-1A SAR and Sentinel-2 optical imagery were employed within the Google Earth Engine (GEE) platform, utilizing four machine learning classifiers: Random Forest (RF) and Support Vector Machine (SVM), in both default and tuned configurations. Observations indicate a linear increase in land subsidence, with a mean subsidence velocity of 1.11 ± 0.73 mm year−1, specifically across the coastal area. The 2021 coastal inundation model showed high agreement with the previous study, with a kappa coefficient of 0.856. The tuned RF classifier achieved the highest accuracy in land-cover classification for the Kota Bharu and Tumpat areas, with a kappa coefficient of 0.876. The final coastal inundation risk map indicates that the Kelantan River estuary is highly vulnerable to flood and is also experiencing significant land subsidence. Given the ongoing subsidence, the findings of this study provide valuable guidance for policymakers prioritizing high-risk areas for targeted coastal management strategies.