Civil Engineering and Architecture Vol. 14(3), pp. 1509 - 1519
DOI: 10.13189/cea.2026.140309
Reprint (PDF) (1272Kb)

High-Resolution Climate Risk Modeling for Coastal Zones Using an Innovative Downscaling Technique

Heri Sulistiyono ^1,*, I Wayan Yasa ¹, Eko Pradjoko ¹, Dewandha Mas Agastya ¹, M. Ari Firdaos ¹, Bing Chen ^2
1 Department of Civil Engineering, University of Mataram, Indonesia
² Department of Civil Engineering, Memorial University of Newfoundland, Canada

ABSTRACT

Coastal communities face mounting threats from climate-induced disasters, driven by rising sea levels and intensifying hydrometeorological extremes. This study introduces a reproducible, high-resolution downscaling framework that integrates numerical extrapolation, regression analysis, and artificial neural networks (ANN) to assess future coastal hazard scenarios. The approach combines the interpretability of regression with the nonlinear pattern-recognition capabilities of ANN, enabling robust projections of inland tidal inundation and extreme rainfall under IPCC AR6 CMIP6 scenarios (historical 1994–2020; SSP2-4.5 and SSP5-8.5 to 2100). Inputs include bias-corrected regional climatology, station and reanalysis records (1994–2024), local hydrometeorological observations, drainage maps, population and infrastructure layers, and the 6 July 2025 flood event for calibration. Ensemble selection, quantile-mapping bias correction, and hybrid regression–ANN modeling, are used to preserve linear relationships while capturing complex interactions. Methods include correlation-based indicator selection, cross-validated training with RMSE and MAE metrics, numerical flood simulations, and geospatial hazard–exposure–vulnerability mapping. Applied to Ampenan, West Nusa Tenggara, the framework projects inland tidal flooding up to 300 meters and rainfall intensities exceeding 3,000 mm, validating its utility for localized risk assessment. Outputs include ensemble-median projections with 5–95% uncertainty ranges and prioritized adaptation recommendations, including structural reinforcements, ecosystem-based buffering, and early-warning systems. The framework is replicable and scalable, offering a transparent tool for climate-informed planning across diverse coastal contexts. Limitations include data resolution and evolving land-use dynamics, which may affect long-term accuracy. Practically, the model supports municipal planning, infrastructure investment, and targeted interventions for vulnerable populations. Socially, it enhances inclusive risk communication and community preparedness. By bridging statistical and machine learning techniques, this study contributes a novel methodology for disaster risk assessment and strengthens resilience in the face of accelerating climate change.

KEYWORDS
Coastal Disasters, Climate Change, Downscaling Methodology, Hybrid Regression-ANN, Ampenan Coastal Area

Cite This Paper in IEEE or APA Citation Styles
(a). IEEE Format:
[1] Heri Sulistiyono , I Wayan Yasa , Eko Pradjoko , Dewandha Mas Agastya , M. Ari Firdaos , Bing Chen , "High-Resolution Climate Risk Modeling for Coastal Zones Using an Innovative Downscaling Technique," Civil Engineering and Architecture, Vol. 14, No. 3, pp. 1509 - 1519, 2026. DOI: 10.13189/cea.2026.140309.

(b). APA Format:
Heri Sulistiyono , I Wayan Yasa , Eko Pradjoko , Dewandha Mas Agastya , M. Ari Firdaos , Bing Chen (2026). High-Resolution Climate Risk Modeling for Coastal Zones Using an Innovative Downscaling Technique. Civil Engineering and Architecture, 14(3), 1509 - 1519. DOI: 10.13189/cea.2026.140309.