EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Integrated Hydro-Hazard Index (HHI) for Drought-Flood Risk Assessment: A Multi-Temporal Machine Learning Approach

Nutchanat Buasri, Patiwat Littidej, Benjamabhorn Pumhirunroj, Jatuphum Juanchaiyaphum, Donald Slack

Sustainability · 21.07.2026 · 18 (14) · 7448

DOI: 10.3390/su18147448 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Sustainability, online erschienen am 21.07.2026. Verfasst von Nutchanat Buasri und 4 weiteren Autor:innen. Die Arbeit wurde bisher 4-mal zitiert (Crossref) und nennt 61 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Mahasarakham University.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Climate change is intensifying hydrological extremes, yet most frameworks assess drought and flood hazards independently, limiting integrated risk management. This study proposes a two-dimensional analytical framework to characterize the drought-flood continuum, moving beyond single-index approaches. We introduce the Hydro-Hazard Index (HHI) as a directionality metric (HHI = Flood Severity − Drought Severity) to classify the dominant hazard type, and the Total Severity Index (TSI = Flood Severity + Drought Severity) as a complementary metric to quantify overall hazard magnitude. Analyzing multi-temporal data from 115 hexagonal units (2018–2024), we employed dynamic features (trends, changes, volatility) and four machine learning models to classify areas as “flood-prone” based on validated flood records. Our results show HHI values ranging from −2.44 to 8.81, with 20.9% of areas classified as Flood-Dominated (mean HHI = 4.58) and 79.1% as Normal (mean HHI = 0.76). Crucially, the two-dimensional analysis revealed that areas with identical HHI values can have vastly different TSI values, under scoring the importance of our dual-index approach. Random Forest achieved the highest performance in predicting flood-prone status (Accuracy = 0.913, AUC = 0.967, Recall = 1.00), with flood_volatility as the most important predictor (24.2%). Spatial autocorrelation confirmed strong clustering of high-risk areas (Moran’s I = 0.716, p < 0.001). By analyzing flood and drought as distinct but interacting dimensions, this framework provides a more robust and nuanced tool for integrated risk assessment. While acknowledging limitations related to data availability and the need for further independent validation, the proposed framework supports sustainable water resource management and climate adaptation planning under increasing hydrological uncertainty.

Förderung

  • Mahasarakham University

Zitieren

APA

Buasri, N., Littidej, P., Pumhirunroj, B., Juanchaiyaphum, J., & Slack, D. (2026). Integrated Hydro-Hazard Index (HHI) for Drought-Flood Risk Assessment: A Multi-Temporal Machine Learning Approach. Sustainability, 18(14), 7448. https://doi.org/10.3390/su18147448

Vancouver

Buasri N, Littidej P, Pumhirunroj B, Juanchaiyaphum J, Slack D. Integrated Hydro-Hazard Index (HHI) for Drought-Flood Risk Assessment: A Multi-Temporal Machine Learning Approach. Sustainability. 2026;18(14):7448. doi:10.3390/su18147448

BibTeX
@article{Buasri2026,
  author = {Buasri, Nutchanat and Littidej, Patiwat and Pumhirunroj, Benjamabhorn and Juanchaiyaphum, Jatuphum and Slack, Donald},
  title = {Integrated Hydro-Hazard Index (HHI) for Drought-Flood Risk Assessment: A Multi-Temporal Machine Learning Approach},
  journal = {Sustainability},
  year = {2026},
  volume = {18},
  number = {14},
  pages = {7448},
  doi = {10.3390/su18147448}
}
Kostenlos starten →