A novel hybrid approach based on cellular automata and a digital elevation model for rapid flood assessment

Obaja Triputera Wijaya, Tsun Hua Yang*

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

Abstract

An efficient inundation model is necessary for emergency flood responses during storm events. Cellular automata (CA)-based flood models have been proven to produce rapid results while maintaining a certain degree of accuracy. However, the need for computational resources dramati-cally increases when the number of grid cells increases. Digital elevation model (DEM)-based models generate results even faster, but the simplified governing equations within the models fail to reflect temporal flood evolution. To achieve rapid flood modeling while maintaining model sim-plicity, a novel two-dimensional hybrid inundation model (HIM) was developed by combining the CA-and DEM-based concepts. Given the temporal flood evolution generated by the CA concept, final finer-scale predictions were obtained by applying the DEM-based concept. The performance of this model was compared to those of widely used, physically based hydraulic models using three UK Environment Agency (EA) benchmark test cases. The HIM yielded consistent prediction results but was faster than the CA-based model. Finally, a comparison was made against flood observa-tions, and the overall root mean squared error (RMSE) for flood depth was 0.388–0.400 m. Consid-ering the uncertainty in the observed flood depths, the HIM shows promising potential to serve as an intermediate tool for emergency response in practical cases.

Original languageEnglish
Article number1311
JournalWater (Switzerland)
Volume13
Issue number9
DOIs
StatePublished - 1 May 2021

Keywords

  • Cellular automata
  • DEM-based
  • Hybrid inundation model
  • Urban flood

Fingerprint Dive into the research topics of 'A novel hybrid approach based on cellular automata and a digital elevation model for rapid flood assessment'. Together they form a unique fingerprint.

Cite this