Building and calibrating a 3Di subgrid flood model for Hanoi, without a surveyed terrain model, drainage data, or rain gauges
Abstract
Urban hydrodynamic flood models typically rely on a surveyed terrain model, an explicit pipe drainage network, and rain gauge measurements, none of which were available for the inundation following Typhoon Matmo in My Dinh and Tay Mo, Hanoi, on 7 October 2025. The present study addresses this issue by documenting the construction and calibration of a 32.8 km^2 3Di subgrid model under severe data constraints, demonstrating how unmonitored catchments can be simulated using openly accessible sources. Specifically, a 1 m digital elevation model was systematically repaired for river and lake bathymetry, unsurveyed water bodies, road underpasses, and an excavation pit to re-establish essential overland flow paths. In addition, assigned infiltration rates across land-cover classes were utilized to stand in for the unrecorded subsurface sewer system, while half-hourly satellite precipitation from the IMERG Late product provided rainfall forcing for a closed two-dimensional domain. Model calibration was conducted against eight crowdsourced depth observations. The simulation results demonstrated that six observation points retained surface water and were closely aligned with the reported values within 2.2 cm. However, two peripheral locations along Tay Mo Road and Road 70 remained completely dry in the model. Because all eight points were utilized during parameter adjustment and no independent validation dataset was withheld, this correspondence reflects a calibration fit rather than an unconstrained predictive test.
Related
- Calibrating subgrid parametrizations of single-column ocean models via simulation-based inference
- Improving low-cost NO₂ sensor reliability under real-world air quality conditions by calibrating distributions instead of data points
- Online learning of subgrid-scale models for quasi-geostrophic turbulence in planetary interiors
- Subgrid parametrizations mediate the large‐scale effects of soil macroporosity on the water cycle in the ORCHIDEE land surface model
- Advancing the microphysical and subgrid treatment of precipitation in the LMDZ AGCM: parametrization development and first applications in polar contexts
BibTeX
@misc{nguyen2026building,
title = {{Building and calibrating a 3Di subgrid flood model for Hanoi, without a surveyed terrain model, drainage data, or rain gauges}},
author = {Minh Hoa Nguyen and Mai Anh Tran},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.20789},
eprint = {2610.20789},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.20789},
note = {CC BY-NC-SA 4.0},
}
RIS
TY - UNPB
TI - Building and calibrating a 3Di subgrid flood model for Hanoi, without a surveyed terrain model, drainage data, or rain gauges
AU - Minh Hoa Nguyen
AU - Mai Anh Tran
PY - 2026
DA - 2026/10/14
PB - arXiv
DO - 10.48550/arxiv.2610.20789
UR - https://arxiv.org/abs/2610.20789
AB - Urban hydrodynamic flood models typically rely on a surveyed terrain model, an explicit pipe drainage network, and rain gauge measurements, none of which were available for the inundation following Typhoon Matmo in My Dinh and Tay Mo, Hanoi, on 7 October 2025. The present study addresses this issue by documenting the construction and calibration of a 32.8 km^2 3Di subgrid model under severe data constraints, demonstrating how unmonitored catchments can be simulated using openly accessible sources. Specifically, a 1 m digital elevation model was systematically repaired for river and lake bathymetry, unsurveyed water bodies, road underpasses, and an excavation pit to re-establish essential overland flow paths. In addition, assigned infiltration rates across land-cover classes were utilized to stand in for the unrecorded subsurface sewer system, while half-hourly satellite precipitation from the IMERG Late product provided rainfall forcing for a closed two-dimensional domain. Model calibration was conducted against eight crowdsourced depth observations. The simulation results demonstrated that six observation points retained surface water and were closely aligned with the reported values within 2.2 cm. However, two peripheral locations along Tay Mo Road and Road 70 remained completely dry in the model. Because all eight points were utilized during parameter adjustment and no independent validation dataset was withheld, this correspondence reflects a calibration fit rather than an unconstrained predictive test.
ER -