Author(s): Jun-Young Kim; Yoon-Seo Lee; Sung-Jo Kim; Seung-Oh Lee

Linked Author(s):

Keywords: Principal Component Analysis (PCA); Stage-Response Function; Urban Pluvial Flood

Abstract: Urban pluvial flood response varies significantly by local topography and drainage, yet generalized models often fail to capture these local differences. This study utilizes an in-house Python-based urban inundation model to identify key factors governing stage responses and to derive a Stage-Response Function applicable across various rainfall magnitudes. We constructed a dataset using simulations under design rainfall conditions with return periods of 30, 50, 100, and 500 years. Principal Component Analysis (PCA) was applied to extract dominant modes of spatial water level variations, which were then linked to catchment physical characteristics via regression analysis. Results indicate that the derived function successfully reproduces stage behaviors in validation areas (unlearned regions), confirming the transferability of the model. Furthermore, the inundation depth distribution, calculated by combining the derived stage with the Digital Elevation Model (DEM), shows high spatial agreement with physical model outputs. The proposed Stage-Response Function enables rapid analysis based on local attributes and input conditions, eliminating the need for repetitive calculations of complex physical models. Consequently, this approach contributes to enhancing the efficiency of urban pluvial flood assessment.

DOI:

Year: 2026