Probabilistic flood inundation mapping of ungauged rivers: Linking GIS techniques and frequency analysis

SummaryThis study presents an exhaustive methodology of floodplain mapping at ungauged rivers. To present our methodology, we selected the Halilrud basin and Jiroft city in southeastern Iran as an example of hazardous regions. To estimate flood quantiles in different return periods at ungauged reaches, we used regional flood frequency analysis. By using the well-known L-moments approach and related criteria, a homogeneous region was formed and the 3-parameter Log normal distribution was identified as the robust regional frequency distribution. The hydro-geomorphic characteristics and the land use properties of the catchments were then extracted using RS&GIS techniques to establish multivariate regional regression models between hydro-geomorphic characteristics and flood quantiles. After delineation of the catchments for the ungauged reaches, flood quantiles as an important factor in flood inundation at outlets of these reaches with different probabilities were estimated using the regional regression models. To delineate flood hazard maps and to enhance the accuracy of the hydraulic modeling, we applied satellite stereoscope images of Cartosat-1 along with the Rational Polynomial Coefficients to extract a high resolution DTM and detailed parameterization of the channel required by 1D hydraulic model HEC-RAS. The GIS-based HEC-Geo RAS pre- and post-processor were also used for careful optimization of the geometry features for real visualization of the flood prone areas. Information of some historical flood events was also used to evaluate the hydraulic model performance in predicting flood inundations. Finally, vulnerable areas were crossed with extracted land use mapping from IRS-P6 satellite images to differentiate the critical infrastructures and the valuable land use classes affected by floods in different return periods.

► We select the robust regional distribution function using an L-moment approach.
► To estimate flood quantiles at ungauged sites regional regression models are used.
► We introduce new techniques for channel geometry using Cartosat-1 images.
► New satellite data-set enhances the accuracy of the hydraulic modeling.
► Reliable flooding prone areas are delineated to decrease probable economic losses.