The Importance of Digital Elevation Model Selection in Flood Simulation and a Proposed Method to Reduce DEM Errors: A Case Study in Shanghai

Digital Elevation Models (DEMs) play a critical role in hydrologic and hydraulic modeling. Flood inundation mapping is highly dependent on the accuracy of DEMs. Various vertical differences exist among open access DEMs as they use various observation satellites and algorithms. The problem is particularly acute in small, flat coastal cities. Thus, it is necessary to assess the differences of the input of DEMs in flood simulation and to reduce anomalous errors of DEMs. In this study, we first conducted urban flood simulation in the Huangpu River Basin in Shanghai by using the LISFLOOD-FP hydrodynamic model and six open-access DEMs (SRTM, MERIT, CoastalDEM, GDEM, NASADEM, and AW3D30), and analyzed the differences in the results of the flood inundation simulations. Then, we processed the DEMs by using two statistically based methods and compared the results with those using the original DEMs. The results show that: (1) the flood inundation mappings using the six original DEMs are significantly different under the same simulation conditions—this indicates that only using a single DEM dataset may lead to bias of flood mapping and is not adequate for high confidence analysis of exposure and flood management; and (2) the accuracy of a DEM corrected by the Dixon criterion for predicting inundation extent is improved, in addition to reducing errors in extreme water depths—this indicates that the corrected datasets have some performance improvement in the accuracy of flood simulation. A freely available, accurate, high-resolution DEM is needed to support robust flood mapping. Flood-related researchers, practitioners, and other stakeholders should pay attention to the uncertainty caused by DEM quality.

Sequential data assimilation for real-time probabilistic flood inundation mapping

Real-time probabilistic flood inundation mapping is crucial for flood risk warning and decision-making during the emergency period before an upcoming flood event. Considering the high uncertainties involved in the modeling of a nonlinear and complex flood event, providing a deterministic flood inundation map can be erroneous and misleading for reliable and timely decision-making. The conventional flood hazard maps provided for different return periods cannot also represent the actual dynamics of flooding rivers. Therefore, a real-time modeling framework that forecasts the inundation areas before the onset of an upcoming flood is of paramount importance. Sequential data assimilation (DA) techniques are well known for real-time operation of physical models while accounting for existing uncertainties. In this study, we present a DA hydrodynamic modeling framework where multiple gauge observations are integrated into the LISFLOOD-FP model to improve its performance. This study utilizes the ensemble Kalman filter (EnKF) in a multivariate fashion for dual estimation of model state variables and parameters where the correlations among point source observations are taken into account. First, a synthetic experiment is designed to assess the performance of the proposed approach; then the method is used to simulate the Hurricane Harvey flood in 2017. Our results indicate that the multivariate assimilation of point source observations into hydrodynamic models can improve the accuracy and reliability of probabilistic flood inundation mapping by 5 %–7 %, while it also provides the basis for sequential updating and real-time flood inundation mapping.

Stage-Based Flood Inundation Mapping

New methods allow the direct computation of flood inundation maps from lidar data, independently of discharge estimates, hydraulic analysis, or defined cross sections. One method projects the interpolated profile of measured flood levels onto surrounding topography, creating a smooth inundation surface that is entirely based on data and geometrical relationships. A second method computes inundation maps for any simple function that relates the water surface to the elevation of the channel bottom, exploiting their known, sub-parallel character. A final method theoretically combines the elevation of the channel bottom and the upstream catchment area for points along the thalweg, all defined by lidar data. The conceptual simplicity and realism of these maps facilitate data-based planning.

Flood Inundation Mapping with Limited Observations Based on Physics-Aware Topography Constraint

Spatial classification with limited observations is important in geographical applications where only a subset of sensors are deployed at certain spots or partial responses are collected in field surveys. For example, in observation-based flood inundation mapping, there is a need to map the full flood extent on geographic terrains based on earth imagery that partially covers a region. Existing research mostly focuses on addressing incomplete or missing data through data cleaning and imputation or modeling missing values as hidden variables in the EM algorithm. These methods, however, assume that missing feature observations are rare and thus are ineffective in problems whereby the vast majority of feature observations are missing. To address this issue, we recently proposed a new approach that incorporates physics-aware structural constraint into the model representation. We design efficient learning and inference algorithms. This paper extends our recent approach by allowing feature values of samples in each class to follow a multi-modal distribution. Evaluations on real-world flood mapping applications show that our approach significantly outperforms baseline methods in classification accuracy, and the multi-modal extension is more robust than our early single-modal version. Computational experiments show that the proposed solution is computationally efficient on large datasets.

Machine-learning and HEC-RAS integrated models for flood inundation mapping in Baro River Basin, Ethiopia

This study presents an integrated machine-learning and HEC-RAS models for flood inundation mapping in Baro River Basin, Ethiopia. ANN and HEC-RAS models were integrated as a predictive hydrological and hydraulic model to generate runoff and the extent of flood, respectively. Daily rainfall and temperature data of 7-years (1999–2005), daily discharge (1999–2005) and 30 m × 30 m gridded Topographical Wetness Index (TWI) were used to train a predictive ANN hydrological model in RStudio. The predictive performance of the developed ANN hydrological model was evaluated in RStudio using Nash–Sutcliffe Efficiency (NSE) values of 0.86 and 0.88 during the training period (1999–2005) and testing period (2006–2008), respectively, with the corresponding observed daily discharge. The validated ANN predictive hydrological model was linked with HEC-RAS to generate the flood extent along the river course. The HEC-RAS model result was calibrated and validated using the water body delineated using Normal Difference Water Index (NDWI) from LANDSAT 8 imagery based on historical flood events of 2005 and 2008. It was found that about 96% of an agreement was made between the flood-prone areas generated in HEC-RAS and the water body delineated using NDWI. Therefore, it is logical to conclude that the integration of a machine-learning approach with the HEC-RAS model has improved the spatiotemporal uncertainties in traditional flood forecasting methods. This integrated model is powerful tool for flood inundation mapping to warn residents of this basin.