Impacts of DEM Type and Resolution on Deep Learning-Based Flood Inundation Mapping

Mohammad Fereshtehpour,Mostafa Esmaeilzadeh,Reza Saleh Alipour,Steven J. Burian

DOI: https://doi.org/10.1007/s12145-024-01239-0

2024-02-17

Abstract:The increasing availability of hydrological and physiographic spatiotemporal data has boosted machine learning's role in rapid flood mapping. Yet, data scarcity, especially high-resolution DEMs, challenges regions with limited access. This paper examines how DEM type and resolution affect flood prediction accuracy, utilizing a cutting-edge deep learning (DL) method called 1D convolutional neural network (CNN). It utilizes synthetic hydrographs as training input and water depth data obtained from LISFLOOD-FP, a 2D hydrodynamic model, as target data. This study investigates digital surface models (DSMs) and digital terrain models (DTMs) derived from a 1 m LIDAR-based DTM, with resolutions from 15 to 30 m. The methodology is applied and assessed in an established benchmark, the city of Carlisle, UK. The models' performance is then evaluated and compared against an observed flood event using RMSE, Bias, and Fit indices. Leveraging the insights gained from this region, the paper discusses the applicability of the methodology to address the challenges encountered in a data-scarce flood-prone region, exemplified by Pakistan. Results indicated that utilizing a 30 m DTM outperformed a 30 m DSM in terms of flood depth prediction accuracy by about 21% during the flood peak stage, highlighting the superior performance of DTM at lower resolutions. Increasing the resolution of DTM to 15 m resulted in a minimum 50% increase in RMSE and a 20% increase in fit index across all flood stages. The findings emphasize that while a coarser resolution DEM may impact the accuracy of machine learning models, it remains a viable option for rapid flood prediction. However, even a slight improvement in data resolution in data-scarce regions would provide significant added value, ultimately enhancing flood risk management.

Computational Engineering, Finance, and Science

What problem does this paper attempt to address?

The problem this paper attempts to address is: How do the type and resolution of Digital Elevation Models (DEM) affect the accuracy of flood inundation predictions based on deep learning?

Specifically, the paper explores the following two main aspects:
1. **The impact of DEM type and resolution on flood prediction**: By using different types of DEMs (such as Digital Surface Model DSM and Digital Terrain Model DTM) and data with different resolutions (ranging from 15 meters to 30 meters), the study investigates how these factors influence the prediction of flood depth and extent.
2. **Application in data-scarce regions**: Particularly in flood-prone areas with scarce data (such as Pakistan), how to utilize limited DEM data for rapid and accurate flood prediction.

The paper employs an advanced deep learning method—1D Convolutional Neural Network (1D CNN), and trains and evaluates the model using synthetic hydrological maps and water depth data generated by the LISFLOOD-FP hydrodynamic model. The study site is selected in Carlisle, UK, which is a commonly used benchmark area for flood risk research. The model performance under different DEM types and resolutions is evaluated using metrics such as RMSE, bias, and fit index.