*3.4. LIDAR as a Source of Information for Hydrodynamic Model Verification*

Based on previous research, LiDAR proved to be an efficient method to provide terrain data with high resolution as compared to other DEM sources [72]. According to Sampson et al. [73], LiDAR-derived DEMs are considered the most reliable DEMs for flood modeling to date. In summary, hydrological modeling studies showed that the vertical accuracy of DEMs does affect the accuracy of hydrologic predictions [70]. *3.4. LIDAR as a Source of Information for Hydrodynamic Model Verification*  Based on previous studies, it was found that LiDAR data are capable of producing high-resolution DEMs for flood simulation modeling, which can be an efficient tool in floodplain inundation management. Hence, they are commonly used for hydrodynamic model verification. Based on previous studies, it was found that LiDAR data are capable of producing high-resolution DEMs for flood simulation modeling, which can be an efficient tool in floodplain inundation management. Hence, they are commonly used for hydrodynamic model verification. Courty et al. [74] mentioned that inundation areas from LIDAR-derived DEM were the closest to reality as reported by Li and Wong [69]; therefore, they used LiDAR-derived DEM as a reference when comparing DEMs generated from Advanced Land Observing Satellite (ALOS) World 3D-30m (AW3D30), SRTM, and Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) for flood modeling purposes. Based on the flood simulation results, AW3D30 performed better than SRTM, while ASTER was the worst performer of all global DEMs.

Courty et al. [74] mentioned that inundation areas from LIDAR-derived DEM were the closest to reality as reported by Li and Wong [69]; therefore, they used LiDAR-derived DEM as a reference when comparing DEMs generated from Advanced Land Observing Satellite (ALOS) World 3D-30m (AW3D30), SRTM, and Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) for flood modeling purposes. Based on the flood simulation results, AW3D30 performed

better than SRTM, while ASTER was the worst performer of all global DEMs.

Hashemi et al. [75] also used LiDAR-derived DEMs as a reference when investigating the quality of DEMs generated from an unmanned aerial vehicle (UAV) used in flood modeling. These studies concluded that the reliability of floodplain maps is dependent on the quality of DEM. Van de Sande et al. [76] adopted LiDAR DEM data as ground truth referring to the terrain elevation. Hence, the flood risk assessment of publicly available DEMs such as ASTER and SRTM DEM was compared with flood risk based on LiDAR DEM. The inundation maps of these publicly available DEMs were smaller than inundation maps produced using LiDAR DEM. The underestimations of the flood risk influence the credibility when making appropriate decisions regarding flood risk management and mitigation. Hashemi et al. [75] also used LiDAR-derived DEMs as a reference when investigating the quality of DEMs generated from an unmanned aerial vehicle (UAV) used in flood modeling. These studies concluded that the reliability of floodplain maps is dependent on the quality of DEM. Van de Sande et al. [76] adopted LiDAR DEM data as ground truth referring to the terrain elevation. Hence, the flood risk assessment of publicly available DEMs such as ASTER and SRTM DEM was compared with flood risk based on LiDAR DEM. The inundation maps of these publicly available DEMs were smaller than inundation maps produced using LiDAR DEM. The underestimations of the flood risk influence the credibility when making appropriate decisions regarding flood risk management and mitigation.

*Remote Sens.* **2020**, *12*, x FOR PEER REVIEW 12 of 20

Furthermore, most small river basins in many countries are not characterized by high-quality DEMs such as LiDAR data [77]. Hence, aerial photographs or globally available DEMs such as ASTER and SRTM are commonly used, which leads to low accuracy of flood prediction due to the significant effect of low-accuracy DEMs. Therefore, this study proposed using corrected DEMs generated from aerial photographs as an option in flood modeling. The correction of DEM was performed based on field measurements to determine vertical errors. Then, a reference DEM that was developed from LiDAR data was used to validate the performance of the original and corrected DEM. The impact of DEM accuracy was evaluated using the flood model. The results from the model indicated that the flood prediction of corrected DEM was better than that of the original DEM when compared with the simulated result of the reference DEM, as shown in Figure 4. However, the authors suggested that the proposed method was not suitable for urban areas. Furthermore, most small river basins in many countries are not characterized by high-quality DEMs such as LiDAR data [77]. Hence, aerial photographs or globally available DEMs such as ASTER and SRTM are commonly used, which leads to low accuracy of flood prediction due to the significant effect of low-accuracy DEMs. Therefore, this study proposed using corrected DEMs generated from aerial photographs as an option in flood modeling. The correction of DEM was performed based on field measurements to determine vertical errors. Then, a reference DEM that was developed from LiDAR data was used to validate the performance of the original and corrected DEM. The impact of DEM accuracy was evaluated using the flood model. The results from the model indicated that the flood prediction of corrected DEM was better than that of the original DEM when compared with the simulated result of the reference DEM, as shown in Figure 4. However, the authors suggested that the proposed method was not suitable for urban areas.

**Figure 4.** Comparison of flood extent and flood depth obtained from flood model simulation: (**a**) **Figure 4.** Comparison of flood extent and flood depth obtained from flood model simulation: (**a**) original DEM; (**b**) corrected DEM; (**c**) reference DEM [77].

#### original DEM; (**b**) corrected DEM; (**c**) reference DEM [77]. *3.5. LiDAR DEM for Flood Hazard and Flood Risk Mapping*

used for flood planning purposes.

Open

*3.5. LiDAR DEM for Flood Hazard and Flood Risk Mapping*  Flood risk assessment and management rely on the accuracy of flood extent simulated using a flood model. Most flood risk mapping is based on a conceptual risk approach that uses DEMs to predict the flood hazard according to the projected water levels and to indicate the vulnerability of areas to flood events with damage to properties and livelihood. Hazard mapping is an important Flood risk assessment and management rely on the accuracy of flood extent simulated using a flood model. Most flood risk mapping is based on a conceptual risk approach that uses DEMs to predict the flood hazard according to the projected water levels and to indicate the vulnerability of areas to flood events with damage to properties and livelihood. Hazard mapping is an important element in assessing risk and designing mitigation measures for flood-prone areas.

element in assessing risk and designing mitigation measures for flood-prone areas. Flood hazard is usually generated based on the outcome of hydrological models that simulate the water movement across the floodplain like flood extent, water velocity, or water depth [11,37,78]. In addition, flood hazards can also be produced using a statistical or machine-learning approach integrated with GIS technology by using fluvial stage records and topographic data [79,80]. Flood hazard and flood risk maps indicate the flood-prone area with possible destructive impact, which is Flood hazard is usually generated based on the outcome of hydrological models that simulate the water movement across the floodplain like flood extent, water velocity, or water depth [11,37,78]. In addition, flood hazards can also be produced using a statistical or machine-learning approach integrated with GIS technology by using fluvial stage records and topographic data [79,80]. Flood hazard and flood risk maps indicate the flood-prone area with possible destructive impact, which is used for flood planning purposes.

For instance, existing digital mapping was not sufficient enough to provide a high accuracy of flood risk maps for Annapolis Royal, Nova Scotia, Canada, an area that is vulnerable to coastal flooding [81]. Hence, the need for a high-resolution DEM was studied to produce accurate

For instance, existing digital mapping was not sufficient enough to provide a high accuracy of flood risk maps for Annapolis Royal, Nova Scotia, Canada, an area that is vulnerable to coastal flooding [81]. Hence, the need for a high-resolution DEM was studied to produce accurate inundation maps based on sea level and climate change. As the sea level rises, water inundates the nearby lands; thus, it is important to define the extent of the flood inundation. The predicted results were compared with the benchmark of a past storm event to test the model. Based on the prediction results, mitigation structures such as dykes could be suggested if coastal development is planned to take place in any of the risk areas.

Puno et al. [13] conducted flood simulations at different return periods with LiDAR-derived DEMs as a primary source of elevation data in the hydrologic model. The model was calibrated by comparing the predicted flood simulation with a real flood event in 2016. Flood hazard maps were generated from the simulated flood events using GIS and LiDAR-derived DEM. The generated maps were validated through an interview with the affected localities. The authors found that using LiDAR data in the hydrologic model could produce high-resolution flood hazard maps that can offer more accurate decisions and actions in disaster management and mitigation.

Ogania et al. [14] evaluated the effect of DEM resolutions on generating flood hazard maps using hydraulic modeling software for disaster preparedness and mitigation. This study presented the performance of three different DEM resolutions, which were LiDAR, IfSAR, and SAR DEMs in flood modeling studies. The accuracy of each generated flood map was evaluated using a confusion matrix approach by comparing the generated maps with the actual flood data. This paper revealed that LiDAR-derived DEMs provide a more defined flood extent and clear distribution of flood hazards. Furthermore, they offer more accurate flood maps compared to other DEM data sources, which aligned with the findings from previous researchers such as Hailes and Rientjes [82] and Schumann et al. [65].

Mihu-Pintilie et al. [83] used high-density LiDAR data with 2D hydraulic modeling to improve urban flood hazard maps. This study simulated four different multi-scenarios at different discharge values. Because LiDAR data provide a precise representation of the hydraulic conditions such as channels and roads, the combination of 2D hydraulic and LiDAR DEMs produced accurate information regarding flood hazard vulnerability. Flood hazard maps were generated based on flood depth classification according to the Japanese criteria of the Ministry of Land Infrastructure and Transport (MLIT). The criteria suggested five hazard classes of very low, low, medium, high, and extreme classified as H1, H2, H3, H4, and H5, respectively. Figure 5 shows that all hazard classes were encountered according to scenario 1 (s1). However, most of the affected areas were assigned with the very low or low class of hazard (H1 and H2).

LiDAR datasets were implemented in a new procedure of flood hazard estimation proposed by Guerriero et al. [84]. The authors developed algorithms of interpolation of multiple probability models of hydrometric time-series data combined with topography derived from LiDAR data for the production of flood hazard maps. Flood hazard maps produced from this method were compared with a flood event observation in 2015 for validation. This suggested method can be considered as another option for hydraulic simulations to provide flood hazard analysis.

In conclusion, high-resolution DEMs have great influence on producing accurate and reliable maps in the field of flood simulations. Using these maps helps in disaster risk reduction and management, especially in identifying specific areasthat need to be prioritized for providing appropriate flood risk management measures to be taken to combat flood disaster. Previous studies implied that LiDAR-derived DEMs improve the accuracy of flood parameters; hence, they can help in producing high-quality flood hazard and flood risk mapping.

**Figure 5.** Flood hazard map based on flood depth classification according to the Ministry of Land Infrastructure and Transport (MLIT) [83].

#### **4. Challenges and Future Perspectives**

The frequency of flood disasters all over the world is increasing due to climate change and rapid urbanization. Future climate projections could provide an additional understanding of extreme climate changes, including the risk of flood events [85]. Furthermore, studies on flood mapping

and monitoring increased with the advancement of current technologies to reduce the impact of flood disasters. LiDAR data acquisitions seem to be a promising approach to solve the problems associated with the inadequate representation of topographic data. Both airborne and terrestrial LiDAR systems are active imaging techniques operating with light that allow the systems to collect data during daylight or nighttime. Previous studies revealed that LiDAR technology has many advantages, which makes it suitable for flood modeling, particularly in flat areas and complex urban environments. Depending on the spatial scale, LiDAR data offer different advantages for accurate terrain mapping compared to other sources. Moreover, LiDAR could be advantageous to provide information in small-scale flood risk management by having small important topographic features such as dykes, ditches, and levees [15,51,86,87]. Furthermore, the integration of LiDAR technology with any remotely sensed products may be used to increase the effectiveness of this technology, especially in flood modeling.

However, there are some restrictions in using LiDAR-derived DEM in the context of flood applications. The main drawback of both LiDAR systems is the process of classifying ground from non-ground data for DEM generation, which is needed in simulations of the flood model. Ground surface information is not easily extracted, especially in areas with complex terrain surface and features such as buildings and vegetation [88]. The ground filtering process proves to be a challenging task as it can affect the accuracy of the LiDAR products [8,49,89]. Several filtering algorithms were developed by previous researchers to process LiDAR data. However, the LiDAR data must be correctly processed because they could influence the outcomes of flood mapping [69]. The algorithms perform differently depending on the specific surface conditions. This means that not all algorithms are competent in producing high-quality LiDAR-derived DEM data [90]. A filtering algorithm should be selected based on its ability to produce the desired result [91]. Common filtering algorithms used in LiDAR data processing include elevation threshold with expand window (ETEW), maximum local slope, adaptive triangulated irregular network (TIN), and progressive morphology [90,92,93]. Filtering problems are expected to be better solved with the evolution of machine learning [88].

In addition, the sensitive response of flood inundation to small changes in topography representation gives rise to several challenges [21]. Collecting small-scale features needs a high resolution of DEM data, but the data are rarely available, especially for developing countries. Not all countries can afford to use LiDAR data due to economic constraints. The high cost and the difficulty of processing huge LiDAR datasets could be the main reason why LiDAR data are not used in some developing countries. Even developed countries like the United States and the United Kingdom do not have LiDAR data available for the entire country. Another challenge when using LiDAR data is the need for huge data storage due to the high-point-density data. High-point-density data need a longer computational time to process [94,95]. Between airborne and terrestrial LiDAR systems, the time required for flood model simulations using terrestrial LiDAR is 10 times longer than that required for airborne LiDAR [96].

Furthermore, even though high-resolution DEMs offer detailed information topography, they take a longer time to process or analyze the data. Abucay and Tseng [97] carried out a visibility analysis that could be used in identifying flood-prone areas using various DEM sources. The authors reported that the LiDAR-derived DEM required 28 min to complete the visibility analysis, followed by the SAR DEM that took 19 s, while ALOS and ASTER GDEM both required only 3 s to complete the process. Nevertheless, the computational time problem may be solved with future advancements in computer technology. Moreover, the LiDAR system cannot penetrate water bodies as its laser beam is absorbed by the water. Therefore, the inaccurate elevation measurement of water-covered areas influences cross-section attributes, leading to inaccuracies in hydrodynamic simulations [98].

#### **5. Conclusions**

Detailed topographic information is a crucial input parameter for flood modeling and monitoring. The performance of flood modeling is highly dependent on the DEM accuracy [10], especially in

small-scale flood modeling studies. Flood model simulation results show differences in water depth and inundation when using detailed DEMs, proving that DEM accuracy has a significant impact on flood hazard estimation [21,41]. Therefore, the need for high-resolution DEM explains the interest in exploring new technology to generate detailed elevation data. In this review, the promising applications in numerous flood studies demonstrate that the LiDAR system is capable of offering high-density and high-resolution DEM data to improve the flood model input, thus resulting in a higher accuracy of flood modeling results. However, LiDAR data also face several difficulties that need to be addressed in the future regarding the filtering process for DEM generation and enormous point density data that need huge data storage, resulting in a longer computational time to simulate flood models. Additionally, integration between terrestrial and airborne LiDAR or any remotely sensed products seems to be a promising approach to solve the problems associated with the inadequate representation of topographic data in topographically complex areas [99]; hence, more investigation and research work for the expansion of LiDAR systems can be foreseen in upcoming applications of flood detection and monitoring.

**Author Contributions:** N.A.M. executed the manuscript writing, coordinated the paper revisions, and contributed to the workflow implementation. A.F.A. contributed to the workflow implementation, as well as the manuscript compilation and revisions. S.K.B. proposed the research idea. M.R.M. and A.M. supervised the final manuscript. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by Universiti Putra Malaysia, grant number UPM/800-3/3/1/GPB/2019/9678700.

**Acknowledgments:** The authors wish to acknowledge the assistance of the Department of Biological and Agricultural Engineering, Faculty of Engineering, Universiti Putra Malaysia for supplying the facilities for this study. The authors would also like to appreciate the support for this study from the Institute of Aquaculture and Aquatic Sciences.

**Conflicts of Interest:** The authors declare no conflict of interest.

## **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
