Flood vulnerability assessment at the local scale using remote sensing and GIS techniques: a case study in Da Nang City, Vietnam

Flooding has been recognized as one of the most devastating calamities among climate-induced natural disasters across many regions over the world. Under the context of recent climate change with occurrences of heavy rain and different climate stressors, floods have caused overwhelming loss for human society throughout history. Understanding of the flood impact in relation to climate change and unpredictable catastrophes has become an urgent need in many lowland regions worldwide. Various studies have attempted to identify flood-prone areas and evaluate the flood hazard potential on different scales from locally to regionally (Forkuo 2011; Ho & Umitsu 2011; De Risi et al. 2014; Do & Nagasawa 2014; Kazakis et al. 2015; Abdelkarim et al. 2020). However, fewer studies have comprehensively assessed the flood hazard in a vulnerable context which is more relevant to the community. The flood itself might not be a hazard since it sometimes can bring benefits to surrounding regions (FAO 2011) or less damage if the local community is well-resilient to flood. Assessing flood vulnerability has become a vital need in disaster risk management and developing strategies to mitigate the damage caused by floods.

There is still a significant research gap between flood hazard zonation and flood vulnerability assessment. While previous studies on flood hazard zonation have primarily focused on the physical vulnerability of floods and based on a pixel-based approach, research on flood vulnerability assessment focuses on the physical, social, and economic vulnerabilities as well, but the assessment is primarily based on an administrative scale, which is not always effective. Several methods for flood hazard assessment have been developed using various hydrological, meteorological, and geomorphological data (e.g., Ballais et al. 2005; Forkuo 2011; Manfreda et al. 2011; Kwak 2017). Some authors also tried to characterize the flood hazard potential using remote sensing and Geographical Information System (GIS) data. Kazakis et al. (2015) have developed an index-based approach for regional-scale flood hazard assessment in the Rhodope–Evros region, Greece, using multiple physical indicators extracted from remote sensing and GIS data including elevation, flow accumulation, slope, rainfall, land use, geology, and distance from the drainage network. This study, however, did not integrate the socio-economic data for flood hazard assessment, except the land use. Ho & Umitsu (2011) has developed a flood hazard map for Thu Bon alluvial plain, Vietnam, based on a landform classification utilizing the Shuttle Radar Topographic Mission (SRTM) and LANDSAT ETM remote sensing data. Do & Nagasawa (2014) have generated the potential flood hazard map for the Hoa Chau commune in the Hoa Vang district, Da Nang City, Vietnam based on the integration of inundation and flow direction maps using the ALOS PALSAR remote-sensed and ASTER GDEM elevation data. Nguyen et al. (2021) have developed an observational network for a real-time flood warning system in the Vu Gia-Thu Bon river basin based on an automatic hydro-meteorological approach. Although these studies have almost accurately identified the potential flood hazard areas, those studies mostly focused on evaluating the flood hazard exposure or partly combined with the social condition, mainly the land-use data with the lack of consideration of other aspects in the flood risk evaluation, especially the resilience and adaptive capacity (AC) of the community.

This study aims to identify the flood vulnerable areas at a local scale according to the different susceptibility degrees under the impacts of the natural and socio-economic conditions. Our primary approach integrates GIS and satellite image processing combined with field survey data verification on an index-based approach. Such a process to assess the impact of natural disasters is considered as an effective way with the advantage of quick and accurate updating temporal and spatial changes of natural disasters. Specifically, we calculate the flood vulnerability index (FVI) from each component of the vulnerability function to assess flood impact quantitatively. This index is elaborated by the Analytical Hierarchical Process (AHP) (Saaty 2008) to determine the weights for component variables, thereby determining the FVI for each area. The output enables identifying areas with different flood vulnerability levels from low, moderate to high, and very high. As such, this study provides a baseline for proposing a number of adaptive solutions for areas under high flood vulnerability. The proposed flood vulnerability map is assessed by field survey data and the statistical report on flood damage that previous studies have never implemented. Our FVI toolkit is useful for assessing the urban flood vulnerability and facilitating the adaptation and coping capacities of the local communities.

Vietnam is reported as one of the regions most affected by climate change and natural disasters (Eckstein et al. 2020). The areas in Central Vietnam, especially Da Nang and Quang Nam provinces located downstream of the Vu Gia – Thu Bon River system, have experienced several severe floods and are often inundated in the rainy season. Climate-induced flooding occurs almost annually in Da Nang City, especially under the impact of recent climate change, and the flood damage has become more exacerbated. We set a case study area in Da Nang City that has faced several historical floods causing huge damage to economic and urban life. A local government statistical report informed that from 1998 to 2020, flooding in Da Nang killed 83 people and injured 100 people, 15,633 households were affected, and 130,861 houses were destroyed leading to more than 2,600 billion Dongs economic loss (Figure 3). Flooding is the most devastating natural disaster in the history of Da Nang City. Assessing the socio-economic vulnerability of floods is vital in managing natural disaster risks and developing strategies to mitigate the damage caused by floods for Danang City. Previous studies only evaluated the vulnerability of Da Nang City based on quantitative analysis using statistical data on the damage caused by climate change and natural disasters on a provincial scale (ISET International and Da Nang City Government 2009; ISET International and Da Nang Climate Change Coordination Office (CCCO) 2017). This approach was costly, time-intensive, and labor-intensive by the statistical measurement and social surveys. In this sense, we develop a comprehensive flood vulnerability assessment method using GIS and remote sensing applications as an effective and cost-benefit approach that is easily applicable to developing countries such as Vietnam.

Based on the vulnerability assessment approach suggested by IPCC (2014) and UNESCO-IHE, this study has attentively focused on selecting criteria for flood vulnerability assessment in Da Nang City. We have selected 20 indices for FVI calculation through the evaluation of the physical conditions, a review of previous research related to flood vulnerability (Ballais et al. 2005; Manfreda et al. 2011, Balica et al. 2012, Tran et al. 2017), and a questionnaire survey to scientific experts on flood studies and local stakeholders (Appendix 6). The exposure component includes seven criteria, in which the flood depth, elevation of the topography, and distance to the river channels are the main causative factors. Regarding the flood sensitivity, this study has chosen six criteria including local income, poverty index, population density, land use, tourism hotspot, and vulnerable population group including children under 5 years and elderly people over 60 years. The indicators were selected based on the analysis of their relationship to the economic condition of the study area and the potential to be affected by floods as well as natural disasters. AC is measured by the ability of a system to cope with or overcome the difficulties caused by the flood and other disasters. Seven indices have been selected for measuring the AC, including the Normalized Difference Vegetation Index (NDVI), and the density of the community service facilities such as healthcare centers, food markets, Base Transceiver Stations (BTS), road density, drainage system, and education index.

The open geospatial data sources have been used in this study. Freely accessible SRTM digital elevation model (DEM) version 4 with the spatial resolution of 30 m (https://earthexplorer.usgs.gov) and the Landsat 8 OLI satellite images with 30-m spatial resolution provided by USGS (https://earthexplore.usgs.gov) were used to extract the physical indicators related to flood exposure such as elevation, slope, flow accumulation, Topographic Wetness Index (TWI), and Modified Normalized Difference Water Index (MNDWI). The statistical data on the socio-economic condition of Da Nang City provided by the Statistical Yearbook of Da Nang City in 2020 were used for evaluating the flood sensitivity and AC. We obtained information on the community service such as the number of tourism hotspots, healthcare facilities, food market centers, and BTS via the Open Data Portal (https://congdulieu.vn) built by the Da Nang City Government. These are very important variables in the determination of flood sensitivity as well as AC. To evaluate the accuracy of the flood vulnerability map, the field survey on historical flood signs and statistical data on flood damage reported by the local government are also used.

The most important task in an index-based vulnerability assessment is to select appropriate indicators. The previous practices (Balica et al. 2012; Nasiri et al. 2016; Van et al. 2022) have tried to assemble a list of indicators using criteria such as suitability, following a conceptual framework considering data availability and sensitivity to formats, usefulness, and ease of the collection of data. Understanding each concept and considering the relationship of certain indicators to the FVI may help characterize the vulnerability of different systems, by which actions can be identified to decrease it (Balica et al. 2012).

Elevation and distance to the river channel are considered the second most important criteria in flood exposure in alignment with relevant studies (Ballais et al. 2005; Manfreda et al. 2011; Kazakis et al. 2015; Tran et al. 2021). Flood is defined as the inundation of the land surface and hence it closely depends on the topography of the surface. Flood frequency increases with decreasing elevation, meaning that lower elevations are more susceptible to flooding (Choubin et al. 2019). In this study, using SRTM DEM (Figure 7(a)), the topographic indices including slope (Figure 7(b)), TWI, and flow accumulation (Figure 7(c)) are extracted. In addition to elevation, the distance from the river channel (DIST) plays a vital role in flood hazard since floods are caused mainly by water overflows from rivers. Areas near the river channels are at a higher exposure to flooding hazards (Tran et al. 2017, 2021). Due to its important role, DIST has been assigned as a major criterion for flood exposure. The distance from the river channel in this study was determined by the Euclidean distance method, which is measured by the nearest distance between points to the river channel (Zhang et al. 2019). The calculation of DIST was conducted using the module of r.grow.distance in the GRASS GIS software with the Euclidean method (Figure 7(e)). Distance from the drainage network and elevation are assigned equal importance since flooded areas are often located at low elevations and near the drainage network (Kazakis et al. 2015).

These bands are selected to maximize the reflectance of water features by using green light wavelengths and minimize the low reflectance of SWIR by water features by taking advantage of the high reflectance of vegetation and soil features in the SWIR band (Hasan et al. 2013). The generation of the MNDWI in this study is based on Landsat 8 OLI data from 2016 to 2021 using the formula mentioned in Xu (2005) via the Google Earth Engine which is a web-based computing platform (https://code.earthengine.google.com).

Vulnerability encompasses a variety of concepts and elements including sensitivity or susceptibility to harm and the lack of capacity to cope and adapt (IPCC 2014). Based on the definition of sensitivity as well as considering the driving factors on flood hazard in the study area, this study has selected six criteria for the evaluation of flood sensitivity including average income, population density, poverty index, land use, tourism hotspot, and the ratio of vulnerable population group. Income was identified to have a close relationship to flood sensitivity since the high-income household can self-supply its demand and adapt well during floods, and thereby is less sensitive to flood vulnerability. Land use and population are widely known as principal factors for flood hazards. The population index is calculated by the number of people in an area unit (population density) and normalized on a standard scale. In addition, the poverty index, which was extracted from the statistical data on poor and near-poor households, was also evaluated as a component of flood sensitivity since it reflects the severity of a community under the impact of disasters. Both population and poverty data were collected from the statistical yearbook of Da Nang City (Da Nang GSO 2020).

Land use influences the infiltration rate and the inter-relationship between surface and groundwater (Kazakis et al. 2015). The land-use map in 2020, provided by the Department of Natural Resource and Environment (DONRE), Da Nang City, has been utilized to generate the land-use index in flood sensitivity. Based on the analysis of the characteristics of each land-use type, including the settlement, agriculture, tourism, industrial and other infrastructure types, bare land, public spaces, and permanent water, in relation to flood susceptibility, and referring to previous studies (Mishra 2013; Kazakis et al. 2015; Tehrany et al. 2015; SEPA 2018), we have assigned the values for land-use index ranging from 0 to 1. The settlement land-use types including urban and rural residential areas were recognized as the most vulnerable land use when floods occurred and hence were given the highest vulnerable value of 1. Following the settlement, agriculture and tourism land-use types were considered the highly vulnerable areas to flooding hazards and thus were assigned a value of 0.8. Subsequently, the industrial land and other infrastructure land use including the traffic area were given a moderate value of 0.6. Bare land was considered less vulnerable than other land-use types since it has less impact on human life and hence was assigned a value of 0.4. Public areas such as parks, squares, or monuments along with the forest type were arranged in the low vulnerable class in the land-use map with the value of 0.2. Finally, permanent water was assigned as a value of 0 since it has no flood susceptibility.

Da Nang City is considered one of the famous tourist destinations in the Coastal Central of Vietnam. Recently, Da Nang City has experienced an increasing occurrence of natural disasters including floods reported by the city government (Da Nang Steering Committees for Disaster Prevention and Search and Rescue 2020). Therefore, it is very necessary to assess flood vulnerability to the tourism of Da Nang City toward the sustainable tourism development. This study has selected an indicator on the tourism hotspots that can represent the potential impact of flood hazards on tourism which is the main economic activity in Da Nang City (Da Nang GSO 2020). This index was generated by taking the number of tourism hotspots including the hotels, resorts, travel agencies, and sightseeing places via the density analysis in the GIS to obtain the average values for each area unit in square kilometers. Information on the tourism hotspots in Da Nang City was collected from the Da Nang Open Data portal (https://congdulieu.vn) and then standardized in a GIS format and the density was calculated using the spatial analysis tool in GRASS GIS (r.stats.zonal).

It is necessary to consider the most vulnerable group in the population to evaluate flood susceptibility. Children and the elderly are more vulnerable to flood hazards since they have a low ability to self-evacuate and self-care compared to other groups experiencing negative impacts from disasters (Mason et al. 2021). This study has calculated the ratio of children under 5 years and people over 60 years in the total population as a vulnerable group indicator for the determination of flood sensitivity. The population data were also taken from the Statistical Yearbook of Da Nang City in 2020 at the commune level (Da Nang GSO 2020).

The AC is related to social conditions and strategies responding to flooding. AC is assessed in terms of the ability to remain and cope with hazards (Turner et al. 2003). The AC can be understood as the flood resilience or coping capacity of a system or a community to mitigate threats and damages of floods. Among three components, AC often has a negative relation with vulnerability (Smit & Wandel 2006).

In this study, seven indicators, namely, the NDVI, density of the community service facilities such as healthcare centers, food market, BTS, road density, drainage system, and an index related to the education level have been investigated to determine the AC.

The NDVI was calculated from time-series Landsat 8 OLI data from 2016 to 2021 using Google Earth Engine (https://code.earthengine.google.com) (Figure 7(f)).

The road density is calculated by the total length of the road network in a unit area (km²), which is obtained via the Open Street Map (OSM) (https://openstreetmap.org). Road density is a relatively important indicator of AC since it reflects the government's investment in enhancing the transportation infrastructure with respect to the citizen's demand supplementation, and represents local resilience to evacuate during flood, and even capacity to deliver necessary items and supporting post-flood recovery efforts.

Education is an indicator that reflects the capacity of communities to respond to floods. A community with a higher population of well-educated people will increase its ability to adapt and respond to natural disasters (Hoffmann & Blecha 2020). This study has taken the ratio of teachers in each thousands of individuals as an indicator for education in Da Nang City.

Other parameters reflecting resilience to flood hazards have been utilized in this study including the density of food markets, healthcare facilities, BTS, and drainage systems. These data were calculated using the spatial data analysis with the help of the Da Nang City Open Data Portal (https://congdulieu.vn) and the Da Nang Department of Natural Resource and Environment (DONRE). The density food markets, healthcare facilities, and BTS represent the accessibility of a local community to the primary service for food, healthcare, and communication. These indicators were calculated by density analysis in the GIS environment as applied to the tourism hotspot.

As suggested by local stakeholders and experts, the drainage system is considered the most important parameter in flood resilience. This index is measured by taking the total length of the drainage convert extracted from the drainage map provided by the Da Nang Department of Natural Resource and Environment (DONRE). Subsequently, the length density is calculated by the ratio of drainage convert length in a unit area (km/km²). Due to the direct relationship of this index on the drainage capacity of the study area, the drainage system has been assigned the highest important value compared to other criteria in the AC component.

The details of all input criteria used for flood vulnerability assessment and the method of generation as well as supported processing tools are explained in Table 1. In this study, we used the Google Earth Engine to develop the indices which require the calculation from time-series remote sensing data such as the MNDWI and the NDVI. The Google Earth Engine provides an interactive platform for geospatial processing at the scale, which is powered by the Google Cloud Platform. By coding the algorithm in GEE, with the support of automatically connecting to remote sensing databases such as Landsat or any other system via cloud service, we quickly obtain the output indices from time-series satellite data. The GRASS GIS version 7.8 (https://grass.osgeo.org) and QGIS version 3.16 (https://grass.osgeo.org) Open Source Software were used to generate different indices including slope, flow accumulation, DIST, TWI, and visualization of the resulting maps. In this study, QGIS and GRASS GIS have been used extensively to update the spatial and attribute data, normalize, and calculate most of the indices for flood vulnerability assessment.

The results of weighting calculation using the AHP method for flood vulnerability assessment are shown in Table 2. The CR for the analysis of exposure, sensitivity, and AC is always less than 0.1, satisfying the condition of AHP validation and revealing that this weighting scheme could be applied reasonably for this case study. Table 2 indicates that among the criteria, the flood depth is the major component affecting the flood exposure. Meanwhile, land use is the most important contributing factor in flood sensitivity with the highest weight of 0.31. Subsequently, among the indicators related to AC, the drainage system occupies the highest weight (0.30) and therefore mostly affects the AC. For a better preparedness for urban flood vulnerability, it is evident that the local government needs to not only focus on building warning systems in the highly flood inundated area but also carefully master plan for any urban land-use development projects and enhance the capacity of the drainage system. In addition, it is necessary to pay attention to other contributing factors such as elevation, flow accumulation, population density, tourism hotspots, vulnerable population groups, road density, and NDVI that relatively affect the urban flood vulnerability in Da Nang City.

Results of the FVI calculation for Da Nang City show that more than 20% of the study area has high and very high vulnerable levels to flood. Considering areas from moderate to high and very high vulnerabilities, it can be seen from Table 3 that the city has more than 60% of the area under the FVI levels from moderate and higher. This is a significant number requiring the city to have an urgent response to flood risk vulnerability, and the local inhabitants need to be aware of this vulnerability.

Among the 56 communes and wards of Da Nang City, 11 are under high and very high levels of the FVI as shown in Table 5. It can be seen from Table 5 that there are seven communes that are extremely vulnerable communes in Da Nang belonging to the Hoa Vang district. Here, communes of Hoa Tien, Hoa Chau, Hoa Phuoc, and Hoa Phong have more than 40% coverage of high and very high vulnerability levels. These areas require robust strategies to respond to and mitigate the impact of flood hazards in the future. Moreover, from Figure 9 and Table 5, it is clear that flood vulnerability assessment should be localized and not always based on the administrative units since a commune may have different FVI levels. For instance, the Hoa Khuong commune has all kinds of vulnerability levels from very low to very high (Figure 9), and the coverage of the area under high and very high FVI in this commune is only 12.0% (Table 5). However, the actual areas of high and very high FVI in Hoa Khuong are significant compared to other communes in Da Nang City as shown in Table 5 (612.6 ha). The high and very high FVI areas in Hoa Khuong, along with Hoa Phong and Hoa Tien communes (Figure 9) that are the major agriculture hotspots of Da Nang City and the home of many villages, are annually affected by flood damage as seen by the number of households affected in Table 7. There is an urgent need for flood responsibilities and adaptation plans for this area, and the proposed solutions should be inter-regional considering the most vulnerable area. It is unfair if the governmental investment budgets or the deployment of flood control projects are distributed based on the average FVI by districts or communes. Instead, it is more important to consider the most vulnerable areas and apply an inter-regional solution.

Table 5

Statistics of flood vulnerability for some typical communes in Da Nang City

View Large

The accuracy assessment of the vulnerability map is still challenging as seen in many studies till now. There has been no generalized workflow that could conclusively validate the flood vulnerability mapping results as yet. Both IPCC and UNESCO-IHE do not suggest any method for vulnerability assessment, but foremost, the input data source used to calculate the indicators should be verified, and then the vulnerability index is used as a baseline to propose measures for the locality as well as predict the future impact of the disasters. Balica et al. (2012) pointed out that since the vulnerability assessment is based on indicators, its main limitation is the accuracy of the data on which it is based. For the results to be valid, all data must be derived from reliable sources (Balica et al. 2012). Most recently, Van et al. (2022) have also confirmed that the accuracy of the input data remains to be the weakness of their approach on flood vulnerability assessment. Many studies (Balica et al. 2012, Nasiri et al. 2016; Hussain et al. 2021; Van et al. 2022) reflect accuracy assessment as an existing issue of their vulnerability studies that needs to be addressed in their future works. In this study, apart from the selection of the data from reliable sources provided by local government departments such as the Da Nang Statistical Office, Da Nang City Government or Da Nang Department of Natural Resource and Environment, we attempt to ensure the reliability of the flood vulnerability map by using field survey flood depth data and the local statistical data on flood damage.

Field survey flood pillar data for Da Nang City, including 52 points, have been used to validate the flood exposure map developed from this study. Our survey indicates that all these flood pillar points are located in areas with exposure levels from moderate to very high. There are 42 of the 52 flood pillar points (80% of total pillars) lying in the high and very high levels in the flood exposure map. Based on the assumption that the exposure map that represents the physical component of the FVI has been verified and the data on flood sensitivity and AC were published by the local official organizations, our proposed flood vulnerability map for Da Nang City could be valuable for both local inhabitants and stakeholders. The flood pillar points in Figure 9 also indicate that the flood hazard in Da Nang City is the most serious in the low-lying southern area which matches with the most vulnerable area in the FVI map.

Apart from the field survey flood pillar data, the local government reports also affirm the high flood vulnerable areas in Da Nang City. According to the annual reports of Da Nang Steering Committees for Disaster Prevention and Search and Rescue, Da Nang City Government about disaster loss and prevention in Da Nang City, Hoa Vang district is invariably the most seriously affected region in the city as indicated in Table 6. Regarding the FVI map retrieved from this study, Hoa Vang also has the largest area of the high and very high levels of vulnerability represented in Table 4, following the Ngu Hanh Son, Cam Le, and Lien Chieu districts. The correspondence between FVI map and the statistical data on flood damage confirms the reliability of the flood vulnerability assessment developed in this study.

Table 7

Comparison between the statistical data of households affected by flood and the FVI in Hoa Vang District, Da Nang City

View Large

Although the results offer valuable insights into flood vulnerability assessment at a local scale, this study could not overcome the limitations that most vulnerability studies are facing, and further investigations could be considered to enhance the reliability of the results. First, the FVI criteria were selected based on the literature review and expert discussions that are still sensitive and the same indicator scheme cannot be applied to all study areas. Moreover, although the AHP is a popular weighting method that is widely used in many applications, its judgment scale is still subjective. In future work, sensitivity analysis of indicators could be considered to verify the uncertainty of the flood vulnerability assessment method.

Flood vulnerability assessment is always an important need in any lowland river basin, especially in cities like Da Nang that are coastal. The present study demonstrates the utilization of Remote Sensing and the GIS approach integrated with the multi-parametric AHP method for flood vulnerability assessment in Da Nang City, Vietnam. This study has generated the GIS database and a robust indicator scheme for flood vulnerability assessment considering the physical, social, and economic conditions of the study area. This study has first time applied the approach of IPCC (2014) in the flood vulnerability assessment for Da Nang City at the local scale and identified the areas that are highly vulnerable to floods. The AHP method was applied to determine the weights for each parameter and compute the FVI, including three main components: exposure, sensitivity, and AC. The flood vulnerability map has been generated based on categorizing the FVI into five levels: very low, low, moderate, high, and very high. The AHP analysis of criterion priorities and weighting calculation results showed that the flood depth, land-use condition, and drainage system are the key factors affecting the flood vulnerability level of Da Nang City. Results of vulnerability assessment also revealed that more than 60% of the study area is under moderate-to-very high flood vulnerability levels. This could be a valuable input for local government to take timely measures for mitigating flood hazard. This study also indicates that the identification of highly vulnerable areas could help in devising flood response strategies and government budget allocations to appropriately address local needs. This study has also, for the first time, attempted to evaluate the accuracy of the flood vulnerability map by using the field survey of historical flood signs as well as local reports on the flood damage. The correlation R²=0.79 between the FVI and statistical data on the household affected by the flood again confirms the reliability of the flood vulnerability assessment results for Da Nang City in this study. The present study provides the scientific basis for designing sustainable policies on flood hazard mitigation. The proposed method affirms the potential of applying the AHP model using Open Remote Sensing and GIS data for flood vulnerability assessment based on IPCC's approach for Da Nang City.

As the methodology is still under development, there is no standardized way to measure vulnerability, and the assessment. depends on the characteristics of each study area, the hazard that occurs, and its frequency. The FVI model developed from this study can be used as a toolkit to evaluate and manage the urban flood vulnerability and facilitate the adapting capacity. The application of FVI workflow should carefully consider the local physical and socio-economic conditions as well as the flood occurrence in each case study.

This work was supported by the Vietnam Ministry of Education and Training (MOET) under the project no. B2021-DNA-14. The authors would like to express sincere thanks to MOET for their financial support on this study. We are also grateful to the Graduate School of Global Environmental Studies (GSGES), Kyoto University, Japan, for their support on this study. The authors also would like to thank the Da Nang Steering Committees for Disaster Prevention and Search and Rescue and the Vietnam Institute of Meteorology, Hydrology and Climate Change (IMHEN) for their help with field data verification. We thank the USGS, NASA SRTM, Google Earth Engine, and the Da Nang Open Data Portal and Open Source Geospatial Foundation for providing open data and Open Source Software for the data processing in this study. We also thank the anonymous reviewers for their valuable comments and suggestions that improved the quality of the earlier versions of the manuscript.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.