Indicator set establishment for flood vulnerability assessment

The flood vulnerability assessment in this study was implemented based on the results of 23 proposed indicators, as shown in Table 1. These indicators belong to 3 components, namely, exposure (E,7 indicators), sensitivity (S, 7 indicators), and adaptive capacity (AC, 9 indicators), of the IPCC (2007) approach13. The indicators were proposed according to modification of the published indicators for suitability of the practical data and information of the study area.

Sociological investigation

The sociological investigation of the study area was conducted with in-depth and semistructured interviews using questionnaires in October 2023 and April 2024. A total of 200 households (20 questionnaires per residential group of the town) were surveyed via semistructured interviews. Additionally, the in-depth investigation was implemented with an environmental officer and five town officials from Phong Nha. This survey aimed to identify and assess factors that contribute to or exacerbate flood exposure, to identify individuals or groups sensitive to floods, to evaluate the adaptability of the local population and to provide input data sources for calculating vulnerability indicators for each residential group and all of Phong Nha town. The interviewed households were evenly distributed across the whole town based on the calculation of sample size using the following equation:

where N is the population of the study area; e is the margin of error of 5.06%; and z is the z score at the sampling confidence level (p) of 85%.

Data normalization

The indicators were collected from different sources, such as questionnaires from household surveys and reports from the Phong Nha Town People's Committee in 2019-20239, as indicated in Table 1. Because of different units, these data were required to normalize in the 0-1 scale range.

In this research, all indicators were normalized according to four methods. The detailed normalization results are also shown in Table 1. The first method is the calculation based on the percentage calculation taken from the ecosocial report of the Phong Nha Town People's Committee, including indicators S3 and S4. The second method is applied to the min–max data standardization theory (UNDP, 2006)22 for quantitative indicators, as shown in Eq. 2, where the correlation function between the indicator and vulnerability is a direct relationship for standardization. Indicators E1, E4, and S6 were calculated according to Eq. 2.

where x is the normalized value of indicator i of household j; Xij is the value of indicator i corresponding to household j; and the max and min values represent each indicator's maximum and minimum values, respectively.

The third method is for indicators coming from questionnaires with households answering yes or no over the total number of households interviewed. In particular, this method was applied for E3, E6, S2, S5, AC3-AC6, and AC9, with values of 1 and 0 corresponding to households' yes or no answers, respectively. The final method is converting to a scale of 0-1 based on the scoring method applied for indicators E2, E5, S1, S7, AC1-AC3, AC7, and AC8.

Assessment of the vulnerability of households to floods

The indicators of each component, E, S, and AC, were calculated for each residential group. The values of the components E, S, and AC of each residential group and the whole of Phong Nha town are the average values of all indicators belonging to each component on a scale of 0-1. The vulnerability of each residential group and the whole town of Phong Nha to flooding were calculated using Eq. 3 as follows (IPCC, 2007)13:

The obtained vulnerability values on scales of -1 and +2 were then converted into the 0-1 scale by applying Eq. 2 with the range for min, max is from -1 to 2. The exposure, sensitivity, adaptive capacity, and vulnerability on a scale of 0-1 were divided into 5 levels, namely, very high, high, medium, low, and very low, corresponding to 0.8-1.0, 0.6-0.8, 0.4-0.6, 0.2-0.4, and 0-0.2, respectively.

Exposure

In the past 5 years, floods have occurred annually in the study area, and one of the most severe floods recorded was the October 2020 flood. This flood occurred continuously for many days in October, with heavy rainfall. The severity of the damage was recorded and is shown in Table 2. According to the Phong Nha Town People's Committee (2020)23, the impacts included 4 deaths, 17 injuries, and extensive damage to residential structures, with 6 completely damaged houses. A total of 2,610 houses and 114 school facilities were affected, with 42 houses needing to be evacuated in an emergency. Additionally, floods impacted the agriculture, forestry, and fisheries of Phong Nha town. A total of 43.7 ha of agricultural area was washed away, and 4,369 animals died. The estimated damage from this flood reached 20,878.2 million people. Figure 2 shows flood occurrence in the same house but in a different year and the evidence of flooding inside the house of a respondent in Phong Nha town.

Figure 3 shows the normalized values of E1, E3-E4, and the average of E5-E7 for each residential group in Phong Nha town. The frequency of floods impacting households in the past five years varies from 0.20 to 0.75. Notably, Tram Me, Ha Loi, Xuan Tien, and Cu Lac 1 experienced the highest flood frequencies, with values of 0.75. The percentage of households impacted by floods in the last five years ranges from 0.40 to 0.99 across residential areas. Na stands out with the highest percentage of affected households at 0.99, followed by Phong Nha and Cu Lac 1. The population density varies significantly among the residential areas, ranging from 0.01 to 0.73. Xuan Son had the highest population density at 0.73, followed by Na and Phong Nha. The proportion of damaged households during floods in October 2020 varied from 0.46 to 0.95 across the 10 residential areas. Most areas experienced extensive damage, with Tram Me and Cu Lac 1 reporting values of 0.90 and 0.84, respectively. Additionally, 8/10 of the residential areas are close to the Son River within 1 km, leading to high values of E2. In contrast, Ban Rao Con had the lowest flood frequency, distance to the Son River, number of affected households, population density, and number of houses damaged by floods in October 2020.

The exposure values of E1-E7 for all of Phong Nha town are shown in Figure 4. The flood frequency in 5 recent years is 0.58 (E1). A substantial 78% of households experienced flood damage (E3), indicating high effects of floods within the community. Moreover, with a value of 0.83 for E2, many households reside near the Son River, increasing exposure to flood risks. Although the population density is moderate at 0.51 (E4), it can exacerbate flood impacts, especially in densely populated areas. The flood impact indicators (E5-E7) of the October 2020 flood event provide valuable information on its consequences. During the October 2020 floods, 85% of houses were affected (E5), underscoring the widespread impacts on residential structures. Figure 5 shows the flood levels of houses in Phong Nha town. Approximately 52% and 18% of the residential houses were flooded at depths of 1-3 m and above 3 m, respectively. Additionally, 80% of households faced clean water scarcity (E6) during floods, emphasizing the importance of access to clean water for postdisaster recovery. Additionally, 42% of livestock perished (E7), suggesting economic losses for households reliant on livestock. These findings collectively underscore the multifaceted nature of flood impacts on communities. The average value of the indicators of the exposure component (E) was 0.69, which indicates high exposure to floods.

Sensitivity

Figure 6 shows the values of the sensitivity indicators S1-S7 for the 10 residential groups. The vulnerable group indicates the average values of 4 indicators of S1-S4, including children and elderly individuals, households with a third child, foreigners, and females. The results show that the Phong Nha and Xuan Tien residential groups stand out as having the most vulnerable objects. In particular, Phong Nha exhibits a notable value for foreigners (S1, 1.0) and a moderate value for vulnerable age groups (S1, 0.67). Regarding livelihood impacts, family members involved in agriculture (S5) in Tram Me and Na demonstrated the highest values of 0.91-0.92, highlighting a significant reliance on agricultural activities for livelihoods. There was a significant difference in income level among the 10 residential groups. Gia Tinh (S6, 0.44) and Na (S6, 0.41) exhibit a low proportion of households with incomes lower than the basic threshold; otherwise, Ban Rao Con (S6: 0.98) shows the highest value. Additionally, Ban Rao Con exhibits a notable presence of 0.81 single-story houses and wooden houses.

Figure 7 shows the sensitivity values of S1-S7 for Phong Nha town. The study results showed that 76% of households had children under five years old and above 65 years old. In addition, 75% of households had a third child in the family (S2). The study area is one of the travel destinations. Therefore, foreigners unfamiliar with this region's conditions should be considered sensitive. Thirty-nine percent and 50% of the population are foreigners (S3) and females (S4), respectively. When a flood occurs, 77% of family members involved in agriculture are susceptible to rising water levels due to flooding. According to the General Statistics Office of Vietnam (2022)24, the average monthly income per person in urban areas over five years from 2019 to 2023 was 5.7 million people. Thirty-six percent of households had an income lower than this basic value. Fifty-three percent of households lived in temporary and one-floor houses, which are considerably vulnerable to flooding, as shown in Figure 8. Note that 21% and 32% of households lived in temporary and one-floor houses, respectively (S7). Quantitative assessment of the S1-S7 indicators revealed an average sensitivity component (S) value of 0.58, indicating moderate sensitivity.

Adaptive capacity

Figure 9 shows the values of the AC1-AC9 indicators for 10 residential groups in Phong Nha town. The values of almost all of these indicators indicate considerable variation among the 10 residential groups, except for AC5. Specifically, the AC1 (income source diversity) and AC3 (preparedness for proactive response) indicators in these areas vary from 0.12-0.62 and 0.1-0.95, respectively. Most residents in these areas had limited participation in property insurance, so the AC5 values in these areas were equally low. Xuan Tien and Ha Loi demonstrated high adaptive values of income source diversity, preparedness for proactive response, education level, and emergency medical access. Conversely, Ban Rao Con indicated a low adaptive capacity of these criteria.

The adaptive capacity to flood all of Phong Nha town is illustrated in Figure 10. Diversity of livelihood was recognized by households possessing more than 2 livelihoods (see Figure 11), resulting in the value of AC1 being 0.42. Eighty-nine percent of households did not participate in annual disaster training courses, which led to an AC2 value of 0.11. The survey revealed that 85% of households participate in health insurance (AC4); however, only 3% have property insurance (AC5). In an emergency, 68% of households were equipped with disaster mitigation tools such as food and water storage, flashlights, medicine, and contact numbers (AC3); however, due to subjective experience in response, some of these households are located far away and at high altitudes, so they are not equipped with natural disaster prevention tools. In addition, the interview results showed that a low percentage of respondents (64%) had a degree higher than a secondary education degree (AC6). It was noted that 86% of residents had good experience with floods (AC7). Up to 79% of households updated their disaster information through television, radio, and network societies (AC9). The average value of the adaptive capacity component was 0.56, indicating medium adaptive capacity.

Vulnerability

The values of exposure (E), sensitivity (S), adaptive capacity (AC), and vulnerability (V) on a scale of 0-1 for the 10 residential groups are summarized in Figure 12. The E values of the residential groups of Gia Tinh, Na, Tram Me, Phong Nha, Ha Loi, and Cu Lac 1 were significantly greater than the AC values. In particular, the AC value was approximately 2 times lower than the E value in the Tram Me residential group. This area also had a medium S value. Therefore, the greatest vulnerability is 0.94 at scales of -1 and +2 according to Eq. 3. After converting to a scale of 0-1, Tram Me obtained a vulnerability value of 0.65, indicating high vulnerability. The opposite trend was observed for the Ban Rao Con residential group, for which the AC value was greater than the S value. Therefore, this region exhibited a moderate vulnerability of 0.56. The Xuan Tien residential group demonstrated the lowest vulnerability at 0.55 because of the equal values of exposure and adaptive capacity.

Figure 13 shows the values of exposure (E), sensitivity (S), and adaptive capacity (AC) of households in the study area to floods in 5 recent years (2019-2023) on a 0-1 scale, which were 0.69, 0.58, and 0.56, respectively. The indicator-based quantitative assessment result indicated that the vulnerability value was 0.57 on a scale of 0-1, revealing a medium vulnerability level. Despite the medium vulnerability of the whole of Phong Nha Town to floods, it is still important to enhance the adaptive capacity of the town in general, especially for some residential groups in the town, especially those with high vulnerability, such as Tram Me and Cu Lac 1.

Discussion and recommendations

The results show that two residential groups, Tram Me and Cu Lac 1, in Phong Nha town exhibited high vulnerability to floods despite the medium vulnerability of the whole town. The findings from in-depth and semistructured interviews reveal that residents have accumulated experience dealing with frequent disasters such as floods. However, in-depth interviews with the vice president of Phong Nha town and officials indicate that this experience is insufficient in response to different and unprecedented flood situations. The households live relatively close to the Son River, resulting in many houses being flooded up to the roof or inundation depths of 3-5 meters. Additionally, the prevalence of Type IV housing and temporary structures within the study area exacerbated the impacts and damage sustained during the flood event. In addition, a lack of preparedness and readiness, such as water storage tanks, backup generators, participation in a training course on disasters, and property insurance of households, were identified as shortcomings of the study area.

The assessment of vulnerability among each residential group in Phong Nha town provides valuable insights for local authorities and organizations, guiding the design and implementation of interventions to reduce vulnerability to future flood events. The importance of enhancing adaptive capacity underscores the need to strengthen resilience and minimize flood vulnerability. To reduce vulnerability within the study area, the following actions are recommended: (1) reinforcing infrastructure and construction measures capable of withstanding disasters, focusing on supporting households with temporary shelters as well as impoverished, vulnerable, veteran, and ethnic minority households; (2) implementing an early warning system to enhance response time to disasters; (3) enhancing community awareness and practices related to disaster prevention; (4) encouraging participation in training courses in disasters and buying property insurance; and (5) providing vocational training to create employment opportunities and increase livelihood diversity for individuals during flood seasons, such as English language training, handicraft workshops, photography, and souvenir making.