Abstract

The emergy theory provides a new approach for flood risk assessment from an ecological perspective. By employing the emergy method, we used five indicators (rainfall runoff, medical workers and students per 10,000 people, social fixed assets investment, unit land GDP, and land-use types) from three dimensions (natural environment, population, and social economy) and the GIS technique to assess the potential impact and risk of a flood disaster on different regions in Ya'an City. Our findings revealed regional differences in the distribution of flood risks in Ya'an City: Lushan County and Yucheng District face the highest risks, followed by Tianquan County and Mingshan District, and Shimian County has the lowest risk. The index method was employed to analyze the regional differences. By training a back-propagation neural network with data on flood disasters in the study area, we produced a flood risk distribution map. We found that Mingshan District, Lushan County and Yucheng District have higher risks than other regions. The results largely agree with what we obtained using the emergy method. Our study shows that flood risk assessment based on the emergy theory can provide a scientific basis for flood control and disaster relief initiatives.

HIGHLIGHTS

  • Emergy theory generates a new approach for flood risk assessment.

  • Research findings were verified by the reality of the study area, ensuring the validity.

  • Analysis for BP neural network training was conducted in the area experiencing a flood.

  • The flood risks in the study area assessed by using the emergy theory are in agreement with the reality.

INTRODUCTION

Floods are one of the most common hazards worldwide, have a wide-ranging impact and present a serious threat to socio-economic development and people's lives and properties (Foudi Osés-Eraso & Tamayo 2015). In July 2020, most parts of Sichuan, Chongqing, Hubei, Anhui and Zhejiang around China experienced multiple rounds of heavy rainfall, resulting in severe flood disasters around many regions, leading to some major losses to the sustainable development of our national economy, jeopardizing the security of people's lives and properties, and affecting social stability. Recently, flood risk assessment has been a heated topic in hydrology and catastrophology. Analysis of the flood hazard is key to disaster risk assessment and control and provides a timely solution to early warning, flood control, and disaster relief.

Currently, most efforts for flood risk assessment in China and abroad have been made with historical flood data, hydrological and hydrodynamic models, and index methods (Tanoue et al. 2016; Abdulrazzak et al. 2019; Yongzhi et al. 2021). Bhuiyan & Baky (2014) employed ArcGIS to map the flood hazard distribution around the low-lying areas of Bangladesh, and used digital elevation model (DEM) data to extract index factors, to work out the flood risk analysis diagram for different scales of floods. Kazakis et al. (2015) employed the analytic hierarchy process to determine the weight of total flow, altitude, land-use type and other attributes, and superposed different parameter information following their weighted values, and completed flood risk assessment. Chengguang et al. (2015) selected 10 assessment indexes based on disaster-inducing factors, hazard-inducting environment, and hazard-affected bodies, built up a random woodland intelligence algorithm-based flood risk assessment model, and then used the GIS technology to assess the flood risks in the Dongjiang River Basin. Sun et al. (2017) used the GIS software to summarize the flood risk analysis mode from the three aspects of historical data, system indexes, and remote sensing data, commented on the features and the short-comings of some common models, and proposed that extended flood risk analysis models and enhanced application of the 3S technology to flood risk analysis would be the direction for us to enhance and extend flood risk analysis in the future. Hanif et al. (2020) investigated the impact of short-term and long-term climate changes on the hydrological regime and flood risks, generated a flood map to identify the vulnerability, and formulated some flood adaptive strategies for 43 regions around the globe.

The indexes in these aforementioned research works are usually not interrelated, and for these works have determined the weights by analyzing the importance of the factors to ultimately work out the risk assessment indexes. They have ignored the potential relation between different indexes and would not help to assess the flood risks accurately. The emergy analysis approach, created by American ecologist H. T. Odum, refers to a kind of emergy accounting of the ecological–economic system to assess the status and function of different types of energy in the system, so it is a quantitative research method to assess flood risks (Huang & Odum 1991). This approach converts different types of incomparable energy into the same standard emergy for comparison to quantitatively analyze the natural system and the human socio-economic system, the real value of resources and the environment, and their interrelationship (Luo et al. 2019). Wu et al. (2018), based on the emergy theory, calculated the rainstorm intensity, susceptibility, and adaptive capacity to assess the vulnerability of each district of Zhengzhou in Henan province to flood disasters, and the results showed that different districts have different demand for rainfall flood control, and the districts of higher vulnerability demand higher. Chang & Huang (2015) pointed out that the emergy approach works to highlight the exposure, susceptibility, and adaptive capacity, and spatialized calculation of the emergy indexes can help assess the risk of floods. Therefore, the infrastructure management and regional planning are critical factors to flood risks. Based on the emergy theory, this paper assesses the flood risks in Ya'an and is expected to provide a new research direction for flood risk assessment in the future.

This paper is organized as follows: First, the introduction introduces some approaches to flood risk assessment in China and abroad and the status quo of research on application of the emergy theory to risk assessment. The Research Methods section elaborates on the principles of the emergy-based risk assessment model, an overview of the survey region, and data sources and processing. The Results and Discussion section uses the emergy method and the back propagation (BP) neural network-based index method to assess the risk of floods and analyze the results. The last section concludes the paper and points out some prospects for future efforts.

METHODS

The basis for application of emergy theory to floor risk assessment modeling

The emergy theory and method proposed by American ecologist H. T. Odum work to convert different types of incomparable energy into standardized comparable values based on the emergy conversion rate. This approach measures the emergy of different types of energies in the ecological system that consists of the human social system and the natural ecological system on the benchmark of solar energy. Based on solar energy, the energy in the ecosystem can be converted into the solar energy of a unified standard through the emergy conversion rate equation (Zhao 2013):
formula
(1)
where E stands for emergy (seJ), τ is the emergy conversion rate (seJ/J or seJ/g), and B is the volume of energy or substance (J/g).

The flood disaster risk energy system is established based on the causal relationship between the systems comprised of the city system and the nature–agriculture system to demonstrate the different energy and material flow between the ecosystem and the economic system (Wangyang et al. 2019). When rainfall takes place, the rainwater-stored energy (N1) may endanger the rainfall area. After deducting buildings' and vegetation's retainment, evapotranspiration and infiltration losses, rainfall is then converted into an overland runoff, and the energy stored in the total overland runoff accumulation represents the exposure in this region (N2). The exposure intensity (N3) is the ratio of the emergy stored in the volume of runoff to the emergy stored in the rainfall. When extreme weather takes place, the sum of the energy loss of the city system (N4) and the energy loss of the nature–agriculture system (N5) is called susceptibility (N6). The possibility of flood disaster in the system has positive correlation with the energy it stores. Thanks to extreme weather, the environment also changes accordingly, and the disaster-stricken region is also affected; this impact is denoted by the potential impact index (N7). The potential impact index is related to the exposure and susceptibility and the product of susceptibility and exposure intensity. In addition to the natural environment, the factors affecting a region's disaster risk include the demographic factor (N8) and the socio-economic factor (N9). They together demonstrate the adaptive capacity of a system (N10) and the coping capacity of this region to any natural hazard. Finally, we use the potential impact and adaptive capacity to assess flood disaster risk in the survey region (N11). Table 1 lists the definitions of different energy flows. For the working principles, see ‘Risk assessment modeling’.

Table 1

Emergy flow of flood hazard risk

Energy flow (seJ)DescriptionDefinition
N1 Potential exposure I Rainfall 
N2 Exposure L Total volume of runoff 
N3 Exposure intensity N2/N1 
N4 Natural–agricultural system susceptibility Property losses in the natural–agricultural system caused by extreme weather 
N5 City system susceptibility Property losses in the city system caused by extreme weather 
N6 Susceptibility M N4 + N5 
N7 Potential impact index N6 × N3 
N8 Population indicator Demographic factor in the survey region 
N9 Social and economic indicator Social and economic factors in the survey region 
N10 Adaptive ability N N8 + N9 
N11 Risk N7/N10 
Energy flow (seJ)DescriptionDefinition
N1 Potential exposure I Rainfall 
N2 Exposure L Total volume of runoff 
N3 Exposure intensity N2/N1 
N4 Natural–agricultural system susceptibility Property losses in the natural–agricultural system caused by extreme weather 
N5 City system susceptibility Property losses in the city system caused by extreme weather 
N6 Susceptibility M N4 + N5 
N7 Potential impact index N6 × N3 
N8 Population indicator Demographic factor in the survey region 
N9 Social and economic indicator Social and economic factors in the survey region 
N10 Adaptive ability N N8 + N9 
N11 Risk N7/N10 
Table 2

Factors affecting adaptive capacity

Contributing factorsClassificationDefinition
Society Medical workers per 10,000 people Medical workers/(total population of a region/10,000), indicating local medical capacity 
Enrolled students per 10,000 people Students enrollment/(total population of a region/10,000), indicating overall education 
Economy Unit land GDP Region GDP/region area, indicating local economic effectiveness, which has positive correlation to the adaptive capacity 
Total social fixed assets In positive correlation to the adaptive capacity 
Demography Population density Year-end resident population/region area, the region of higher-density population distribution seeing lower adaptive capacity 
Proportion of people below 14 and above 65 years old Number of this group/resident population, this group being highly susceptible, and less adaptive to a natural disaster 
Contributing factorsClassificationDefinition
Society Medical workers per 10,000 people Medical workers/(total population of a region/10,000), indicating local medical capacity 
Enrolled students per 10,000 people Students enrollment/(total population of a region/10,000), indicating overall education 
Economy Unit land GDP Region GDP/region area, indicating local economic effectiveness, which has positive correlation to the adaptive capacity 
Total social fixed assets In positive correlation to the adaptive capacity 
Demography Population density Year-end resident population/region area, the region of higher-density population distribution seeing lower adaptive capacity 
Proportion of people below 14 and above 65 years old Number of this group/resident population, this group being highly susceptible, and less adaptive to a natural disaster 

Risk assessment modeling

A risk refers to how a system is adversely affected, which can be indicated by exposure, susceptibility, adaptive capacity, and potential impact indexes. By analyzing these constituent factors of risks, and exploring their inter-relations, we can put forward some emergy indicators to represent the risk of floods and thus accurately assess the risk of floods in a region.

  • (1)

    Exposure L

Exposure refers to the extent to which a region is involved in a natural disaster; for instance, in a flood, a region that is more susceptible to a flood has higher exposure (Wanzhi et al. 2019). When a flood takes place, the exposure of a region to the flood can be represented by the total emergy of the accumulative runoff in this region, i.e., the product of the accumulative runoff and its corresponding emergy conversion rate.
formula
(2)
where L refers to exposure (seJ); r is the runoff depth (mm); τ is the emergy conversion rate (seJ/J or seJ/g).
  • (2)

    Susceptibility M

Susceptibility refers to the extent to which the system in a region is affected by extreme weather (Chang & Huang 2015). As a response of a system to extreme weather, susceptibility is not just decided by the scale of a hazard like a flood, but also dependent on relevant features of the disaster-stricken region. Different land-use types stock different energies, so when a flood disaster occurs, the impact on different regions of different land-use types also differs. Thus, we can analyze the susceptibility based on the emergy power intensity of different land types (Jun et al. 2015). Therefore, the susceptibility of a region is denoted by the total energy stored by different land types.
formula
(3)
where M refers to the susceptibility of the region (seJ), ρ is the emergy power intensity of different land-use types (seJ/m2), and A is the area of different land-use types (m2).
  • (3)

    Adaptive capacity N

In addition to exposure, the risk of a system is also decided by its adaptive capacity. Adaptive capacity refers to the capacity of a system to make continuous progress and adjustments to adapt itself to the environment when any natural disaster is triggered by extreme weather (Wahab & Ludin 2018). The adaptive capacity of a system constantly changes along with the changing environment, and different regions and groups have different adaptive capacity to the environment. For instance, an underdeveloped rural village has a lower adaptive capacity than a developed region of complete infrastructure, and adults have a much stronger adaptive capacity to the environment than children and the aged. Therefore, the adaptive capacity of a region should be assessed from the three aspects of society, economy, and demographics. The classification and interpretation of each influencing factor are shown in Table 2.

  • (4)

    Potential impact index I

When extreme weather occurs, the impact on the system can be denoted by the potential impact index (Yangfan 2018). The risk of a flood disaster is in positive correlation to the potential impact. The product of exposure and susceptibility denotes the potential impact index.
formula
(4)
formula
(5)
where I refers to potential impact index (seJ), I is exposure intensity, M is the susceptibility of the region (seJ), L is exposure (seJ), and q is potential exposure (seJ).
  • (5)

    Risk index V

Risk index refers to the ratio of the potential impact to the adaptive capacity, indicating the damage of the disaster to a system in case of extreme weather (Chao et al. 2020). The risk is positively correlated to the system's potential impact index, but in negative correlation to the adaptive capacity.
formula
(6)
where V refers to the index of vulnerability, I is the potential impact index (seJ), and N is the adaptive capacity.

Overview of the study area

Ya'an (101°56′–103°23′E, 28°51′–30°56′N), located on the western margin of the Sichuan Basin, is in the transition zone from the Basin to the Qinggai-Tibet Plateau. It is a long strip stretching about 220 km long from north to south and spreads about 70 km wide from east to west. Ya'an City has jurisdiction over six counties and two districts (Figure 1), covering a total area of 15,046 km2. The city boasts quite diversified topographic features. Mountains cover the largest area within its borders, with higher terrain in its south, west, and north, and low-lying land in its center and east. Thanks to its unique ‘windward slope’ and ‘horn mouth’ terrains to lift up water vapor in a cyclic process for coagulation to form precipitation, the city has 218 annual mean rainfall days and is called the ‘Rainy City’. In all four seasons in a year it rains, so all districts and counties in Ya'an make a potentially high-risk region. Statistics showed that during the flood season in 2018, Ya'an experienced 18 regional rainstorms, 465,100 people in 137 townships (towns and sub-districts) in 8 counties (districts) of the city were affected, with a total economic loss up to 2.01 billion yuan. These figures indicate that the rainstorm and flood disasters exert quite a strong impact on the economic development of Ya'an, and are important factors hindering the economic development of Ya'an City.

Figure 1

Elevation map of Ya'an.

Figure 1

Elevation map of Ya'an.

Data sources

The data sources for this study and their specifications are shown in Table 3.

Table 3

Data presentation

MethodCategoryDescriptionSourceSpecification
Emergy theory Hydrology and water resource Rainfall runoff Ya'an Municipal Communique on Water Resources 2018 Various water resources data for Ya'an in 2018 
Society and economy Medical workers Ya'an Statistics Yearbook and Sichuan Statistics Yearbook Number of medical workers in each district or county 
Student enrollment Number of students enrolled in each district or county 
People below 14 and above 65 years old Population of this group in each district or county 
Social fixed assets investment Social fixed assets investment in each district or county 
GDP Total output value for each district or county 
Land-use type L Main land-use type data Natural Resources and Planning Bureau of Ya'an City, Ya'an Statistics Yearbook, and Resource and Environment Science and Data Center, China Academy of Sciences Various land-use resources data 
BP neural network Disaster-inducing factors TRMM 3B43 satellite precipitation NASA, USA https://disc.gsfc.nasa.gov/ Global 0.25° × 0.25° monthly precipitation raster data 1998–2016 
Annual runoff depth isogram Hydrology and Water Resources Bureau of Sichuan Province http://www.schwr.com/ 1956–2000 Synchronization series 
Land cover UMD University of Maryland, USA https://www.umd.edu/ Global land cover classification at 1 km spatial resolution 
Soil type Food and Agriculture Organization (FAO) http://www.fao.org/home/en/ Global land cover classification at 1 km spatial resolution 
DEM Geospatial data cloud http://www.gscloud.cn/ SRTM V4.1 data merged, disposed, and generated as SRTM V4.1 data 
 Disaster Flood-stricken points in Ya'an in history Flood Control & Drought Relief Headquarters, People's Government of Sichuan Province/Information collected online 79 villages as of the provincial headquarters and 10 villages collected online 
MethodCategoryDescriptionSourceSpecification
Emergy theory Hydrology and water resource Rainfall runoff Ya'an Municipal Communique on Water Resources 2018 Various water resources data for Ya'an in 2018 
Society and economy Medical workers Ya'an Statistics Yearbook and Sichuan Statistics Yearbook Number of medical workers in each district or county 
Student enrollment Number of students enrolled in each district or county 
People below 14 and above 65 years old Population of this group in each district or county 
Social fixed assets investment Social fixed assets investment in each district or county 
GDP Total output value for each district or county 
Land-use type L Main land-use type data Natural Resources and Planning Bureau of Ya'an City, Ya'an Statistics Yearbook, and Resource and Environment Science and Data Center, China Academy of Sciences Various land-use resources data 
BP neural network Disaster-inducing factors TRMM 3B43 satellite precipitation NASA, USA https://disc.gsfc.nasa.gov/ Global 0.25° × 0.25° monthly precipitation raster data 1998–2016 
Annual runoff depth isogram Hydrology and Water Resources Bureau of Sichuan Province http://www.schwr.com/ 1956–2000 Synchronization series 
Land cover UMD University of Maryland, USA https://www.umd.edu/ Global land cover classification at 1 km spatial resolution 
Soil type Food and Agriculture Organization (FAO) http://www.fao.org/home/en/ Global land cover classification at 1 km spatial resolution 
DEM Geospatial data cloud http://www.gscloud.cn/ SRTM V4.1 data merged, disposed, and generated as SRTM V4.1 data 
 Disaster Flood-stricken points in Ya'an in history Flood Control & Drought Relief Headquarters, People's Government of Sichuan Province/Information collected online 79 villages as of the provincial headquarters and 10 villages collected online 

Data processing

The rainfall runoff data for this study were obtained from Ya'an Municipal Communique on Water Resources 2018. The number of medical workers, enrolled students, people below 14 and above 65 years old, social fixed assets investment, and GDP were obtained from Ya'an Statistics Yearbook and Sichuan Statistics Yearbook. The main land-use data of Ya'an were obtained from Natural Resources and Planning Bureau of Ya'an City, Ya'an Statistics Yearbook, and Resource and Environment Science and Data Center, China Academy of Sciences (Figure 2) in combination with the Geographic Information System (GIS). The data for the BP neural network were normalized to remove the dimension before the network training. Specifically, the data were normalized by Matlab to obtain the weights of various indexes and conduct overlay analysis in the GIS.

Figure 2

Types of land use in Ya'an City.

Figure 2

Types of land use in Ya'an City.

Data obtained are arranged as shown in Table 4.

Table 4

Data of Ya'an City by type

RegionShimianHanyuanBaoxingLushanTianquanXingyangYuchengMingshan
Accumulative runoff (100 million m349.29 27.30 42.33 22.59 60.27 44.98 23.57 9.85 
Medical workers (10,000 people) 111.00 65.00 73.00 67.00 102.00 78.00 171.00 56.00 
Enrolled students (10,000 people) 1,769.00 1,298.00 1,033.00 1,197.00 1,373.00 1,355.00 1,559.00 1,224.00 
Social fixed assets investment (100 million RMB Yuan) 51.04 45.54 25.96 28.59 48.24 47.30 106.49 62.83 
Unit land GDP on average (10,000 RMB/km2306.91 373.68 107.86 362.72 268.62 417.05 1,672.62 1,314.24 
Farmland (km287.00 295.00 42.00 84.00 123.00 94.00 192.00 166.00 
Woodland (km2798.00 501.00 818.00 256.00 1,971.00 416.00 475.00 147.00 
Area of land for mining and industry (km23.00 3.00 1.00 1.00 2.00 1.00 3.00 8.00 
Area of land for residence (km2153.00 17.00 212.00 43.00 50.00 20.00 193.00 43.00 
Area of land for transportation purposes (km20.56 0.61 0.17 0.48 8.08 0.99 3.75 0.55 
Water bodies (km236.00 74.00 4.00 4.00 3.00 3.00 10.00 4.00 
RegionShimianHanyuanBaoxingLushanTianquanXingyangYuchengMingshan
Accumulative runoff (100 million m349.29 27.30 42.33 22.59 60.27 44.98 23.57 9.85 
Medical workers (10,000 people) 111.00 65.00 73.00 67.00 102.00 78.00 171.00 56.00 
Enrolled students (10,000 people) 1,769.00 1,298.00 1,033.00 1,197.00 1,373.00 1,355.00 1,559.00 1,224.00 
Social fixed assets investment (100 million RMB Yuan) 51.04 45.54 25.96 28.59 48.24 47.30 106.49 62.83 
Unit land GDP on average (10,000 RMB/km2306.91 373.68 107.86 362.72 268.62 417.05 1,672.62 1,314.24 
Farmland (km287.00 295.00 42.00 84.00 123.00 94.00 192.00 166.00 
Woodland (km2798.00 501.00 818.00 256.00 1,971.00 416.00 475.00 147.00 
Area of land for mining and industry (km23.00 3.00 1.00 1.00 2.00 1.00 3.00 8.00 
Area of land for residence (km2153.00 17.00 212.00 43.00 50.00 20.00 193.00 43.00 
Area of land for transportation purposes (km20.56 0.61 0.17 0.48 8.08 0.99 3.75 0.55 
Water bodies (km236.00 74.00 4.00 4.00 3.00 3.00 10.00 4.00 

RESULTS AND DISCUSSION

Calculated results

Based on the above data, we calculated the exposure, susceptibility, and adaptive capacity of Ya'an City with the method mentioned in the Methods section.

Exposure

Exposure is the product of accumulative runoff and its corresponding emergy conversion rate (Jun et al. 2015), which can be calculated via Equation (2). The results are shown in Table 5.

Table 5

Results of exposure calculation

IndicatorsShimianHanyuanBaoxingLushanTianquanXingyangYuchengMingshan
Accumulative runoff (100 million m349.29 27.30 42.33 22.59 60.27 44.98 23.57 9.85 
Exposure (1020seJ) 27.99 15.51 24.05 12.84 34.24 25.55 13.39 5.60 
IndicatorsShimianHanyuanBaoxingLushanTianquanXingyangYuchengMingshan
Accumulative runoff (100 million m349.29 27.30 42.33 22.59 60.27 44.98 23.57 9.85 
Exposure (1020seJ) 27.99 15.51 24.05 12.84 34.24 25.55 13.39 5.60 

Susceptibility

A region's susceptibility to disaster can be denoted by the total energy stored by different land-use types (Chang & Huang 2015). In this study, the susceptibility of different land-use types can be calculated with Equation (3). The results are shown in Table 6.

Table 6

Sensitivity calculation results

RegionShimianHanyuanBaoxingLushanTianquanXingyangYuchengMingshan
Farmland (km287.00 295.00 42.00 84.00 123.00 94.00 192.00 166.00 
Emergy (106 seJ) 1.07 3.63 0.52 1.03 1.51 1.16 2.36 2.04 
Woodland (km2798.00 501.00 818.00 256.00 1,971.00 416.00 475.00 147.00 
Emergy (106seJ) 4.82 3.03 4.94 1.55 11.91 2.51 2.87 0.89 
Area of land for residence (km23.00 5.00 2.00 12.00 5.00 5.00 21.00 6.00 
Emergy (108 seJ) 3.38 5.64 2.25 13.53 5.64 5.64 23.68 6.76 
Area of land for mining and industry (km23.00 3.00 1.00 1.00 2.00 1.00 3.00 8.00 
Emergy (107seJ) 5.44 5.44 1.81 1.81 3.63 1.81 5.44 14.52 
Area of land for transportation purposes (km20.56 0.61 0.17 0.48 8.08 0.99 3.75 0.55 
Emergy (107seJ) 1.59 1.73 0.48 1.36 22.90 2.81 10.63 1.56 
Water bodies (km236.00 74.00 4.00 4.00 3.00 3.00 10.00 4.00 
Emergy (102 seJ) 9.74 20.02 1.08 1.08 0.81 0.81 2.71 1.08 
Susceptibility (1024seJ) 4.15 6.43 2.54 13.90 8.44 6.15 25.39 8.42 
RegionShimianHanyuanBaoxingLushanTianquanXingyangYuchengMingshan
Farmland (km287.00 295.00 42.00 84.00 123.00 94.00 192.00 166.00 
Emergy (106 seJ) 1.07 3.63 0.52 1.03 1.51 1.16 2.36 2.04 
Woodland (km2798.00 501.00 818.00 256.00 1,971.00 416.00 475.00 147.00 
Emergy (106seJ) 4.82 3.03 4.94 1.55 11.91 2.51 2.87 0.89 
Area of land for residence (km23.00 5.00 2.00 12.00 5.00 5.00 21.00 6.00 
Emergy (108 seJ) 3.38 5.64 2.25 13.53 5.64 5.64 23.68 6.76 
Area of land for mining and industry (km23.00 3.00 1.00 1.00 2.00 1.00 3.00 8.00 
Emergy (107seJ) 5.44 5.44 1.81 1.81 3.63 1.81 5.44 14.52 
Area of land for transportation purposes (km20.56 0.61 0.17 0.48 8.08 0.99 3.75 0.55 
Emergy (107seJ) 1.59 1.73 0.48 1.36 22.90 2.81 10.63 1.56 
Water bodies (km236.00 74.00 4.00 4.00 3.00 3.00 10.00 4.00 
Emergy (102 seJ) 9.74 20.02 1.08 1.08 0.81 0.81 2.71 1.08 
Susceptibility (1024seJ) 4.15 6.43 2.54 13.90 8.44 6.15 25.39 8.42 

Adaptive capacity

When a natural disaster occurs due to extreme weather, the system, to alleviate the resulted losses, can constantly make progress and adjustment to adapt to the changing environment (Wanzhi et al. 2019). The results of the adaptive capacity calculation are shown in Table 7.

Table 7

Results of adaptive capacity calculation

RegionShimianHanyuanBaoxingLushanTianquanXingyangYuchengMingshan
Medical workers (10,000 people) 111.00 65.00 73.00 67.00 102.00 78.00 171.00 56.00 
Emergy (1022seJ) 1.49 0.87 0.98 0.90 1.37 1.04 2.29 0.75 
Enrolled students (10,000 people) 1,769.00 1,298.00 1,033.00 1,197.00 1,373.00 1,355.00 1,559.00 1,224.00 
Emergy (1022seJ) 4.14 3.04 2.42 2.80 3.21 3.17 3.65 2.87 
Social fixed assets investment (100 million RMB Yuan) 51.04 45.54 25.96 28.59 48.24 47.30 106.49 62.83 
Emergy (1021seJ) 3.27 2.92 1.67 1.83 3.09 3.03 6.83 4.03 
Unit land GDP on average (10,000 RMB Yuan/km2306.91 373.68 107.86 362.72 268.62 417.05 1,672.62 1,314.24 
Emergy (1020seJ) 1.97 2.40 0.69 2.33 1.73 2.68 10.75 8.45 
Adaptive capacity (1022seJ) 5.98 5.34 3.04 3.35 5.65 5.54 12.48 7.36 
RegionShimianHanyuanBaoxingLushanTianquanXingyangYuchengMingshan
Medical workers (10,000 people) 111.00 65.00 73.00 67.00 102.00 78.00 171.00 56.00 
Emergy (1022seJ) 1.49 0.87 0.98 0.90 1.37 1.04 2.29 0.75 
Enrolled students (10,000 people) 1,769.00 1,298.00 1,033.00 1,197.00 1,373.00 1,355.00 1,559.00 1,224.00 
Emergy (1022seJ) 4.14 3.04 2.42 2.80 3.21 3.17 3.65 2.87 
Social fixed assets investment (100 million RMB Yuan) 51.04 45.54 25.96 28.59 48.24 47.30 106.49 62.83 
Emergy (1021seJ) 3.27 2.92 1.67 1.83 3.09 3.03 6.83 4.03 
Unit land GDP on average (10,000 RMB Yuan/km2306.91 373.68 107.86 362.72 268.62 417.05 1,672.62 1,314.24 
Emergy (1020seJ) 1.97 2.40 0.69 2.33 1.73 2.68 10.75 8.45 
Adaptive capacity (1022seJ) 5.98 5.34 3.04 3.35 5.65 5.54 12.48 7.36 

The above calculated results reveal the following findings:

  • (1)

    Table 5 shows that the bigger the accumulation runoff in the system, the more energy it will store, and the higher the exposure will be. Tianquan County has the highest annual precipitation, and generates the largest accumulative runoff, resulting in high emergy storage, so Tianquan County is the region with the highest exposure. Mingshan District sees relatively low annual precipitation, so its accumulative runoff is relatively small, leading to relatively low emergy storage, so the exposure is also relatively low.

  • (2)

    Based on the emergy power density of different land-use types, we calculate the emergy of different land-use types for all districts and counties of Ya'an City. It can be seen from the results that areas with high energy values, such as residential areas, industrial and mining areas, and transportation, are highly sensitive, so floods are at greater risk. The areas with a low energy value, such as water and cultivated land, are less sensitive and less risky. Table 6 shows that compared with other districts and counties, Yucheng District has a relatively high proportion of land for residence, mining, and industrial purposes, and the sensitivity is high, so it is vulnerable to flood threats. However, Hanyuan County and other areas with large areas of cultivated land and woodland are less sensitive. Therefore, these areas are less threatened by flood disasters, thus featuring relatively low susceptibility.

  • (3)

    The stronger the system's adaptive capacity is, the stronger capacity it has against a flood disaster. Table 7 shows that different districts and counties in Ya'an City have different adaptive capacity. Yucheng and Mingshan districts boast better economic strength, medical and healthcare conditions, and popular education to see a stronger adaptive capacity. Hanyuan and Tianquan counties see favorable economic growth, but relatively low popular education, so they have a relatively poor adaptive capacity. Baoxing County, which is economically underdeveloped and has poorer medical services and a lower level of popular education, presents the poorest adaptive capacity among districts and counties in Ya'an.

Risk assessment

Based on the calculated results of exposure, susceptibility, and adaptive capacity, we use the calculation method mentioned in the Methods section to calculate the potential impact index and risk of a flood disaster to various districts and counties of Ya'an City. The calculated results are shown in Table 8.

Table 8

Results of risk assessment for districts and counties of Ya'an City

RegionShimianHanyuanBaoxingLushanTianquanXingyangYuchengMingshan
Potential impact index (1024seJ) 3.32 5.14 2.03 11.12 6.75 4.92 20.31 6.74 
Vulnerability (1020.56 0.96 0.67 3.32 1.19 0.89 1.63 0.92 
RegionShimianHanyuanBaoxingLushanTianquanXingyangYuchengMingshan
Potential impact index (1024seJ) 3.32 5.14 2.03 11.12 6.75 4.92 20.31 6.74 
Vulnerability (1020.56 0.96 0.67 3.32 1.19 0.89 1.63 0.92 

With the calculated results in Table 8, the GIS software was used to obtain the potential impact indexes and a risk distribution map for various districts and counties of Ya'an City (Figure 3).

Figure 3

Potential impact indices and risk profiles.

Figure 3

Potential impact indices and risk profiles.

The potential impact index is related to the exposure and susceptibility of a region, which indicates the possible losses of a system in case of extreme weather (Yangfan 2018). Figure 3(a) presents the potential impact of a flood disaster on different districts and counties in Ya'an City. Yucheng District, thanks to a high population density, social fixed assets investment, and economic development, boasts a high level of accumulative social assets, so it will see the highest potential impact in case of a flood disaster. Baoxing and Shimian Counties, with a low population density and slow economic growth, see lower potential indices than other regions.

The risk of a system means that in the same case of a flood disaster, the higher risk emergy it has, the higher risk the system will see, and the more serious the impact will be. Figure 3(b) shows that Lushan County sees the highest risk index, followed by Yucheng District, and Shimian County witnesses the lowest risk index. Shimian County sees the lowest risk as this county has more medical workers per 10,000 people than others, and high social fixed assets investment, so effective control measures can be taken in the case of a flood disaster.

Model effect analysis

We built a BP neural network-based flood risk assessment model to conduct a comprehensive assessment of the emergy theory's performance in flood disaster risk assessment. The artificial neural network uses a lot of simple processing units (neurons) to form a nonlinear dynamic system, which has learning, memorization, calculation, identification, predication, and many intelligent processing functions. It is different from the so-called electronic brain—electronic computer and the AI system deduced based on symbols, being an AI system that imitates our human brain neural system on a different level and to a different extent (Kuang 2017).

The BP neural network is an error back-propagation learning algorithm, and its working principle is to convert an input vector into an output vector via the hidden layer and to map from the input space onto the output space. Its weight works to realize forward mapping using the network input of the current weight and the expected output to meet the mapping requirements for contrastive learning (Jing et al. 2020). The BP neural network does not only have an input layer node and an output layer node, but also has one or more hidden layer nodes. After signal input, we should first propagate it forward to the hidden layer nodes; under the activation function, the output signal from the hidden node is propagated to the output node, to ultimately output the results. Its main operation procedures are to: (1) provide the neural network with training examples, i.e., learning samples, including input and expected output, (2) determine the permissible error of the network between the actual output and the expected input, and (3) change all connection weight values in the network, to make the generated output closer to the expected output, until it meets the demand for the determined permissible error.

To assess the flood disaster risk in Ya'an City, we first should consider the local natural and geographic environment and the socio-economic conditions to choose the assessment indicator factors from hydro-meteorology and topography geology and society. We select precipitation, runoff depth, drainage basin water storage, slope, elevation, population density, and GDP factors in this paper as assessment indicators to form the assessment indicator system, standardize the data, reduce its dimensions, and divide each indicator into five grades through the natural-break point method (Table 9). After standardization, we use the BP neural network, in combination with the actual flooding position, and conduct learning training to obtain the weights of all assessment indicators (Table 10). Then we use the raster calculator on the GIS platform to overlay all indicator factors according to their respective weights to obtain the hazard distribution map of Ya'an City (Figure 4). The training accuracy based on the BP neural network is 0.768.

Table 9

Range of indicators

Risk rankingVery lowLowMediumHighVery high
Precipitation 0–0.176 0.176–0.282 0.282–0.478 0.478–0.639 0.639–1 
Runoff 0–0.231 0.231–0.412 0.412–0.592 0.592–0.792 0.792–1 
Drainage basin water storage 0–0.369 0.369–0.522 0.522–0.643 0.643–0.749 0.749–1 
Slope 0–0.478 0.478–0.604 0.604–0.718 0.718–0.843 0.843–1 
Elevation 0–0.420 0.420–0.576 0.576–0.714 0.714–0.847 0.847–1 
Population density 0–0.141 0.141–0.471 0.471–0.510 0.510–0.945 0.945–1 
GDP 0–0.146 0.146–0.379 0.379–0.641 0.641–0.829 0.829–1 
Risk rankingVery lowLowMediumHighVery high
Precipitation 0–0.176 0.176–0.282 0.282–0.478 0.478–0.639 0.639–1 
Runoff 0–0.231 0.231–0.412 0.412–0.592 0.592–0.792 0.792–1 
Drainage basin water storage 0–0.369 0.369–0.522 0.522–0.643 0.643–0.749 0.749–1 
Slope 0–0.478 0.478–0.604 0.604–0.718 0.718–0.843 0.843–1 
Elevation 0–0.420 0.420–0.576 0.576–0.714 0.714–0.847 0.847–1 
Population density 0–0.141 0.141–0.471 0.471–0.510 0.510–0.945 0.945–1 
GDP 0–0.146 0.146–0.379 0.379–0.641 0.641–0.829 0.829–1 
Table 10

Weights of indicators

FactorRainfallRunoff depthDrainage basin water storageSlopeElevationPopulation densityGDP
Weight 0.271 0.064 0.075 0.084 0.113 0.265 0.128 
FactorRainfallRunoff depthDrainage basin water storageSlopeElevationPopulation densityGDP
Weight 0.271 0.064 0.075 0.084 0.113 0.265 0.128 
Figure 4

Flood risk distribution of Ya'an City.

Figure 4

Flood risk distribution of Ya'an City.

The risk distribution map shows that most regions with relatively high flood hazard risk are in the east part of Ya'an City: Lushan County, Yucheng District, and Mingshan District face the highest risks. Yucheng and Mingshan districts boast relatively high-density population distribution and are more likely to be affected by a flood disaster's damage, so corresponding disaster control and relief efforts should be made. Other districts and counties in Ya'an do not feature high risk, but their vegetation should also be protected, to prevent any flood disaster, and provide bases for local governments to carry out flood control and disaster relief. Compared with some previous research efforts, we have found both indicate that a region, more developed in the economy with more accumulative assets, sees higher potential impact and boasts higher adaptive capacity (Wu et al. 2018).

In order to verify the objectivity of the research results, we compared the historical flood event in Ya'an City. On 22 August 2019, 47 towns and villages in seven counties (districts) of the city were affected, with 96,000 affected people and 22,500 emergency relocations, resulting in a direct economic loss of 3.078 billion yuan. The areas with the largest losses are Lushan and Yucheng, and the disasters are mainly concentrated in Lushan County. The actual loss is consistent with the experimental results. Overall, the prediction matches the actual flood disaster situation, which represents the actual situation of the region.

CONCLUSIONS

Using emergy analysis, we constructed a flood risk assessment indicator system for Ya'an City that involved susceptibility and adaptive capacity of the hazard-affected region. In combination with the BP neural network, we have made a risk assessment and contrastive analysis of the study area. The results are as follows:

  • (1)

    Two methods have been applied to various districts and counties of Ya'an City. The results show that more developed regions like Yucheng and Mingshan districts have the highest potential impact index but lower risk than other regions. This is because these two economically developed districts, well-equipped with medical facilities, have a high adaptative capacity and thus can effectively address various emergencies in case of a flood disaster. Lushan County is just the opposite. This study reveals that a system with a strong adaptive capacity can effectively lower its risk of a flood disaster.

  • (2)

    This study assesses the flood risks of various districts and counties of Ya'an City. In actual practice, the emergy theory has often been used to investigate the sustainable development capacity of an ecological–economic system. However, studies on its application to flood disaster risk assessment are rare, most of which stay in the stage of theoretical or methodological analysis and show defects in real-world application. This paper selects Ya'an area as the research area, and applies emergy theory to the risk analysis of Ya'an City, making the analysis results more objective. Moreover, it is also of great research value to explore whether the emergy theory could be applied to other fields.

ACKNOWLEDGEMENT

This study is funded by the Yangtze River Collaborative Fund for Hydro-sciences, Natural National Science Foundation of China (SFC; U2040210).

DATA AVAILABILITY STATEMENT

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

REFERENCES

Abdulrazzak
M.
,
Elfeki
A.
,
Kamis
A.
,
Kassab
M.
,
Alamri
N.
,
Chaabani
A.
&
Noor
K.
2019
Flash flood risk assessment in urban arid environment: case study of Taibah and Islamic universities' campuses, Medina, Kingdom of Saudi Arabia
.
Geomatics, Natural Hazards and Risk
10
,
780
796
.
Chang
L.
&
Huang
S.
2015
Assessing urban flooding vulnerability with an Emergy approach
.
Landscape and Urban Planning
143
,
11
24
.
Chao
X.
,
Nan
M.
,
Fei
L.
,
Xiaoming
L.
&
Zhiyun
O.
2020
Research on Urban Green Infrastructure Management in Macao from the Perspective of Ecosystem Service Demand. Classical Chinese Garden
.
Chengguang
L.
,
Chen
X.
,
Shiwei
Z.
,
Wang
Z.
&
Xushu
W.
2015
Flood risk assessment model based on random forest and its application
.
Shuili Xuebao
46
,
58
66
.
Foudi
S.
,
Osés-Eraso
N.
&
Tamayo
I.
2015
Integrated spatial flood risk assessment: the case of Zaragoza
.
Land Use Policy
42
,
278
292
.
Hanif
A.
,
Dhanasekar
A.
,
Keene
A.
,
Li
H.
&
Carlson
K.
2020
Flood risk assessment methodology for planning under climate change scenarios and the corresponding change in land cover
.
Journal of Water and Climate Change
11
(
4
),
1370
1382
.
Huang
S.
&
Odum
H. T.
1991
Ecology and economy: emergy synthesis and public policy in Taiwan
.
Journal of Environmental Management
32
(
4
),
313
333
.
Jing
K.
,
Xiaoxi
M.
,
Hailian
W.
,
Lu
W.
,
Kai
S.
&
Aitao
T.
2020
Research progress of artificial neural networks in material science
.
Materials Reports
34
(
21
),
21172
21179
.
Jun
X.
,
Wei
S.
,
Xinping
L.
,
Si
H.
,
Like
N.
&
Gippel
C. J.
2015
Revisions on water resources vulnerability and adaption measures under climate change
.
Advances in Water Science
26
(
02
),
279
286
.
Kuang
W.
2017
Overview of the development of artificial neural networks
.
Frontiers of Science and Technology
6
(
2
),
165
167
.
Luo
P.
,
Xiaorong
H.
,
Xiaoyue
W.
&
Lang
Z.
2019
Research on the sustainability of water eco-economic system in Chengdu based on emergy analysis
.
Water Power
45
(
09
),
12
16
.
Sun
Z.
,
Zhu
X.
&
Pan
Y.
2017
Flood risk analysis: progress, challenges and prospect
.
Journal of Catastrophology
32
,
125
130
.
Tanoue
M.
,
Hirabayashi
Y.
&
Ikeuchi
H.
2016
Global-scale river flood vulnerability in the last 50 years
.
Scientific Reports
6
,
36021
.
Wahab
A. M.
&
Muhamad Ludin
A. N.
2018
Flood vulnerability assessment using artificial neural networks in Muar Region, Johor Malaysia
.
IOP Conference Series, Earth and Environmental Science
169
,
12056
.
Wangyang
Y.
,
Chunbo
J.
,
Jian
L.
&
Qi
Z.
2019
Hydrologic-hydrodynamic model and its application in floodrisk analysis
.
Journal of Hydroelectric Engineering
38
(
08
),
87
97
.
Wanzhi
L.
,
Di
Y.
,
Xiaoli
F.
&
Tiaofeng
Z.
2019
Risk assessment of rainstorm and flood disasters based on the hazard grades/indices in Qinghai Province
.
Journal of Glaciology and Geocryology
41
(
3
),
680
688
.
Wu
Z.
,
Shen
Y.
&
Wang
H.
2018
Assessment of vulnerability to flood disasters based on Emergy theory
.
South-North Water Diversion and Water Resources Science and Technology
16
,
9
14
.
Yangfan
X.
2018
Flood Hazard Assessment Based on Spatial Multi-Criteria Decision Making and Heterogeneous Spatial Data Integration
.
Huazhong University of Science & Technology
,
Wuhan, China
.
Yongzhi
L.
,
Wenwen
T.
,
Wenting
Z.
,
Xingnan
Z.
&
Shuai
N.
2021
Review of flood disaster risk analysis based on disaster chain
.
Water Resources Protection
37
(
1
),
20
27
.
Zhao
L.
2013
Ecological Economics
.
China Economic Publishing House
,
Beijing
.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Licence (CC BY 4.0), which permits copying, adaptation and redistribution, provided the original work is properly cited (http://creativecommons.org/licenses/by/4.0/).