Natural hazard triggered technological risks in the Yangtze River Economic Belt, China

With the continuous occurrence of natural disasters, natural hazard triggered technological accident (Natech) risks also follow. At present, many countries have performed much research on Natech risks. However, there is still a lack of Natech research at the regional or watershed level in China. The Yangtze River Economic Belt (YREB) is not only an industrially intensive development area but also an area with frequent natural disasters. In this study, we selected the YREB as a typical case to study the Natech risk triggered by floods, geological disasters, and typhoons at the regional or watershed level. Four types of risk indicators representing risk sources, natural hazard factors, control levels, and vulnerabilities were developed to assess the spatial patterns of the Natech risks of the YREB. The results show that the Natech risk triggered by floods and typhoons is more serious in eastern area and central area than in western zone and that the Natech risk triggered by geological disasters is more serious in the west part. Approximately 7.85% of the areas are at relatively high-risk and above the Natech risk level based on the comprehensive assessment of three types of Natech risks. The combined population of these areas accounts for approximately 15.67% of the whole YREB, and the combined GDP accounts for approximately 25.41%. It can be predicted that the occurrence of Natech risks in these areas will cause serious harm to both the people and the economy. This work will provide the basis and key management direction for Natech risk management in the YREB.

www.nature.com/scientificreports/ The contribution of this work can be summarized as follows. (1) An index system is developed for regional Natech risk assessment, which can be used to quantify the impacts of Natech risk on the economic, environmental and social aspects. (2) The Natech risk triggered by natural disasters, such as floods, geological disasters, and typhoons, is evaluated; the findings can assist decision makers in making effective policies for risk management. (3) The framework of this work provides a reference for risk assessment at the macro level.

Data and method
Data. Risk sources of industrial enterprises. The basic information of more than 140 thousand industrial enterprises in China comes from the China Environmental Statistics Database (ESD) for 2015 32 . The risk sources collected information included the codes, names, longitudes and latitudes, scales of industrial enterprises and the industrial sectors to which the enterprises belong. Then the Q value is collected. It is defined as the ratio of the risky substance quantity that was used and stored in the enterprise to the threshold quantity relating to the physical toxicity, environmental hazard, and diffusion characteristics of the substance 33 . The Q value is a significant impact variable in the occurrence of an accident 35 . If an enterprise has more than one risk substance, the Q value equals m n=1 Q n , where Q n is the sub Q value of risk substance n 34,35 . The final Q value of each chemical enterprise is collected and determined mainly through the following channels. First of all, the data is collected from of the national investigation on environmental risks and chemicals from enterprises in key industrial sectors, which was conducted in 2010. Secondly, the data is collected from of the investigation on environmental incident risks among enterprises in Jiangsu Province in 2015. Third, the data is collected from of investigations on enterprise environmental risks in some regions such as Urumqi City (2015), Tianjin Binhai New District (2015), etc. The Q value is obtained through the above channels. And the Q values are assumed that did not change between the investigation years and assessment year. For the enterprises not included in these investigations, their Q values indirectly are obtained by using the enterprises' scale and sector information from ESD of 2015. That is, the Q value of other enterprises is taken as the average of the Q value of existing enterprises in the same industry with the same enterprise scale 36,37 . Therefore, the Q values can represent the intensity of risk sources. This Q value dataset is applied to assess the risks of environmental incidents in the YREB 36 as well as throughout China 37 .
Hazard data on natural disasters. The hazard data on natural disaster risk sources mainly come from remotesensing data and basic geographic data, including flood, geological disaster and typhoon disaster data.
The flood hazard factor data are collected from the following sources. (1) Flood inundation range data of the YREB in 1998YREB in , 2002YREB in , 2006, and 2010 are obtained by Landsat and Modis satellite image interpretation. (2) Digital elevation data mainly come from the Environment Data Cloud Platform of the Chinese Academy of Science (EDCP-CAS, http:// www. resdc. cn/ DataL ist. aspx). (3) The primary site coordinates, flood level, peak discharge, and flood return period during a flood, are mainly derived from the annual report of the Ministry of Water Resources (http:// www. mwr. gov. cn/ sj/) and site information of the China Water Station.
The geological disasters factor data are collected from the following sources. (1) The data on earthquake points in China and surrounding areas since 1900 mainly come from the EDCP-CAS (http:// www. resdc. cn/ data. aspx? DATAID= 296). The data mainly include important information such as the longitude and latitude coordinates, occurrence time, magnitude, focal depth, and location. (2) The data of seismic intensity caused by geological disasters come from the EDCP-CAS (http:// www. resdc. cn/ data. aspx? DATAID= 290). The data includes 7 major types of geological disasters: collapse, subsidence, debris flow, ground subsidence, ground fissures, landslides, and slopes. (3) The geological deformation data came from the European Space Agency (ESA) sentinel online (https:// scihub. coper nicus. eu/ dhus/#/ home).
The typhoon hazard factor data are collected from the following sources. (1) The data on typhoon tracks in the Western Pacific come from EDCP-CAS (http:// www. resdc. cn/ data. aspx? DATAID= 293). The data mainly include important information such as the track points and paths of 1782 typhoons from 1951 to 2018 and the maximum wind force. (2) The precipitation data affected during the typhoon are derived from the precipitation station data of the China Meteorological Administration from 2000 to 2018. (http:// www. resdc. cn/ data. aspx? DATAID= 282).
Other data. Control mechanism data. (1) The data on the proportion of investment in regional environmental management come from the "China Environmental Statistics Yearbook" from 2010 to 2017. (2) The original data on enterprise environmental irregularities come from the "Institute of Public & Environmental Affairs" (IPE, http:// www. ipe. org. cn/ index. html). (3) Statistics on the number of emergencies between 2010 and 2018 come from "Annual Statistic Report on Environment in China". (4) Data on energy conservation and environmental protection expenditure comes from provincial statistical yearbooks and government websites.
Vulnerability data. (1) The population data in 2015 come from EDCP-CAS. (2) The GDP data per unit area of each district or county come from the Statistical Yearbook of 11 provinces and cities in the YREB. (3) The vector data on the national water attribute come from the "National Geomatics Center of China" (http:// www. webmap. cn/ main. do? method= index). (4) The sensitive point vector data of education and medical treatment come from the EDCP-CAS (http:// www. resdc. cn/ data. aspx? DATAID= 330). Construction of indicator system. The indicator system was constructed considering four aspects: the risk source, control level, natural disaster hazard factors, and vulnerability. These indicators are shown in Fig. 2. Calculation of Natech risk indices and determination of index weight. The calculations of the risk indices of Natech risks triggered by floods, geological disasters, and typhoons are described as follows. We used the analytical hierarchy process (AHP) 38 to calculate the weights of the indicators. The pair-wise compari-  Table S1-S5 in the Supplementary Materials (SM). The relative weight of each index is shown in Tables 1, 2 and 3, and the consistency checking result (CR < 0.1) indicates each index's reasonability and effectivity.
Risk index calculation of Natech risk triggered by floods.

Risk level classification standard of enterprise risk sources
The Q value and the threshold quantity of risky substances refer to the grading standard method provided by the Ministry of Ecology and Environment (MEE) 33 and can be found in the "Classification method of for environmental accident risk of the enterprise (HJ941-2018)". The calculation formula of Q is as follows.
where the w 1 , w 2 , . . . , w n means that the stock of each risk substance, t The W 1 , W 2 , . . . , W n means that the critical quantity of each risk substance, t The quantification of Q refers to the tenfold conversion relationship for environmentally hazardous substances in German inventory law. Because of the actual situation of existing data processing and the lower hierarchy of the classification method, two new levels have been added. The method is mentioned in the "enterprise environmental risk grading assessment method (preparation instruction)" 39 . As a result, the hazards of risk sources were recategorized into 6 levels according to the size of Q. The classification criteria as follows.
Q < 1, represented by Q0. 1 ≤ Q < 10, represented by Q1. 10 ≤ Q < 100, represented by Q2. 100 ≤ Q < 1000, represented by Q3. 1000 ≤ Q < 10,000, represented by Q4. Q ≥ 10,000, represented by Q5. Among them, the risk levels from Q0 to Q5 increase in order. In addition, it is necessary to consider the degree of hazard contribution under different Q value ranges. The hazard degree of different Q-level intervals in each district is calculated. The calculation formula and classification criteria are as follows.
where n is the total number of Q-level classifications in each district or country. HQ (district) represents the degree of hazard in different Q-level intervals of each district or county. Q q represents the number of enterprises in different Q intervals of the district. Cq (hazard) represents the hazard contribution in different Q intervals. HQ * (district) is the standardized treatment of HQ (district). Cq (hazard) is assigned as 15%, 30%, 45%, 60%, 75%, and 90% from Q0 to Q5, respectively.

Flood submerged range and inundation risk level
The impact of topography on flood formation is mainly reflected in the fact that the lower the terrain elevation is, the more vulnerable it is to floods. First, based on the relatively serious and representative floods in history, the floods of 1998, 2002, 2006 and 2010 were selected as the representative objects of the research. The result of the flood inundation range was determined by calculating the difference in water area between the flood period and the normal period. By calculating the proportion of each district or county flood inundation to the total inundation, the temporal and spatial impact of a flood were expressed.
Then, combined with the elevation data of the inundated area, the inundation degree of the inundated area is obtained. Finally, the degree of inundation is divided into six grades. From levels 1 to 6, the terrain increases, and the degree of inundation decreases in turn.

Vulnerability indicators of the population
The vulnerability index of the population is shown in Eq. (3): where V (pop) indicates the vulnerability index of the population. pop (district) is the population of a district or a county. pop min is the minimum population of districts or counties. pop max is the maximum population of districts or counties.

Vulnerability indicators of the sensitive points of medical education
The vulnerability indicator of the sensitive points of medical education is written as Eq. (4): The proportion of each district or county flood inundation in the total inundation (%). Classification according to the part of "Flood submerged range and inundation risk level" The total inundation frequency of each district or county/4 0.3204 100 Superimpose DEM data on submerged area data to classify submerged levels. The proportion of the frequency of emergencies in the administrative region in the total number of regional emergencies from 2009 to 2018 (% where V (LCME) indicates the vulnerability index of medical education. LCME (district) is the number of medical education institutions in a district or county. LCME min is the minimum number of medical education institutions in districts or counties. LCME max is the maximum number of medical education institutions in districts or counties. The details of these indicators are summarized in Table 1.
In summary, the Natech risk index triggered by floods in each district or county is calculated as follows:  where CRF i is the Natech risk index triggered by floods in the ith district or county. SF i is the risk source indicator of Natech risk triggered by floods in the ith district or county. HF i is the hazard factor indicator of Natech risk triggered by floods in the ith district or county. CF i is the control mechanism level indicator of Natech risk triggered by floods in the ith district or county. VF i is the vulnerability indicator of Natech risk triggered by floods in the ith district or county.
Risk index calculation of Natech risk triggered by geological disasters.

Analysis of the magnitude hazard index
Based on the statistical data on earthquakes that occurred from 1900 to 2018 in the YREB, magnitude and focal depth are classified. Seismic magnitude division is based on the Richter scale division standard. Furthermore, based on the divided geological disasters levels, the hazard index of geological disasters in districts or counties is calculated, and their contribution is assigned. The degree of hazard in the region is presented in Eq. (6): where n represents the total number of magnitude-level classifications in each district or country. Hm (district) represents the degree of hazard in different magnitude intervals of each district or county. Q (m) is the number of earthquakes with different magnitudes in each district or county. Cm (hazard) represents the hazard contribution degree in different magnitude intervals.

Analysis of the hazard index of earthquakes depth
According to the criteria of seismic focal depth, the calculation formula and classification criteria of the degree of hazard in different magnitude intervals of each district or county are shown in Eq. (7): where n represents the total number of seismic depth level classifications in each district or country. Hd (district) represents the degree of hazard in the different seismic depth intervals of each district or county. Q (d) is the number of earthquakes with different seismic depths in each district or county. Cd (hazard) represents the www.nature.com/scientificreports/ hazard contribution in different seismic depth intervals. Based on the criteria of seismic focal depth, the seismic depth is divided into (0,30], (30,60], and above 60 and Cd (hazard) is set as 90%, 60%, and 30%, respectively.

Analysis of the hazard index of geological deformation
Based on data availability and the occurrence of geological disasters from 2016 to 2020, we finally chose data from 2016, which had more geological disasters, as the representative data. The geological deformation was calculated by the DInSAR technique 40,41 . Then, based on the geological deformation results, we calculate the average rate of land subsidence. Finally, the standardized results are presented in Eq. (8): where Hs (district) indicates the hazard index of geological deformation. S (district) is the average rate of land subsidence in a district or county. S min is the minimum number of the average rates of land subsidence in districts or counties. S max is the maximum number of the average rates of land subsidence in districts or counties. The details of indicators of are shown in Table 2.
The Natech risk index triggered by geological disasters is calculated by the following formula: where CRG i is the Natech risk index triggered by geological disasters in the ith district or county. SG i is the risk source indicator of Natech risk triggered by geological disasters in the ith district or county. HG i is the hazard factor indicator of Natech risk triggered by geological disasters in the ith district or county. CG i and VG i is the control mechanism level indicator and the vulnerability indicator in the ith district or county. The indicator of CG i and VG i refer to the Natech risk triggered by flood.
Risk index calculation of Natech risk triggered by typhoons.

Analysis of the frequency of typhoons
According to the data from the typhoon center of the Japan Meteorological Agency, the average radius of the seven-level wind circle was approximately 350 km, as influenced by the typhoon. The frequency of typhoons is determined by the number of typhoons that the buffer zone intersects with each county-level administrative area. Then, based on precipitation data of the typhoon period, the precipitation and the affected area caused by each typhoon were calculated in the unit of districts or counties.

Analysis of the regional precipitation affected during typhoons
The meteorological stations were matched with the districts and counties of the YREB, and then the affected districts and counties under the path of each typhoon every year were selected. Then, according to the time period of each typhoon path, the precipitation of the weather station at the same time period was filtered and matched to the affected districts and counties. On this basis, the precipitation brought by the typhoon path buffer was calculated. The superposition of two kinds of precipitation to obtain each typhoon affected period precipitation information. The assessment indicators of Natech risk triggered by typhoons are shown in the Table 3.
The Natech risk index triggered by typhoons is calculated by the following formula: where CRT i is the Natech risk index triggered by typhoons in the ith district or county. ST i is the risk source indicator of Natech risk triggered by typhoons in the ith district or county. HT i is the hazard factor indicator of Natech risk triggered by typhoons in the ith district or county. CT i and VT i is the control mechanism level and the vulnerability indicator in the ith district or county. The indicator of CT i and VT i refer to the Natech risk triggered by flood.
Risk classification of Natech in the YREB. According to the Recommended Method for Risk Assessment of Environmental Incidents in Administrative Areas 42 , the relatively high risk is added to the original risk level. The Natech risk index ( CRF i , CRG i , CRT i ) be classified into one of the five following risk levels: high risk ( CR i ≥60), relatively high risk (50 ≤ CR i <60), medium risk (40 ≤ CR i <50), relatively low risk (30 ≤ CR i <40) and low risk ( CR i <30).

Results and discussion
Distribution of Natech risk triggered by floods. Based  The proportion of risk level from low to high is 45.7%, 28.3%, 9.44%, 4.3% and 0.84%. In addition, 11.40% of (8) Distribution of Natech risk triggered by geological disasters. Based on geological disasters in recent years, the average seismic intensity caused by geological disasters of the same area was analysed. The results are shown in Fig. S3 (SM). The average seismic intensity increases in order from east to west. After calculation based on the original data of geological deformation in Fig. S4 (SM), it is found that the degree of geological deformation is more serious in northwestern and southern Hunan and southern Zhejiang. Based on the calculation results of comprehensive indicators of Natech risk triggered by geological disasters, the risk distribution map of Natech triggered by geological disasters in the YREB is shown in Fig. 4. The proportion of risk level from low to high is 52.2%, 23.27%, 4.67%, 0.37% and 0.00%. In addition, 19.44% of areas are not affected by Natech risk. There are 4 relatively high-risk areas in the YREB, i.e., Ruian City, Linhai City of Zhejiang Province, Panzhou City of Guizhou Province, Jiangyin City of Jiangsu Province. The risk level of most districts and counties is low risk or no risk.  Discussion. In summary, based on the results of the Natech risk analysis, it can be found that the Natech risk triggered by floods and typhoons is more serious in the east and centre than in the west part, such as Jiangsu, Zhejiang, and Hunan. However, the Natech risk triggered by geological disasters is more serious in the west part than in the central area and east area, mainly concentrated in Sichuan, Guizhou, Chongqing and parts of Zhejiang and Jiangsu, mainly concentrated in Sichuan, Guizhou, Chongqing and parts of Zhejiang and Jiangsu. From the results of the Natech risk spatial pattern, the overall Natech risk triggered by floods and typhoons is higher than that triggered by geological disasters, which may be because the current risk sources in the east and centre regions are more densely distributed than those in the west region. And the seismic points are mainly distributed in Sichuan, Chongqing and Yunnan in the west. Therefore, the risk level and high-risk area of Natech risk triggered by geological disasters are relatively weaker than those of the other two. From the perspective of areas not affected by Natech risk, these areas triggered by typhoons account for the largest proportion of the three types of Natech risks. From the perspective of the high Natech risks of all types, there are nine areas of high-risk Natech triggered by floods. The population of these areas is approximately 16.81 million, and the GDP is approximately 3149.4 billion yuan. There are 18 areas of high-risk Natech triggered by typhoons. The population of these areas is approximately 26.49 million, and the GDP is approximately 4131 billion yuan. There are no Natech high-risk areas triggered by geological disasters. In addition, from the comprehensive analysis of the three types of Natech risks, there are 84 districts or counties with relatively high-risk and high-risk in the YREB, accounting for 7.85%. The total population of these areas is approximately 88.04 million, and the total GDP is approximately 10.2 trillion yuan. It can be predicted that the occurrence of Natech risks in these high-risk areas will cause serious harm to both the people and the economy.
In addition, based on the levels of hazard factors shown in Figs. S6-S8 (SM), the risk levels of flood and typhoon disaster factors decreased from east to west. Among them, the areas severely affected by floods were mostly concentrated in the northwest of Jiangxi, around Dongting Lake, Anhui, Shanghai and other areas. Most  www.nature.com/scientificreports/ areas of Sichuan, Yunnan, and Chongqing were severely affected by geological disasters. The severity of the area affected by the typhoon gradually decreased from southeast to northwest. The control mechanism level is shown in Fig. S9 (SM). Anhui, most of Jiangxi and parts of Yunnan and Sichuan had relatively high levels of control. Among them, the frequency of environmental emergencies in the Eastern Coastal Region (including Shanghai, Zhejiang, and Jiangsu) is more prominent. The proportion of investment in environmental pollution control in Shanghai, Sichuan, Hunan and Zhejiang is weaker than that in other provinces. Compared with other provinces, Guizhou has more serious violations of regulations, and Sichuan's expenditure on energy conservation and environmental protection is relatively weak. The vulnerability level is shown in Fig. S10 (SM). Areas with relatively serious vulnerabilities were concentrated in parts of Jiangsu, Shanghai, Hubei, Hunan and other regions, while the vulnerability in northern Sichuan and southern Yunnan was relatively light. The sensitivity analysis is performed to verity the rationality of the results by change the weight of indicators. The Natech risk distribution map is shown in Fig. S11-13 (SM). From the results of the sensitivity analysis, it could be concluded that although there exist differences for the weight of indicators differences, the results are not changed from the final result. According to the risk level and spatial pattern of Natech in the YREB, many regions are below the medium risk level. However, Natech risk has the characteristic of low risk probability and serious consequences of damage. There are still many districts or counties with a relatively high risk of natural disasters. These areas still need to strengthen their risk management. Therefore, it is necessary to pay more attention and manage these areas based on the evaluation results. This work will also provide directions and suggestions for the improvement of enterprise construction location, industrial chain and industrial structure in the future.
Limitations. In the calculation process, the risk source data for the industrial enterprises were collected from the sources published in 2015. If considering recent development and industrial construction, the current Natech risk may change. Thus, it is necessary to carry out further tracking research on Natech risk in this direction. Due to the limitation of data availability, relevant disaster indicators, such as fragilities, seismic peak ground acceleration, the velocity and wave height of flood etc. have not been considered in detail. Further studies are needed to improve and optimize the indicator system if the corresponding data are available. In addition, the selection of relevant indicators, such as the level of control mechanisms, is calculated based on the provincial level, and there may be some deviations for the specific management at the district or county level. In the future, refining the data should be considered.

Conclusions
The YREB is not only an industrial-intensive area but also an area where natural disasters occur frequently. In this paper, the Natech risks triggered by floods, geological disasters, and typhoons are comprehensively analysed and evaluated from risk source indicators, natural hazard factor indicators, control level indicators and vulnerability indicators. Finally, the risk level and spatial patterns of Natech risk in the YREB were determined. This research has identified the high-risk areas of Natech accidents under different natural disasters in the YREB. With the frequent occurrence of natural disasters, this work will help decision makers and management departments to strengthen the priority supervision and management of Natech risk areas. Moreover, it provided directions for strengthening Natech prevention and management in the YREB.
The research is mainly based on spatial multiple indicators to identify the Natech risk distribution in the YREB. The main advantage of this approach is that it provided a comprehensive indicator selection proposal at a large scale, especially for basin-scale Natech risk assessment. In the future, extended analysis and research will be conducted on the loss of containment and diffusion effects of different Natech risks. Current work will help to carry out future research based on different Natech risk levels.