Revealing Risk Stress on the Lanzhou Section of the Yellow River from the Industries alongside It

Abstract

The precise assessment of environmental risks is significant in the transformation from treatment after pollution towards a precaution and management regime. Herein, we employed the linear combination of the Analytic Hierarchy Process (AHP) and Entropy-Weighted (EW) method to exam the impacts of 14 environmental risk factors of 70 enterprises in Lanzhou against the Yellow River aquatic safety. The results showed that enterprises that present a low-risk level to the aquatic safety of the Yellow River accounted for 73% of enterprises, and 27% of enterprises presented a medium level of risk. No enterprise presenting a high level of risk was found. In each administrative district/county, the total value of environmental risks in the Honggu, Xigu, and Anning Districts came to 22.87, 40.51, and 14.17, respectively, due to these districts being the location of a massive number of industries. Furthermore, this study found that the types of enterprises, the species of the risk substance, the shortest distance from the Yellow River, the river′s level, and the population density were the main contributors to the environmental risks. Other issues, such as weak outfall supervision and the shortage of emergency supplies also require increased attention.

1. Introduction

The development of society and the economy is highly water dependent [1,2,3]. Enterprises and industries tend to agglomerate adjacent to water, especially rivers, due to their convenience for water supply and for the transportation of raw materials and intermediate goods [2,3,4,5,6,7]. Consequently, the population increases and cities and towns grow. Interactions between the aquatic environment and human activities are not always nonmalignant. The intentional or unintentional discharge of waste into aquatic media via direct or indirect methods poses great risks to the safety of the aquatic environment [1,2,3,6,7,8,9,10]. The cost is the shortage of a safe water supply and accompanied damage to human health and the commercial interests of enterprises. As the second longest river in China, the Yellow River provides a tremendous amount of water for the residents and enterprises located alongside it, and serves as the largest recipient for waste discharge at the same time [8,11,12]. Several pollution incidents, such as the leakage of benzene into the Yellow River in Lanzhou and the discharge of industrial wastewater into the Yellow River in Bayan Nur over the last two decades, highlight the vulnerability of the Yellow River [6,12,13,14]. Especially for the cities or towns whose water supply is dependent on a sole water source, water pollution significantly disrupts routines, causing chaos in social functioning. Therefore, the precise assessment of the potential risks in addition to post-event remedies are deserving of concern [8,11,12,13,14,15,16,17].

The Analytic Hierarchy Process (AHP) and Entropy-Weighted (EW) methods are two popular methods for environmental risk assessment [16,18,19,20,21]. The AHP is a systematic and hierarchical analysis method that combines qualitative and quantitative analysis. This method is characterized by the use of less quantitative information to mathematize the thinking process of decision making on the basis of in-depth research on the nature, influencing factors, and internal relationships of complex decision-making problems, so as to provide a simple decision-making method for complex decision-making problems with multiple objectives, multiple criteria, or no structural characteristics. It is a model and method for making decisions about complex systems that are difficult to fully quantify [16,18,19,20,21]. AHP is suitable to cope with problems with hierarchical and interleaved evaluation indicators where the target value is difficult to quantitatively describe [16,19,21,22]. Entropy is a measure of the degree of disorder in a system. For a specific index, the entropy value can be used to judge the degree of dispersion of the index. The basic idea of the EW method is to determine the objective weight according to the degree of index variability. Generally, when the information entropy of an indicator is large, its weight is positively related to the information amount which it provides. Thus, the EM method is a typical objective weighting method [18,20,21,22]. Introducing a sole method to analyze environmental risks cannot deliver a precise outcome due to the uncertainty and complexity of the target. However, the subjectiveness of AHP and objectiveness of EW are not suitable to solely introduce in a very complex system, which may generate insufficient and even incorrect analysis results when AHP or EW is solely used. Thus, some balance should be achieved between the objectiveness and the subjectiveness. A combination method integrates the subjectivity of AHP and the objectivity of EW and avoids some of the drawbacks derived from the nature of these methods separately. Cases abound on aquatic environmental assessments, such as lake water quality, water supply stress, etc., and they produce a great deal of useful information for assessing environmental safety [16,18,19,20,21,22].

Lanzhou is the largest city in Gansu province, China, and is located along the upper section of the Yellow River. Due to the limitations of geographic morphology, the Lanzhou urban area winds alongside the Yellow River, with dense population and enterprise distributions occupying the river basin. As the sole aquatic media for water supply and waste, the Lanzhou section of the Yellow River has long been a focus [11,12,13,14,15,17]. However, the comprehensive environmental risks affecting the water safety of the Yellow River from local enterprises and industries have been scarcely investigated and reported. Furthermore, as the most important petrochemical processing hub in Northwest China, many enterprises present high potential risks towards aquatic safety [11,12,13,14,15,17,23,24,25]. Thus, it is crucial to deploy a study considering the potential risks from enterprises and industries in Lanzhou for planning a risk management and control regime.

Herein, we selected 70 representative enterprises in Lanzhou as targets. We employed the re-combined AHP-EW method to thoroughly analyze and rank the risks, the most decisive factors, and the most vulnerable areas according to 14 detailed criteria. The biggest potential risk was found to be from petrochemical-related industries. Additionally, the Honggu, Xigu, and Anning districts were the largest contributors to environmental risk along the Yellow River. We hope that our work could present a panorama regarding the environmental risks from enterprises and industries, and offer some useful clues for decision makers to prepare suitable risk management and control regimes.

2. Preparations and Methods

2.1. Area

Lanzhou is a unique megacity in the upper reaches of the Yellow River where the main stream of the Yellow River passes through the urban area. The Yellow River winds for nearly 50 km in the Lanzhou Basin (Yellow River Valley Basin) [11,12,13,14,15,17]. The two banks of the Yellow River in the urban area are densely populated (about 12.73% of the province′s total population) and have a well-developed economy (about 32.01% of the province′s total gross domestic product). Due to the drought nature in Northwest China, the water source supply is limited. The Yellow River is the main water supply source and pollutant recipient for Lanzhou. Given Lanzhou′s unique geographical location and the dense distribution of population and industrial enterprises, the water environment of the Lanzhou section of the Yellow River is extremely vulnerable to various production and living activities [11,12,13,14,26,27]. The area targeted in this study included all administrative areas in Lanzhou, over six districts (Chengguan District, Anning District, Qilihe District, Xigu District, Honggu District, and Lanzhou New District) and three counties (Yuzhong County, Yongdeng County, and Gaolan County). In Lanzhou, 70 industrial enterprises (with an annual main business income of more than CNY 20 million) were selected as representative enterprises, accounting for about 20% of the city′s approximately 350 industrial enterprises above a designated size. Chen et al. [28] studied the enterprise distribution, nature, and output value in each county/district. Hence, the number of selected enterprises in each district or county was based on their research. Since this research mainly considered the impact of enterprises on the water environment, the selected enterprises had to be water-related enterprises (with quantitative wastewater treatment or discharge). Finally, combined with the main characteristics of the main industries and enterprises in each district and county, the scale of the enterprise (including the number of employees and annual operating income, etc.) was comprehensively considered to sort and establish the enterprises. According to the type of enterprises, there were 15 companies in the petrochemical industry, 13 companies in the food and biopharmaceutical industry, 10 companies in the metallurgy and machinery manufacturing industry, 11 companies in the material industry, 5 companies in the energy and power industry, and 16 companies in other industries. These 70 companies are located in areas with relatively active economic activities in Lanzhou, as shown in Figure 1. The specific company names and geographic coordinates can be obtained upon reasonable request (Chinese version). The numbers 1–70 were used to replace the names of each company below. The large blank area in Figure 1 does not contain the distribution of surveyed enterprises because these areas were mainly inaccessible places such as pasture grasslands, barren grasslands, sandy land, saline–alkali land, mountains, and hills with almost no large-scale industrial distribution.

2.2. Index Selection and Data Source

Based on factor analysis of the emerging risks identification model for the river basin, risk mainly covered three components: risk source (harmfulness), management regime (effectiveness), and recipient (vulnerability) [16,17,18,19,20,21,22]. Hence, three criterion-level indicators were established in this study, namely, risk objective, management regime, and risk recipient. The Guidelines for Risk Assessment of Enterprises in Environmental Emergencies mandated by the government also put forward seven items for risk identification, namely, (1) basic information of the enterprise, (2) risk-recipient surroundings, (3) type and amount of risk matter, (4) manufacturing technology, (5) management of manufacturing safety, (6) control and emergency regime, and (7) exiting emergency source. Of the seven items, items (1), (3), and (4) belong to the risk objective category. Thus, we established three criteria as the detailed index, including the type of enterprise, the species of risk substance, and the environmental risk substance to critical quantity ratio. Items (5), (6), and (7) were credited as being part of the management regime, and we established five criteria as the detailed index, namely, the enterprise emergency plan, the reserve amount of emergency materials, the occurrence of sudden water environmental incidents in the past three years, monitoring of the enterprise′s sewage outlet, and the shortest distance from the Yellow River. Item (2) corresponded to the risk recipient. Due to the current situation of Lanzhou enterprises, geographical features, and existing investigation and research, we proposed six concrete indexes: water environment function, river level, distance from downstream water sources, distance from environmentally sensitive targets, population density, and water environment management and control zoning. The complete details of index establishment are presented in

Table 1. Classification of risk assessment indicators for water pollution-related enterprises in Lanzhou section of the Yellow River catchments.

Criteria	Details	Value
		4	3	2	1
Risk objective	Type of enterprise (C1)	Petrochemical industry	Electroplating, medicine and metallurgy	Machine manufacture and hazards storage	Other
	Species of risk substance K (C2)	K ≥ 10	6 ≤ K < 10	3 ≤ K < 6	K < 3
	Environmental risk substance to critical quantity ratio Q (C3)	Q ≥ 100	10 ≤ Q < 100	1 ≤ Q < 10	Q < 1
Management regime	Enterprise′s emergency plan (C4)	None	-	-	Prepared
	Reserve amount of emergency materials (C5)	None	1–2	3–4	5 and more
	Occurrence of sudden water environmental incidents in the past 3 years (C6)	Severe event	Big event	Small event	None
	Monitoring of the enterprise′s sewage outlet (C7)	None	Personal monitoring	Common online monitoring	Precise online monitoring
	The shortest distance from the Yellow River (km) D (C8)	D ≤ 1	1 < D ≤ 3	3 < D ≤ 5	D > 5
Risk recipient	Water environment function (C9) a	Type Ⅰ, Type Ⅱ	Type Ⅲ	Type Ⅳ	Type Ⅴ
	River level (C10)	Main stream	Primary tributary	Second tributary	Other
	Distance from downstream water sources (km) D (C11)	D ≤ 2	2 < D ≤ 5	5 < D ≤ 10	D > 10
	Distance from environmentally sensitive targets (km) D (C12)	D ≤ 2	2 ≤ D ≤ 5	5 ≤ D≤ 10	D > 10
	Population density σ (C13)	σ ≥ 1200	800 ≤ σ ≤ 1200	300 ≤ σ ≤ 800	σ < 300
	Water environment management and control zoning (C14) b	Priority protection	Common protection	-	None

(a Q =${\displaystyle \sum }_{i=1}^n\frac{w_i}{W_i}$, where, n is the number of hazards; w_i is the largest storage of hazards i; W_i is the critical storage mandated by the Identification of Major Hazards from Hazardous Chemicals; b China′s Surface Water Environmental Quality Standard).

. The data in this research were based on the report from the second national census of pollution sources, and environmental protection, water conservancy, urban construction data, etc. On-site data were collected and processed by means of expert evaluation and consultation.

2.3. Weighting Methods

2.3.1. Analytic Hierarchy Process

AHP is a decision method that coordinates qualification and quantification analysis. It compares every two indexes and identifies the importance degree of each index against an upper index. Then, the weight values are to be given based on the comparison analysis [16,19,21,22]. In this study, 10 experts were invited to assess the value of each index, their assessments are presented in Table 2. Its steps are as follows:

(1)

Define the question and identify the target;

(2)

Establish the judgment matrix;

(3)

Examine the compliance index (CI).

$$CI=\frac{{\lambda }_{max}-n}{n-1}$$

(1)

where λ_max is the largest eigenvalue of the judgement matrix, and n is the number of the experts. Further, we also employed the random index (RI) as a supplement to examine the flexibility of the judgement matrix, and its form is as follows:

$$RI=\frac{{\overline{\lambda }}_{max}-n}{n-1}$$

(2)

where ${\overline{\lambda }}_{max}$ is the weighted average of the largest eigenvalue of the judgement matrix. The judgement matrix is only acceptable when the compliance ratio (R, R = CI/RI) is smaller than 0.1; otherwise, the judgement matrix should be adjusted.

Table 2. Values of each index assessed by 10 experts.

Expert	1	2	3	4	5	6	7	8	9	10
C1	0.4580	0.4075	0.0650	0.4523	0.0884	0.4534	0.0521	0.1252	0.0399	0.1219
C2	0.0673	0.1676	0.0156	0.0905	0.0487	0.1254	0.0521	0.0444	0.0168	0.0501
C3	0.1761	0.0683	0.0077	0.0905	0.1602	0.0690	0.1563	0.0236	0.1236	0.0204
C4	0.0328	0.1354	0.0295	0.0037	0.0084	0.0055	0.0131	0.0059	0.0103	0.0092
C5	0.0029	0.0252	0.1777	0.0072	0.0635	0.0055	0.0361	0.0326	0.0169	0.0200
C6	0.0284	0.0485	0.2797	0.0143	0.0189	0.0291	0.0078	0.0038	0.0034	0.0081
C7	0.0150	0.0485	0.0638	0.0276	0.0292	0.0220	0.0361	0.0253	0.0331	0.0341
C8	0.0062	0.0252	0.1179	0.0534	0.0437	0.0602	0.0131	0.0158	0.0079	0.0594
C9	0.0447	0.0318	0.0097	0.0148	0.0280	0.0126	0.0525	0.0679	0.0607	0.0484
C10	0.0570	0.0136	0.0933	0.0305	0.0570	0.0206	0.0525	0.0777	0.0704	0.0272
C11	0.0071	0.0042	0.0245	0.1158	0.2129	0.1003	0.2704	0.3096	0.2236	0.3003
C12	0.0118	0.0042	0.0156	0.0600	0.1258	0.0601	0.1529	0.1778	0.2236	0.1550
C13	0.0220	0.0073	0.0384	0.0305	0.0849	0.0237	0.0525	0.0515	0.1304	0.0779
C14	0.0707	0.0127	0.0616	0.0089	0.0304	0.0126	0.0525	0.0389	0.0394	0.0680

Table 2. Values of each index assessed by 10 experts.

Expert	1	2	3	4	5	6	7	8	9	10
C1	0.4580	0.4075	0.0650	0.4523	0.0884	0.4534	0.0521	0.1252	0.0399	0.1219
C2	0.0673	0.1676	0.0156	0.0905	0.0487	0.1254	0.0521	0.0444	0.0168	0.0501
C3	0.1761	0.0683	0.0077	0.0905	0.1602	0.0690	0.1563	0.0236	0.1236	0.0204
C4	0.0328	0.1354	0.0295	0.0037	0.0084	0.0055	0.0131	0.0059	0.0103	0.0092
C5	0.0029	0.0252	0.1777	0.0072	0.0635	0.0055	0.0361	0.0326	0.0169	0.0200
C6	0.0284	0.0485	0.2797	0.0143	0.0189	0.0291	0.0078	0.0038	0.0034	0.0081
C7	0.0150	0.0485	0.0638	0.0276	0.0292	0.0220	0.0361	0.0253	0.0331	0.0341
C8	0.0062	0.0252	0.1179	0.0534	0.0437	0.0602	0.0131	0.0158	0.0079	0.0594
C9	0.0447	0.0318	0.0097	0.0148	0.0280	0.0126	0.0525	0.0679	0.0607	0.0484
C10	0.0570	0.0136	0.0933	0.0305	0.0570	0.0206	0.0525	0.0777	0.0704	0.0272
C11	0.0071	0.0042	0.0245	0.1158	0.2129	0.1003	0.2704	0.3096	0.2236	0.3003
C12	0.0118	0.0042	0.0156	0.0600	0.1258	0.0601	0.1529	0.1778	0.2236	0.1550
C13	0.0220	0.0073	0.0384	0.0305	0.0849	0.0237	0.0525	0.0515	0.1304	0.0779
C14	0.0707	0.0127	0.0616	0.0089	0.0304	0.0126	0.0525	0.0389	0.0394	0.0680

2.3.2. Entropy-Weight Method

The EW method is based on conducting the following steps [18,20,21,22]:

(1)

Identify indexes and targets;

(2)

Establish a matrix that includes n indexes and m targets;

$$B=\left(\begin{array}{ccc}{b}_{11}& \cdots & b_{1n}\\ ⋮& \ddots & ⋮\\ b_{m1}& \cdots & b_{mn}\end{array}\right)$$

(3)

(3)

Calculate the proportion P_ij of the index i under the target j;

$$p_{ij}=\frac{r_{ij}}{{{\displaystyle \sum }}_{i=1}^m{r}_{ij}}$$

(4)

(4)

Import entropy;

$$h_j=-{(\ln n)}^{-1}{{\displaystyle \sum }}_{i=1}^m{p}_{ij}\ln p_{ij}, 0\le h_j\le 1;$$

(5)

(5)

Calculate the coefficient of variation of each index difference;

$$a_j=1-h_j, j=1,2,\cdots ,n$$

(6)

(6)

Calculate the weight.

$$w_j=\frac{a_j}{{{\displaystyle \sum }}_{j=1}^n{a}_j}, j=1,2,\cdots ,n$$

(7)

When the value of the evaluation object on a certain index is the same, the entropy value is 1 and the entropy weight is 0. When the value of the evaluation object on an index varies substantially, the entropy value is small and the entropy weight is large. After standardizing the indicators of the main stream of the Yellow River, they were brought into SPSS 23.0 (IBM, Amunc, USA) for entropy weight analysis, and the results are shown in

Table 3. EW analysis results.

Index	Entropy Value	Information Utility Value	Weighting
C1	0.790	0.210	0.065
C2	0.670	0.330	0.102
C3	0.831	0.169	0.052
C4	0.459	0.541	0.167
C5	0.965	0.035	0.011
C6	1	0	0
C7	0.986	0.014	0.004
C8	0.932	0.068	0.021
C9	0.933	0.067	0.021
C10	0.907	0.093	0.029
C11	0.524	0.476	0.147
C12	0.350	0.650	0.200
C13	0.730	0.270	0.083
C14	0.677	0.323	0.099

.

In this study, linear weighting and nonlinear weighting were employed as the recombination method for the AHP and EW method; Equations (8) and (9) were linear weighted and nonlinear weighted, respectively

W_j = μα_j + (1 − μ)β_j

(8)

$$W{\mathrm{j}}_{ }=\frac{\alpha j\beta j}{{{\displaystyle \sum }}_{j=1}^n\alpha j\beta j}$$

(9)

where W_j is the comprehensive weight value, μ is the tending coefficient (if the decision is in favor of the expert’s judgment, μ > 0.5; if the decision is in favor of the subjective data, μ < 0.5); a_j is the weighting value of the AHP method; and β_j is the weighting value of the EW method.

2.4. Environmental Risk Index Calculation

Based on the establishment of the above system, the environmental risk index (ERI) was employed to characterize the environmental risks. A larger ERI value implied a larger probability of pollution [16,18,19,20,21]. The ERI was calculated as follows:

$$ERI={{\displaystyle \sum }}_{i=1}^n{A}_i{W}_i$$

(10)

where A_i is the risk value of index i; W_i is the corresponding weighting value.

2.5. Principal Component Analysis

Principal component analysis (PCA) was calculated using Origin 8 with a PCA plugin.

3. Results and Discussions

3.1. Analysis on Weighting Values via AHP and EW Method

In

Table 4. Weighting values for AHP and EW.

Criteria	Weighting		Index	Weighting
	EW	AHP		EW	AHP
Risk subjective	0.219	0.3839	C1	0.065	0.2264
			C2	0.102	0.0679
			C3	0.052	0.0896
Management regime	0.203	0.1812	C4	0.167	0.0254
			C5	0.011	0.0388
			C6	0	0.0442
			C7	0.004	0.0335
			C8	0.021	0.0403
Risk recipient	0.579	0.4824	C9	0.021	0.0371
			C10	0.029	0.0500
			C11	0.147	0.1569
			C12	0.200	0.0987
			C13	0.083	0.0519
			C14	0.099	0.0396

, it is concluded that methods were distinct. For EW, five factors—C2, C4, C12, C13, and C14—recorded higher values than their counterparts under AHP analysis. C1, C3, C5, C6, C7, C8, C9, C10, and C11 told the reverse story, presenting smaller values compared with those obtained from the AHP method. Due to the differences in detailed indexes, the accumulated influences on the comprehensive value in environmental risks were in larger divergence. In Figure 2, all 70 enterprises were under the low-risk region via the EW method, whereas 24 enterprises were under the medium-risk region and 46 enterprises were under the low-risk region via the AHP method. Furthermore, AHP seemed to disfavor the risk objective more, but benefited the risk recipient, which was the result of more weighting inclined to some indexes vulnerable to objectiveness. However, no enterprise was identified as high risk under either the AHP or EW methods, meaning that the environmental risks are controllable in Lanzhou.

3.2. Analysis on Linear and Nonlinear Recombination of AHP and EW Methods

Due to the objective nature of the AHP method and the subjective nature of the EW method, adapting a sole method could generate overdue or undersize estimations of environmental risks [16,18,19,20,21]. First, we applied the nonlinear weighting method, as shown in

Table 5. Comparison of four weighting methods.

Index	EW	AHM	Nonlinear	μ = 0.5	μ = 0.8
C1	0.065	0.2264	0.1729	0.1457	0.1941
C2	0.102	0.0679	0.0813	0.0849	0.0747
C3	0.052	0.0896	0.0547	0.0708	0.0821
C4	0.166	0.0254	0.0495	0.0957	0.0535
C5	0.011	0.0388	0.0050	0.0249	0.0332
C6	0.001	0.0442	0.0005	0.0226	0.0356
C7	0.004	0.0335	0.0016	0.0187	0.0276
C8	0.021	0.0403	0.0099	0.0306	0.0364
C9	0.021	0.0371	0.0092	0.0291	0.0339
C10	0.029	0.0500	0.0170	0.0395	0.0458
C11	0.147	0.1569	0.2709	0.1519	0.1549
C12	0.199	0.0987	0.2307	0.1488	0.1187
C13	0.083	0.0519	0.0506	0.0675	0.0581
C14	0.099	0.0396	0.0460	0.0693	0.0515

and Figure 3, and the value of the enterprise risk calculated by the nonlinear method is relatively low, with even a medium-risk enterprise of 0 distribution. Further, we took a compromised weighting with an equal weighting value (μ = 0.5) that AHP and EW methods shared. However, a balance between the importance of sensitive factors and the influence of insensitive factors was not struck (Table 5). Some sensitive factors that are usually given much more attention were ignored, such as factors C11 and C12. Thus, we preferred to adapt the nonlinear recombination method which highlighted the importance of significantly concerned factors while limiting the influence exerted from insensitive factors. Based on the principle of the least safety, μ was set to 0.8 to reduce the loss caused by insufficient response and prevention mechanisms due to risk underestimation. The comparisons of five methods in weighting value of each enterprise are presented in Table S2 and Figure 3. As shown in Figure 3, under linear method (μ = 0.8), finally, 19 enterprises were identified as having a medium-risk distribution.

3.3. Similarities and Disparities in the Main Environmental Risk Factors in Different Districts/Counties

In total, 19 enterprises were identified as posing a medium level of environmental risk against the Yellow River based on the linear weighting method (μ = 0.8), and they distributed in 5 counties/districts, namely, Chengguan District, Anning District, Xigu District, Honggu District, and Gaolan District (

Table 6. Number of enterprises of middle risk in each district/county.

Area	Chengguan District	Anning District	Qilihe District	Xigu District	Honggu District	Lanzhou New District	Gaolan County	Yuzhong County	Yongdeng County
Number	3	1	0	12	1	0	2	0	0

). This distribution was consistent with industry distribution in Lanzhou. Xigu District was the area with the densest petrochemical industry distribution in Lanzhou, accounting for at least 20% of the total number of the enterprises over Lanzhou. Additionally, the number of enterprises with a medium level of risk was the highest in Xigu District. Xigu District is located at the upper section in Lanzhou and could exert huge impacts on the downstream districts/counties, such as Chengguan District (the most populated district in Lanzhou). Further, we investigated details in each county/district in the medium level of risk, and found some similarities and disparities (Figure 4). For similarities, five districts/counties all presented obvious pressure regarding the monitoring of the enterprise′s sewage outlet (C7), implying that the lack of a corresponding regime was universal. Due to the short distance to the Yellow River mainstream, Chengguan District, Xigu District, and Anning District showed a high level of pressure on the river level (C8). Chengguan District and Anning District also presented a high pressure on the reserve amount of emergency materials (C5) and the shortest distance from the Yellow River (C8), which was the result of the small area of two districts and the lack of industrial enterprise layout in earlier plans. The urban area of Lanzhou was located in a river basin which was ample in length (more than 30 km) but short in width (smaller than 5 km). Thus, all enterprises located not longer than 4 km away from the Yellow River in Chengguan District and Anning District, and corresponding sensitivity towards C5 and C8, were very high. Despite high pressure on C7, C8, and C10 by Xigu District, plain pressure was on other indexes. The extreme number of other indexes surpassed other districts/counties. For instance, when comparing enterprises for C1, Xigu District had 12 enterprises displaying obvious pressure, whose number was more than all other districts/counties. Similar findings were also observed in C3 (seven enterprises) and C5 (four enterprises). Hence, more concern was placed on other indexes of the Xigu District though no obvious pressures were displayed.

The relationships with each index were an important consideration for decision making. Here, PCA was employed to analyze the linear weighting values (μ = 0.8) for all 70 enterprises via the Monte Carlo significance test. The arrow referred to the index, while its length stood for the correlation. The first pivot accounted for C8, C10, and C11, whose accumulation was responsible for 39.8%. The second pivot was for C1, C3, C4, C5, and C14, whose accumulation was responsible for 29.7%. Loading of Gaolan County, Yuzhong County, Yongdeng County, and Lanzhou New District mainly distributed in the second and the third quadrant, which corresponded to the enterprise′s emergency plan (C4), reserve amount of emergency materials (C5), monitoring of the enterprise′s sewage outlet (C7), and distance from environmentally sensitive targets (C12). Enterprises of Honggu District, Chengguan District, Xigu District, and Anning District were primarily located at the first and fourth quadrants, meaning that environmental safety was vulnerable to the type of enterprise (C1), species of risk substance (C2), environmental risk substance to critical quantity ratio (C3), the shortest distance from the Yellow River (C8), water environment function (C9), river level (C10), and water environment management and control zoning (C13). It can be concluded that the inherent geography affected the urban design, and then the resultant disparities in environmental risk pressure by different counties/districts evolved.

Accumulation of environmental risks based on ERI in each of the districts/counties showed that Xigu District received the highest value of 40.51. Honggu District attained the second-highest value of 22.87 (Figure 5). These two districts are located along the upper stream of the Lanzhou section, presenting high-risk potential toward the Yellow River and the downstream districts/counties. Anning District, Qilihe District, and Chengguan District were distributed alongside the Yellow River Basin and, in turn, their total risk values were 14.17, 1.40, and 11.40, respectively. The above areas represent the main residential areas with relatively few industrial enterprises, and thus, they present relatively low risk potential.

3.4. Similarities and Disparities in Main Indexes of Different Industries

Due to the diversity in nature, size, and geography of different enterprises, their risk pressure against the Yellow River presents some similarities and some disparities. All 70 enterprises were sorted into six types of industry, namely, petrochemical processing industry, food and pharmaceutical manufacturing industry, metallurgy and machine manufacturing industry, materials industry, energy and power industry, and other industries. All industries presented obvious risk pressure on the enterprise′s sewage outlet (C7) (Figure 6), further highlighting that the lack of a corresponding regime was universal and the importance of immediate plan making. Pressure related to the type of enterprise (C1) was obvious in the petrochemical processing industry and presented some degree of pressure on the food and pharmaceutical manufacturing industry, materials industry, and other industries. However, there was no pressure on the metallurgy and machine manufacturing industry and energy and power industry, which was mainly due to the nature of the enterprises. The reserve amount of emergency materials (C5) especially presented high pressure on the food and pharmaceutical manufacturing industry and metallurgy and machine manufacturing industry. The recent release of the Brinell virus in Lanzhou in 2019 highlighted the pressure in this index faced by the food and pharmaceutical manufacturing industry, as the virus caused serious damage to thousands of citizens adjacent to the release source in Lanzhou. Pressure on the shortest distance from the Yellow River (C8) was mainly faced by the petrochemical processing industry, food and pharmaceutical manufacturing industry, materials industry, and energy and power industry. This was due to the geographical constraints of Lanzhou. The topography of the basin caused enterprises to be densely distributed on the flat valley bottom on both sides of the Yellow River. Furthermore, owing to the small area, it was difficult for enterprises to expand to peripheral areas far from the Yellow River, resulting in greater environmental risk pressure. River level (C10) shared a similar trend with C8 on risk pressure. The main reason was also subject to geographical conditions. In Lanzhou, only the terrain around the Yellow River is flat and the area was large enough, but the remaining rivers were limited by factors such as narrow area or large undulating terrain, which make it difficult to attract enterprises to relocate. In terms of other risk pressures, none of the industries showed obvious pressure [15,17,27,28,29,30,31,32].

3.5. Identification of Hazards with Potential Risks against the Safety of the Yellow River

In this study, 64 kinds of hazards were identified in 70 enterprises as having potential risks against the safety of the Yellow River in this study. Their species and amount were summarized in Table S4. At present, the most hazardous chemicals in the Lanzhou section of the Yellow River were petrochemical products such as aromatic compounds and petroleum hydrocarbons, while the overall stock of other highly hazardous inorganic compounds such as sulfuric acid and hydrochloric acid was very small compared with the former and their share was less than 5% of the former. The types and overall stocks of hazardous chemicals were consistent with the industrial pattern of Lanzhou, namely, the overall industrial layout was based on the petrochemical industry. In addition to the obvious preference for the types of hazardous chemicals, the spatial distribution of these hazardous chemicals also exhibited obvious differences. Figure 7 suggested that Xigu District, among all 9 districts/counties, stored the largest quantality of hazards with 299,446.34 tons of storage, accounting for 88.71% of all hazards in Lanzhou. Other counties/districts merely shared 11.39% of the storage. Additionally, Anning District, Honggu District, and Gaolan County had large hazards storage, while other counties/districts stored less than 300 tons of hazards. The huge chemical inventory in Xigu District is related to its own numerous petrochemical-heavy industries. The results of the analysis of the existing hazardous chemical types and various types of inventories are shown in Figure 7. It can be seen that the existing hazardous chemicals in Xigu District are divided into four categories: aromatic aliphatic hydrocarbon compounds, explosive chemicals, acid/base chemicals, and strong oxidants/highly toxic chemicals. Among them, aromatic aliphatic hydrocarbon compounds and explosive chemicals accounted for the main part, accounting for 198,251.3 tons (66.21%) and 94,220.9 tons (31.07%), respectively. Aromatic aliphatic hydrocarbons were dominated by xylene/aniline/methanol/butyl acrylate/styrene, and their stocks equaled over 10,000 tons. Among the explosive chemicals, the stocks of nitrobenzene and ammonium nitrate were the most, and their stocks were also over 10,000 tons. Although the proportion of acid/base chemicals and strong oxidant/highly toxic chemicals was small, only 1.3% and 1.43%, their absolute stocks were huge, 3894.95 tons and 4274.16 tons, respectively. The stock has exceeded the total stock of various hazardous chemicals in six districts and counties including Yuzhong County and Yongdeng County. The distribution of major hazardous chemicals in Xigu District was consistent with the characteristics of pollutants investigated in previous studies [12,14,17].

3.6. Reasons for Environmental Risks Formation and Corresponding Suggestions

The sources of the main environmental risks against the Yellow River in Lanzhou can be divided into three categories. The first was the nature of heavy industry in which the dense distribution in Lanzhou accounted for a large part of Lanzhou industry. Metal mining and dressing, metal smelting and rolling processing, coal mining and washing, and the petrochemical pressing industry, as the pivot upholding Lanzhou industry, present high-risk pressure on the water environment [1,14,15,17]. The second came from the geographical limitations of the river basin in Lanzhou, whose small area constrained the locations of enterprises to be no more than 4 km away from the main stream of the Yellow River. Furthermore, the downstream water source was just 10 km away, making a highly vulnerable condition towards environmental risks caused by pressures exerted by enterprises in Lanzhou. The third focused on the weak response mechanism. Recently, pollution from and release of several environmental hazards has exposed the long-established weakness in the control and supervision regime. Hence, we proposed that coordination and preparation in the form of enterprise emergency plans, ensuring emergency material reserves, and frequently monitoring the water quality of enterprise sewage outlets, were the method to make up for the improvement, so as to achieve the control and prevention of potential pollution risks of enterprises. As for the small number of venture enterprises, which were mainly located in Gaolan County, Yuzhong County, Yongdeng County, and Lanzhou New District of the tributaries of the Yellow River, it was necessary to control the venture enterprise industry away from the water sources and densely populated areas of the tributaries of the Yellow River.

4. Recommendation

Based on our results, we proposed some recommendations for environmental risk control from enterprise:

(1)

Accountability should also be built. The negligence of outlet monitoring has not been corrected for a long time, and thus the potential risks remain. This situation is mainly due to the absence of accountability.

(2)

Furthermore, Lanzhou needs to strengthen the supervision of potential high-risk and existing medium-risk enterprises in the Yellow River Basin, build an active prevention and control risk mechanism with environmental liability insurance and damage assessment as the core, formulate emergency plans for unexpected release of pollutants of high-risk enterprises, and specifically implement relevant emergency pollution response drill activities.

5. Conclusions

This study systematically assessed the potential pollution risks of 70 enterprises in the Lanzhou section of the Yellow River Basin from the perspectives of administrative regions, types of enterprises and hazardous chemicals, and summarized the factors affecting high-risk pressure. Generally, the water pollution risk pressure of the Lanzhou section of the Yellow River was within control levels and there were no high-risk enterprises. From the perspective of administrative divisions, Xigu District had the largest number of enterprises and the largest stock of hazardous chemicals. With the exception of five impact factors that did not exert any obvious pressure (C4, C6, C9, C11, and C12), the remaining nine indexes presented some degree of risk pressure. Anning District, Honggu District, and Gaolan County also showed certain risk pressure, while other districts and counties had less risk pressure. Furthermore, risk pressure concentrated on C5, C7, C8, and C10, especially C7. Almost all enterprise risk pressures showed great pressure in this index. It was very important to improve the monitoring and management mechanism of sewage discharge from enterprises. In addition, the above four items belonged to the control mechanism and risk recipient, respectively, so more attention needs to be paid to the main risk pressure of the Lanzhou section of the Yellow River to the control mechanism and risk recipient. Upon hazardous chemicals, aromatic aliphatic hydrocarbons and explosive chemicals were present in a huge stock, and accounted for more than 90% of the entire hazards’ storage in Lanzhou. Xigu District was the primary area for their storage, so there was a major risk and pressure in Xigu District on the water environment safety of the Yellow River section of Lanzhou.