Ecotoxicology (2011) 20:1131–1140 DOI 10.1007/s10646-011-0618-0

Fuzzy synthetic model for risk assessment on Haihe River basin Jingling Liu • Qiuying Chen • Yongli Li Zhifeng Yang

•

Accepted: 21 February 2011 / Published online: 6 March 2011  Springer Science+Business Media, LLC 2011

Abstract A comprehensive indicator model for risk assessment and a multiple-level theoretical indicator system of the water quality–quantity-ecosystem (WQQE) for the Haihe River basin were constructed in this research. A fuzzy optimization model was used to assess risks for the four water systems of the Haihe River basin, and their risk order from high to lower risk was southern Haihe River system (SH), northern Haihe River system (NH), Tuhaimajiahe River system (TH) and Luanjiyanhai River system (LJ). The highest risk value (SH) was 0.8737. In terms of the WQQE, the secondary parameters for assessment of the four water system risks were 0.3579, 0.7226, 0.9547, and 0.5428 respectively. The results indicated that the main control factors for pollution for LJ, TH, SH and NH differed from each other and involved pollutant level, development of water resources, water flow and quality, ecosystem health and the hydrologic structure. Keywords Haihe River basin  Risk assessment  Fuzzy synthetic model  Indicator system

Introduction Risk assessment is a tool used to organize, structure and compile scientific information in order to help identify existing hazardous situations, anticipate potential problems, establish priorities and provide a basis for regulatory

J. Liu (&)  Q. Chen  Y. Li  Z. Yang State Key Joint Laboratory of Environmental Simulation and Pollution Control, School of Environmental, Beijing Normal University, NO.19 Xinjiekouwaida Street, Haidian District, Beijing 100875, China e-mail: [email protected]

controls and/or corrective actions (World Health Organization 2004). In recent years, the risk assessment process used for determining the need for remediation or redevelopment actions at contaminated sites has become well developed (Rı´o and Gracia 2009; Chapman et al. 2010) and documented (US EPA 1996, 1998; National Research Council (NRC) 1983). Although large amounts of toxic substances may be of major concern, simply detecting a hazardous chemical in the environment should not necessarily be a cause of alarm, depending on the chemicals’ bioavailability and the risk of human exposure (Eduljee 2000). Many environmental problems impact large geographic areas (Yu et al. 2010), and therefore regional-scale (as opposed to local) risk assessments are needed to evaluate the risks these problems pose. Major River basins are heterogeneous and complex groups of ecosystems that serve very important roles in a country’s sustainable development strategy (Yang et al. 2006). Land-based activities such as industrial production and agriculture, combined with population growth and heavy urbanization, bring increasing pressures onto the aquatic environment, including the degradation of wetland habitats (Brazner 1997) and an increased health risk for the natural biota and the human population (European Environment Agency (EEA) 1999; Gomez-Gutierrez et al. 2007). Prior risks assessments on water have tended to focus on quality (Doria et al. 2009; Liu et al. 2009; Wang et al. 2009; Li et al. 2010), and generally lack consideration of the total ecosystem, which can lead to an assessment that does not reflect the causal mechanisms that contribute to environmental degradation of the water basin and consequent environmental health risks. Moreover, these studies have mostly been aimed at risk assessments conducted for individual pollutants (Wcislo et al. 2002; McGraph et al. 2004), or the contaminated site on (Eduljee 2000), a small

123

1132

scale. Liu et al. (2010a) conducted an ecological risk assessment of the Luanhe River basin (sub-basin of Haihe river Basin) based on a relative risk model, and clarified the characteristic of regional risk in sub-basin scale of Haihe River Basin. However, larger-scale assessment studies are still lacking, and assessment theories and methods are being developed. Fuzzy synthetic evaluation (FSE) models based on fuzzy set theory have been used by a number of researchers in a variety of environmental settings (Onkal-Engin et al. 2004; Dahiya et al. 2007; Fisher 2003; Liu et al. 2010b; Feng and Wang 2007; Li et al. 2005). Fuzzy evaluation methods process all the components of a model according to predetermined weights, and decrease the fuzziness by using the membership function. As a result, sensitivity is quite high compared with other index evaluation techniques (Onkal-Engin et al. 2004). In this study, we attempt to use the FSE method to assess the ecological risks of the aquatic environment of the Haihe River basin, the politic, economic, and cultural center of China. The basin is an area facing critical challenges for water resources (Xia et al. 2006). For example, the water resource per capita in the basin is about 305 m3, which is only 1/7 of the national average and 1/24 of the global average. Water shortages have serious ecosystem degradation, including drying-up of river systems, shrinking of lakes and wetlands, and the decreases in inflows into the Bohai Sea. In this study, we have established a conceptual model of the water environment at the basin scale through analysis of the sources and stressors in order to select the most appropriate risk indicators. The entropy method was used to determine the weight of the risk indicator, and combined with an evaluation of the indicator system and the multilevel, multi-objective fuzzy optimization method, water environmental risk assessments were conducted. The

Fig. 1 Location of the Haihe River Basin in China

123

J. Liu et al.

information provides us with a comprehensive understanding of the environmental risks on a basin-wide scale. In addition, a primary objective of the study was to provide the scientific basis for informed risk management decisions.

Materials and methods Study area The Haihe River basin, located between 350 and 430 N latitude and 1120 and 1200 E longitude, is situated with the Bohai Sea on the east, the Yellow River on the south, is adjacent to the Yunzhong and Taiyue Mountains on the West, and the Mongolian Plateau on the north. It is in Hebei Province and includes the two major cities of Beijing and Tianjin; parts of Inner Mongolia autonomous region, and Shanxi, Henan, Liaoning and Shandong Provinces. It includes the southern Haihe River system (SH), the northern Haihe River system (NH), the Tuhaimajiahe River system (TH) and the Luanjiyanhai River system (LJ) (Fig. 1). The basin covers an area of 318,000 km2, of which mountains and plateaus comprise 60% and plains comprise 40%. The predominant climate is the Asian Monsoon climate with cold and dry winters and hot and rainy summers. The annual precipitation ranges from 379.2 to 583.3 mm, about 75% of which falls in the months of June to September (Xia et al. 2006). The main characteristics of the Haihe River basin include limited water resources, unequal rainfall distribution over time and space, and the frequent occurrence of drought years. Overall, it is considered an area of water-resource shortages (Wan et al. 2005).

Fuzzy synthetic model for risk assessment

Creation of primary indicator system Selection of receptor and assessment endpoints The receptor is the subject of the risk assessment, and whose risk is used to infer, analyze, or replace the ecological risk for the entire region (US EPA 1996). The receptors should represent sensitive components of the ecosystem that serve an important functional role. Wetlands include rivers, lakes, reservoirs, estuaries and other water bodies. They provide a habitat for a variety of organisms and may have similar plant communities and soil–water environments, however, they have their own characteristics in relation to human development and resource utilization. In this study, the wetlands water environment has been treated as the receptor in the Haihe River basin. Assessment endpoints are defined as the damage to receptors as a result of the action of uncertain risk sources, and the damage to the regional ecosystem structure and function. Assessment endpoints are the link between scientifically measurable endpoints and the risk mitigation objectives of stakeholders and resource managers (Suter 1991, 1993). In general regional ecological risk assessment, the regional or population species that most closely reflects the significance of regional ecological assessment is also often selected the ecological endpoint. In this study, the water quality, the water quantity and the water ecosystems (WQQE) were selected as the assessment endpoints, based on the water environment of the Haihe River basin. Description of sources The risk sources mainly involve human activities as the result of river basin development. The analysis of the risk source involves qualitative, quantitative and distribution analysis for the potential risks from various sources, in order to provide an in-depth understanding to the risks. Major land use/land cover types in the Haihe River basin are farmland (7–48%), grassland (26–60%) and forest (18.8–51.8%). Rainfall often does not meet crop water demand, hence irrigated lands have not only increased, but are increasingly reliant on groundwater pumping. As irrigation significantly enhances harvests, it has therefore become an indispensable agronomic practice in the region. Hence, as the largest water user, agricultural water demand has a significant impact on the scale of water scarcity in the region. About 1,900 reservoirs with a total storage capacity of over 30 billion m3 have been built to supplement agricultural, industrial and domestic water supplies. Exactly which of these factors has the most impact on runoff is the subject of considerable controversy in the hydrological community (Xia et al. 2006). In order to describe the

1133

effects that human development and resource utilization have on the water environment of the Haihe River basin, risk sources which are grouped into four major categories in this study. They are industrial, agricultural, municipal, and water project (major hydrological projects such as reservoir construction), respectively. Division of risk region The Haihe River basin has four sub-basins (Fig. 1), which were divided into SH, NH, TH and LJ mainly based on water resource condition and degree of development and utilization. These four sub-basins were selected as the risk sub-regions. Conceptual model The conceptual models are working hypotheses about how the hazardous agent or action may affect the endpoint entities (Barnthouse and Brown 1994; US EPA 1998). They include descriptions of the source, receiving environment, and processes by which the receptors come to be exposed directly to the contaminants, and indirectly to the effects of the contaminants on other environmental components (Phenrat et al. 2009). In order to analyze and choose the risk indicator, we established the conceptual model (Fig. 2) of ecological risk assessment for the water environment which suitable to the Haihe river basin. Primary indicator system In order to have complete indicator information, as well as to simplify calculations, principal component analysis was applied to the classification and selection of risk sensitive indicators. In the process of screening indicator, we eliminate the indicator by using the Pearson correlation analysis tools of SPSS software, and then the primary indicator system was established. Creation of FSE model Fuzzy optimal evaluation model Assuming that the multi-objective decision making program set is D = (D1, D2… Dn), objective set is G = (G1, G2… Gm). The fuzzy matrix X can be expressed as: 2 3 x11 x12    x1n 6 x21 x22    x2n 7 6 7 ¼ xij X¼6 . ð1Þ .. .. 7 4 . 5 . . . xm1 xm2    xm3

123

1134

J. Liu et al.

Fig. 2 Conceptual model of ecological risk assessment for basin water environment

Industry risk source

Agriculture risk source

Municipal risk source

Water project risk source

Paper-making industry,

Aquaculture,

Urban runoff,

Dam,

Pharmaceutical Industry,

Livestock,

Traffic,

Gates,

Chemical Manufacturing,

Farming,

Hospital waste,

Water diversion,

Electric Power Industry,

Forestry industry,

Solid waste,

Groundwater extraction,

Metallurgical industry,

Sideline production

Domestic sewage

River regulation

Mining industry

Chemical pollution

Toxic pollutants (organic, inorganic);

Hydrology structure changed

Wetlands Shrinkage;

Vertical (Connectivity); Horizontal (river bed evolution);

Eutrophic pollution (Nitrogen and phosphorus);

Ecosystem structure changed

Estuary degradation; Biodiversity loss;

Vertical (permeability);

Water and soil erosion

Flow;

New pollutants (Estrogen)

Water quality

Water quantity

where xij (i = 1, 2… n; j = 1, 2… m) is the membership degree of the ith evaluation parameter at the jth level. The feature indictor values in matrix (1) are converted into the corresponding indicator relative membership degree. For the primary indicators with large values, the membership degree structure can be expressed as: rij ¼

xij  xmin i xmax  xmin i i

For primary indicators with smaller values, the membership degree structure can be expressed as: rij ¼

xmax  xij i max xi  xmin i

Therefore, the objective value of matrix (1) can be converted into an indicator membership degree matrix. 2 3 r11 r12    r1n 6 r21 r22    r2n 7 6 7 R ¼ 6 .. ð2Þ .. 7 .. 4 . . 5 . rm1

rm2



rmn

If we extract the maximum value of each line in matrix (2), it can be recorded as rg = (rg1, rg2… rgm) = (max r1i, max r2i… max rmi) = (1, 1… 1), which is the ideal superior program; If we extract the minimum value of each line in matrix (2), it can be recorded as rb = (rb1, rb2,… rbm) = (min r1i, min r2i… min rmi) = (0, 0… 0), which is

123

Water ecology

the ideal inferior program. Thus, any program j attaches to superior program rg or inferior programs rb to a certain membership degree ugi or ubj. The optimal fuzzy matrix can be expressed as:   ug1 ug2    ugn U¼ ub1 ub2    ubn 2n where 0 B ugi B 1, 0 B ubj B 1, ugi ? ubj = 1, j = 1, 2… n. Let us suppose that the weighting vector of assessment P indicator is k = (k1, k2… km)T, ki = 1. In order to understand the optimal value of the relative membership degree that solves program j, the following optimization criterion was set up:  X 2 min F ¼ u2gi ki ðrij  rgj Þ  ð3Þ X 2 2 ki ðrij  rbj Þ þð1  ugj Þ The weighted euclidean distance of program j from the superior and inferior can be expressed as: sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi m  X 2 Sgj ¼ ugj ki ðrgj  rij Þ i¼1ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi s m  X 2 Sbj ¼ ubj ki ðrij  rbj Þ ; where ubj ¼ 1  ugj i¼1

Fuzzy synthetic model for risk assessment

1135

Setting objective function (3) derivatives and ordering derivatives equal to 0, allows the multi-objective fuzzy optimal model to be expressed by the following formula, where ugj is the decisive superiority. ,( m  m  X X 2 2 ugj ¼ ki ðrij  rbj Þ ki ðrij  rbj Þ i¼1

þ ( ¼

m  X 2 ki ðrgj  rij Þ i¼1

)

i¼1

ð4Þ

"

m  X 2 1þ ki ð1  rij Þ

#,

i¼1

m X

)1 2

ðki rij Þ

i¼1

The fuzzy optimal method of a multi-objective system Assuming that the system was divided into H levels, H is the highest level. If the first layer (bottom) has m-parallel unit systems, the fuzzy matrix can be obtained base on the formula (4): 2 3    1 ugn 1 ug1 1 ug2 6 2 ug1 2 ug2    2 ugn 7 6 7 U¼6 . ð5Þ .. 7 .. 4 .. . 5 . m ug1

m ug2



m ugn

If iugi = rij, then we have the fuzzy matrix of decision making at level 1: 2 3 r11 r12    r1n 6 r21 r22    r2n 7 6 7 R ¼ 6 .. .. 7 .. 4 . . 5 . rm1

rm2



rmn

Assuming that the full vector of the M sub-system is k = (k1, k2… km)T, we can solve the decisive superiority of n programs based on Eq. 4. The above decision making membership degrees can be used to calculate the unit system of level 2. In this way, the calculation of fuzzy optimization was carried out from the bottom to the top, up to the highest level. We can obtain the output of the highest level unit system because of the only one unit system in the highest level, which is the membership degree vector of decisive making or program. uj ¼ ðu1 ; u2 . . .un Þ: After solving the Eq. 5, we can select the most satisfactory outcome for decision making in accordance with the program membership degree from the largest to the smallest. The introduction of relative membership degree reduced the subjective arbitrariness of the membership function to a certain extent, which made the optimal value of membership degree change in a larger scale, also avoided tending

homogenization of differences between the final rankings based on the fuzzy synthetic evaluation method. Determination of indicator weights The size of the indicator weighting has a direct impact on the results of the evaluation, and the weighting values may change the assessment outcome in relation to priority setting (Ding and Qiu 2010). It is therefore very important to choose the appropriate weighting, and several methods are available. At present, two methods to calculate weights of indicator were used usually, they are objective and subjective method. In the typical method of fuzzy synthetic evaluation, some researchers have calculated weights according to the national environmental quality standards, some have distributed weights with an analytic hierarchy process (Fan 1998; Li et al. 2005), and some have distributed weights according to the knowledge and experience of experts. In order to investigate the relationship between the internal elements, also considering the complexity of the basin water environment system, this study selected the objective method that calculate the weights of proportion to determinate indicator weights—entropy method. The entropy method first appeared in thermodynamics, and was introduced into information theory by Shannon (1948). Advantage in using the entropy method is that the approach uses target value or property value to calculate the weight in the evaluation program, which is a more objective weight. According to the nature and characteristics of a given objective Bi (i = 1, 2… m), each objective value was normalized for all programs Dj (i = 1, 2, m; j = 1, 2, n). For the primary indicator value (Bi), when the value is large, the normalized formula is:

Z ij ¼ Zij = Z^iD þ Z^i i ¼ 1; 2. . .r; j ¼ 1; 2. . .n For the more primary indicator (Bi) when the objective value is smaller, the normalized formula is:

Z ij ¼ 1  Zij = Z^iD þ Z^i i ¼ r þ 1; r þ 2. . .m; j ¼ 1; 2. . .n where Z^iD and Z^i are the max value and the min value of each objective. For any given objective Bi, the larger the difference between the objectives (j = 1, 2… n), the stronger the relative intensity of the objective score between different programs. At the same time, the comparison effect of Bi impact on the program will become larger, which means that it becomes more important to the decision-making process. Entropy is a measurement of uncertainty in the system, and can be used as a measurement of the scale of the amount of information. The overall contribution of Bi to all programs used entropy in the following expression:

123

1136

J. Liu et al.

eðBi Þ ¼ K

n n X

 o Zij =Ei In Zij =Ei i ¼ 1; 2. . . m

Table 2 Principle component load matrix after orthogonal circumgyration of water quantity risk pressure

j¼1

p14

Pn

where Ei ¼ j¼1 Zij (i = 1, 2… m), K [ 0; In is the natural logarithm, e (Bi) C 0. If for a given objective i are all equal, then Zij =Ei ¼ 1=n; at this time the maximum of e (Bi) was selected, namely emax ¼ K  Inðn=1:0Þ: If assuming K = 1/In (n/1.0), then 0 B e (Bi) B 1. For any given objective Bi, the smaller the objective Z*ij (j = 1, 2… n) difference, the larger for e (Bi); When all are equal, e (Bi) = emax = 1. Namely, the contribution of the program to the objective had no basic difference. At this time, the objective could not provide useful information for program comparison, hence its weight coefficient was set to zero, and the objective in the further decision-making could be deleted. However, when the contribution of that program for the objectives had a larger difference, the objective impact on the program is greater, the more it contains and transfers decision-making information, and its weight coefficient is greater. Therefore, the weight coefficient k and the entropy e (B) are inversely proportional to the relationship. The objective weights coefficient is defined as follows: " # m X ki ¼ ½1  eðBi Þ= m  eðBk Þ i ¼ 1; 2. . . m k¼1

Results and discussions Indicator system and weights The loading matrix of principle components after orthogonal circumgyration of water quality risk pressure, water quantity risk pressure and water quantity risk status were listed in Tables 1, 2 and 3. Principal component analysis results of water quality risk indicators are as follows: for the first two characteristic roots k1 = 10.784, k2 = 1.783, k3 = 0.433, E = (10.784 ? 1.783 ? 0.433)/13 = 100.000%. Thus, the number of principal components was taken to be 3. After obtaining eigenvalues and eigenvectors, the primary factor loading matrix after orthogonal circumgyration was calculated. The results showed that the main factor in component 1 includes the amount of COD that flows into the river from

p15

p16

p17

p18

p19

p20

p21

Z1 0.939

0.999 0.636 0.269 0.433

0.649 -0.129 -0.201

Z2 0.212

0.024 0.743 0.945 0.266 -0.133 -0.620

0.938

Z3 0.271 -0.022 0.208 0.187 0.861 -0.749 -0.774

0.283

rural domestic sewage, non-point source pollution, chemical fertilizers, livestock, urban surface runoff and soil erosion. As these indicators relate to non-point source pollution, the first principal component can be considered to represent regional non-point source pollution. Principal component 2 was primarily driven by the emissions output from the million yuan industry, and principal component 3 was driven by the per capita emissions of urban waste water. Principal components 2 and 3 primarily represent point source pollution. For water quantity risk indicators, principal component analysis results are as follows: the first two characteristic roots k1 = 4.534, k2 = 2.522, k3 = 0.944, E = (4.534 ? 2.522 ? 0.944)/8 = 100%. The number of principal components was taken to be 3. After obtaining eigenvalues and eigenvectors, the primary factor loading matrix after orthogonal circumgyration was calculated. The results showed that the loading value of agricultural and industrial water use was the largest, followed by domestic water use. In each factor of principal component 2, the largest loading values were the extent of development and utilization of surface water and the proportion of water storage projects supply water to others. The extent of groundwater exploitation was the largest loading value in principal component 3. The original indicator can be classified as three new indicators, namely, the three principal components. According to the loading of relevant factors of each principal component, representative indicators were selected. For example, principal component 1 was indicative of water conditions, and is associated with industrial, agricultural and domestic water consumption; principal component 2 is indicative of surface water resource development and utilization; and principal component 3 is indicative of groundwater exploitation and utilization. Principal component analysis results of water quantity risk status indicators are as follows: the first two characteristic

Table 1 Principle component load matrix after orthogonal circumgyration of water quality risk pressure p1

p2

p3

p4

p5

p6

p7

p8

p9

p10

p11

p12

p13

Z1

0.878

-0.218

0.536

-0.006

0.755

0.702

0.960

0.964

0.964

0.955

0.964

0.962

0.964

Z2

0.469

-0.948

0.768

0.342

0.620

0.630

0.273

0.262

0.262

0.287

0.262

0.264

0.262

Z3

0.096

-0.231

0.352

0.940

0.215

0.331

0.058

0.050

0.046

0.073

0.050

0.064

0.050

123

Fuzzy synthetic model for risk assessment

1137

Table 3 Principle component load matrix after orthogonal circumgyration of water quantity risk status s3

s4

s5

s6

s7

s8

s9

s10

s11

s12

Z1

0.035

-0.169

0.972

0.994

0.027

0.993

-0.274

0.998

0.898

0.158

Z2

0.109

-0.955

0.232

0.032

0.955

-0.117

0.802

0.053

-0.315

-0.596

Z3

-0.933

0.243

0.041

0.107

-0.297

-0.023

0.532

0.022

-0.306

0.787

roots k1 = 4.888, k2 = 3.269, k3 = 1.834, E = (4.888 ? 3.269 ? 1.834)/10 = 100%. The results showed that precipitation, surface water resources, groundwater resources, natural annual runoff, and rainfall infiltration replenishment were the largest loading values. Water resources, surface water availability and groundwater extraction rates were the largest loading values in principal component 2. Drought indicators were the largest loading values in principal component 3. In the same way, principal component analysis results of water ecology risk pressure indicators showed that length and the number of days of rivers drying up, area ratio of groundwater funnel, the proportion of wetland area were the largest loading value. The original indicator can be classified as three new indicators, namely, the three principal components. Principal component 1 was indicative of water resources, and can be characterized by surface and ground water resources; principal component 2 was indicative of per capita water resources; and principal component 3 was indicative of drought. Based on the primary evaluation indicator system and the analysis of these results for the Haihe River basin, the

risk assessment layered indicator system was established (Table 4).The indicator weight results are listed in Table 5. Risk assessment Based on the data from 2000 to 2004, water environmental risk of the 4 sub-regions (LJ, NH, SH and TH) in the Haihe River basin was assessed by using a fuzzy optimization model. According to the decisive superiority for the first level (Table 6), the main controlling factors in LJ were pollution control, water resources and water quantity, with decisive superiority of 0.9707, 0.9831 and 0.9996 respectively. The main controlling factors in NH were water resources, water resources development and the hydrological structure, with decisive superiority of 0.9610, 0.9279 and 0.9977 respectively. The main controlling factors in SH were non-point source pollution and water consumption with decisive superiority as high as 1. In TH, pollution, water quality control, ecological water and ecological control were the main controlling factors, with decisive superiority all equal to 1.

Table 4 Layered indicator system of risk assessment for the Haihe River Basin water environment Layer 3

Layer 2

Layer 1

Water environment risk

Water quality risk

Pollution

River length more than classic iii water (km)

Pollution intensity of point source

Volume of sewage discharge of million industrial output (t/a), urban per capita waste water discharge (t/a)

Pollution intensity of non-point source

Non-point source output COD (104 t/a)

Pollution control

Indicator layer

Proportion of waste water up to the standards for industry (%) Centralized treatment ratio of urban sewage (%)

Water quantity risk

Water ecology risk

Water resource condition Water status

Per capita water resources (m3), amount of surface water resources (108 m3), amount of groundwater resources (108 m3), drought indicator Industrial water (108 m3), agriculture water (108 m3), domestic water (108 m3)

Water resources development

Extent of surface water development (%), extent of groundwater development (%)

Water quality control

Reuse rate of wastewater (%)

Hydrological structure

Length (km) and the number of days (d) of rivers drying up, proportion of wetland area (%)

Soil geology

Area ratio of water and soil erosion (%), area ratio of over-exploitation of groundwater (%) Area ratio of groundwater funnel (%)

Ecological water

Ratio of ecological water (%)

Ecological control

Area of water and soil erosion under control (104 km2)

123

1138

J. Liu et al.

Table 5 Indicator weights of elements of risk assessment of the Haihe River water environment Indicator weights

Indicator

Indicator weights

Water quality risk (0.1928) Pollution

0.0152

River length more than classic iii water (km)

0.0152

Pollution intensity of point source

0.0153

Volume of sewage discharge of million industrial output (t/a), urban per capita waste water discharge (t/a)

0.0092

Pollution intensity of non-point source

0.0733

Non-point source output COD (104 t/a)

0.0733

Pollution control

0.089

0.0061

Proportion of waste water up to the standards for industry (%)

0.0353

Centralized treatment ratio of urban sewage (%)

0.0537

Per capita water resources (m3), amount of surface water resources (108 m3), the amount of groundwater resources (108 m3), drought indicator

0.0116

Water quantity risk (0.4409) Water resource condition

0.1041

0.0470 0.0414 0.0041

Water status

0.2457

Industrial water (108 m3), agriculture water (108 m3), domestic water (108 m3)

0.0738 0.0643 0.1076

Water resources development Water quality control Water ecology risk (0.3663) Hydrological structure

0.0241

Extent of surface water development (%), the extent of groundwater development (%)

0.0106

0.0670

Reuse ratio of wastewater (%)

0.0670

0.0195

Length (km) and the number of days (d) of rivers drying up, the proportion of wetland area (%)

0.0127

0.0135

0.0047 0.0021

Soil geology

0.1149

Area ratio of water and soil erosion (%), area ratio of over-exploitation of groundwater (%)

0.0583

Area ratio of groundwater funnel (%)

0.1019

0.0566

Ecological water

0.0484

Ratio of ecological water (%)

0.0484

Ecological control

0.0816

Area of water and soil erosion under control (104 km2)

0.0816

For total risk values (Table 6) involving indicator weights, the risk order of the four water systems of the Haihe River basin was SH, NH, TH and LJ. The risk value for SH was the highest at 0.8737. In terms of WQQE, the decisive superiorities for the second level (Table 6) were 0.9547, 0.8655 and 0.8506 respectively in SH, which were all the largest values among the four sub-regions, indicating highest water quality, water quantity and water ecology risk. The ecological risk in LJ was the lowest among the four sub-regions. Although the ecological risk in some subbasins was relatively low, they still demonstrated some risk and the risk assessment was not optimistic. As shown in Fig. 3, the distribution of regional risk is clear. The map can help local governments to optimize the distribution of industrial areas, and establish risk prevention measures and emergency management procedures. According to Table 6, water quality risk had the largest decisive superiority in each sub-region, with water quantity risk second in LJ and SH, and water ecology risk second in NH and TH. Pollution

123

was the most serious threat in NH and TJ, with decisive superiorities of 0.9600 and 1 respectively. It appears clear that measures need to be taken to control pollution so as to reduce water quality risks in the Haihe River basin.

Conclusions Regional environmental risk results from the interaction of many environmental risk sources. In this paper, the fuzzy synthetic evaluation method was used to assess the water environment risk. The assessment results have shown that the order of severity of risk of the sub-basins from high to low was SH, NH, TH and LJ. The main controlling factors for pollution for the 4 water systems in the Haihe river basin differed from each other. The major factors were pollution control, water resources and water quantity control in LJ; non-point source pollution, water consumption and water quality control in SH; pollution, water quality

Fuzzy synthetic model for risk assessment

1139

Table 6 Decisive superiority for the first level and the second level LJ

NH

SH

TH

Pollution

0.0000

0.1014

0.9600

1.0000

Point source pollution

0.3396

0.3060

0.0980

0.8893

The first level

Non-point source pollution

0.0000

0.0347

1.0000

0.0505

Pollution control

0.9707

0.0000

0.9062

0.9908

Water resource

0.9831

0.9610

0.0114

0.9562

Water consumption

0.0000

0.6683

1.0000

0.0200

Water resources development

0.4507

0.9279

0.9143

0.0000

Water quantity control

0.9996

0.0000

0.9985

1.0000

Hydrological structure

0.0384

0.9977

0.7259

0.5448

Soil geology Ecological water

0.2082 0.0000

0.8893 0.8360

0.9948 0.9710

0.0395 1.0000

Ecological control

0.0000

0.0385

0.2000

1.0000

Water quality risk

0.3579

0.7226

0.9547

0.5428

Water quantity risk

0.2719

0.5817

0.8655

0.3104

Water ecology risk

0.2259

0.6023

0.8506

0.4265

0.2662

0.5032

0.8737

0.3877

The second level

The total risk value

Fig. 3 Risk levels of four sub-basins of the Haihe River Basin

control, ecological water and ecological control in TH; and water resources, water resources development and hydrological structure in NH. Water quality presented the largest risk in each sub-region relative to water quantity and water ecology. Pollution overall was the most severe risk in NH and TH, whereas non-point pollution was most critical in SH and point pollution in TH. The results of this ecological risk assessment provide us with a comprehensive understanding of the water

environment of the Haihe River basin, and can also be used for risk management in making decisions regarding revitalization options. However, this analysis provides relative risk among the 4 sub-regions. If we want to know the relationship between the risk components and the spatial variation in each sub-region, further research should be carried out. Acknowledgments This work was supported by National Basic Research Program (973 Program) (Grant No. 2006CB403403) of the Ministry of Science Technology, Program for Changjiang Scholars and Innovative Research Team in University (No. IRT0809) of the People’s Republic of China and the NSFC (No.40940012) of the People’s Republic of China.

References Barnthouse LW, Brown J (1994) Conceptual model development. Chapter 3 in ecological risk assessment issue papers, EPA/630/ R-94/009. US Environmental Protection Agency, Washington Brazner JC (1997) Regional, habitat, and human development influences on coastal wetland and beach fish assemblages in Green Bay, Lake Michigan. J Great Lakes Res 23:36–51. doi: 10.1016/S0380-1330(97)70883-9 Chapman EEV, Dave G, Murimboh JD (2010) Ecotoxicological risk assessment of undisturbed metal contaminated soil at two remote lighthouse sites. Ecotoxicol Environ Saf 73:961–969. doi: 10.1016/j.ecoenv.2010.02.014 Dahiya S, Singh B, Gaur S et al (2007) Analysis of groundwater quality using fuzzy synthetic evaluation. J Hazard Mater 147:938–946 Ding JD, Qiu JP (2010) An approach to improve the indicator weights of scientific and technological competitiveness evaluation of Chinese universities. Scientometrics 86:285–297. doi:10.1007/ s11192-010-0268-7 Doria MDF, Pidgeon N, Hunter PR (2009) Perceptions of drinking water quality and risk and its effect on behavior: a cross-national study. Sci Total Environ 407:5455–5464. doi:10.1016/j.scitotenv. 2009.06.031 Eduljee GH (2000) Trends in risk assessment and risk management. Sci Total Environ 249:13–23. doi:10.1016/S0048-9697(99)00507-0 European Environment Agency (EEA) (1999) State and pressures of the marine and coastal Mediterranean environment. Environmental assessment report No 5, Copenhagen, pp 47–75 Fan BD (1998) Fuzzy comprehensive evaluation model for groundwater quality. China Rural Water Hydropower 9:29–32 Feng ZG, Wang Q (2007) Research on health evaluation system of liquid-propellant rocket engine ground-testing bed based on fuzzy theory. Acta Astronaut 61:840–853. doi:10.1016/ j.actaastro.2007.01.071 Fisher B (2003) Fuzzy environmental decision-making: applications to air pollution. Atmos Environ 37:1865–1877. doi:10.1016/ S1352-2310(03)00028-1 Gomez-Gutierrez A, Garnacho E, Bayona JM et al (2007) Screening ecological risk assessment of persistent organic pollutants in Mediterranean Sea sediments. Environ Int 33:867–876. doi: 10.1016/j.envint.2007.04.002 Li XM, Min M, Tan CF (2005) The functional assessment of agricultural ecosystems in Hubei Province, China. Ecol Model 187:352–360. doi:10.1016/j.ecolmodel.2004.09.006 Li YL, Liu JL, Cao ZG et al (2010) Spatial distribution and health risk of heavy metals and polycyclic aromatic hydrocarbons (PAHs)

123

1140 in the water of the Luanhe River Basin, China. Environ Monit Assess 163:1–13. doi:10.1007/s10661-009-0811-2 Liu JL, Li YL, Zhang B et al (2009) Ecological risk of heavy metals in sediments of the Luan River source water. Ecotoxicology 18:748–758. doi:10.1007/s10646-009-0345-y Liu JL, Chen QY, Li YL (2010a) Ecological risk assessment of water environment for Luanhe River Basin based on relative risk model. Ecotoxicology 19:1400–1415. doi:10.1007/s10646-010-0525-9 Liu L, Zhou JZ, An XL et al (2010b) Using fuzzy theory and information entropy for water quality assessment in Three Gorges region, China. Expert Syst Appl 37:2517–2521. doi: 10.1016/j.eswa.2009.08.004 McGraph D, Zhang CS, Carton O (2004) Geostatistical analyses and hazard assessment on soil lead in Silvermines, area Ireland. Environ Pollut 127:239–248. doi:10.1016/j.envpol.2003.07.002 National Research Council (NRC) (1983) Risk assessment in the federal government: managing the process. National Academy Press, Washington Onkal-Engin G, Demir I, Hiz H (2004) Assessment of urban air quality in Istanbul using fuzzy synthetic evaluation. Atmos Environ 38:3809–3815. doi:10.1016/j.atmosenv.2004.03.058 Phenrat T, Liu Y, Tilton RD et al (2009) Adsorbed polyelectrolyte coatings decrease Fe0 nanoparticle reactivity with TCE in water: conceptual model and mechanisms. Environ Sci Technol 43: 1507–1514. doi:10.1021/es802187d Rı´o LD, Gracia FJ (2009) Erosion risk assessment of active coastal cliffs in temperate environments. Geomorphology 112:82–95. doi:10.1016/j.geomorph.2009.05.009 Shannon CE (1948) A mathematical theory of communications. Bell Syst Tech J 27:379–423 Suter GW II (1993) Ecological risk assessment. Lewis, Boca Raton Suter II GW (1991) Screening level risk assessment of off-site ecological effects in surface waters downstream from the U.S.

123

J. Liu et al. Department of Energy Oak Ridge Reservation, ORNL/ER-8. Oak Ridge National Laboratory, Oak Ridge US EPA (1998) Guidelines for ecological risk assessment, EPA/630/ R-95/002F. Risk Assessment Forum, Environmental Protection Agency, Washington US EPA (1996) Proposed guidelines for carcinogen risk assessment. Federal Register 61:16960–18011 Wan Y, Hu JY, Liu JL (2005) Fate of DDT-related compounds in Bohai Bay and its adjacent Haihe Basin, North China. Mar Pollut Bull 50:439–445. doi:10.1016/j.marpolbul.2004.11.037 Wang LL, Yang ZF, Niu JF et al (2009) Characterization, ecological risk assessment and source diagnostics of polycyclic aromatic hydrocarbons in water column of the Yellow River Delta, one of the most plenty biodiversity zones in the world. J Hazard Mater 169:460–465. doi:10.1016/j.jhazmat.2009.03.125 Wcislo E, Ioven D, Kucharski R et al (2002) Human health risk assessment case study: an abandoned metal smelter site in Poland. Chemosphere 47:507–515. doi:10.1016/S0045-6535(01)00301-0 World Health Organization (2004) IPCS harmonization project-IPCS risk assessment terminology, Geneva Xia J, Feng HL, Zhan CS (2006) Determination of a reasonable percentage for ecological water-use in the Haihe River Basin, China. Pedosphere 16:33–42. doi:10.1016/S10020160(06)60023-4 Yang ZF, Liu JL, Sun T et al (2006) Environmental flows in basins. Science Press, Beijing (in Chinese) Yu GM, Feng J, Che Y et al (2010) The identification and assessment of ecological risks for land consolidation based on the anticipation of ecosystem stabilization: a case study in Hubei Province, China. Land Use Policy 27:293–303. doi:10.1016/ j.landusepol.2009.03.004