Research Framework for Ecosystem Vulnerability: Measurement, Prediction, and Risk Assessment

Yajun WANG; Lifang ZHONG

doi:10.5814/j.issn.1674-764x.2020.05.007

Target layer

Index layer

Research object

Literature sources

Exposure, sensitivity, adaptability

Population pressure, environmental degradation, industrial pollution, income difference, natural background, economic strength, infrastructure, fiscal expenditure, social security

Loess plateau area

Ma et al., 2019

Ecological environment, natural resources, social economy, sustainability

Vegetation index, soil organic matter, rainfall, water resources, engel coefficient, gross domestic product(GDP) index, natural population growth rate, contribution rate of tertiary industry

Hilly mining area

Liu et al., 2018

Pressure, sensitivity, stability

Population pressure, social pressure, environmental pressure, economic pressure, desertification sensitivity, salinity sensitivity, function, vitality, elasticity, structural limitations

Turpan area

Pei et al., 2015

Natural quality, anthropogenic pressure

Precipitation rate, drought index, soil depth, soil parent material, vegetation coverage, population growth rate, population density

Italy

Salvati et al., 2013

Causes and results

Land use type, annual average precipitation, temperature, humidity, population density, per capita income, cultivated land area

Qinghai-Tibetregion

Zhou et al., 2011

Ecological pressure, ecological sensitivity, ecological resilience

Inverse fractal dimension, disturbance index, terrain index, soil sensitivity index, dominance, fragmentation

Wugong mountain

Zhang et al., 2018

Primary index

Secondary index

Tertiary index

Indicators of Natural factor

Terrain, climate, soil, vegetation, geology, water resources

Vegetation coverage rate, terrain distribution, precipitation, soil type, soil erosion rate, total water resources

Indicators of social factors

Social development index

Population density, natural population growth rate, per capita arable land area, urbanization level, poverty rate, unemployment rate, school enrollment rate

Indicators of economic factors

Economic development index

Per capita GDP, proportion of primary industry, proportion of secondary industry, per capita net income of farmers, consumer price index, industrial wastewater discharge, energy consumption

Measurement model

Model content

Model evaluation

Scope of application

PSR (pressure- state-response) model

Pressure indicators based on the effects of human economic and social activities on the environment, the status quo of ecosystem and natural environment represented by the status indicators, and the response indicators are established to prevent the negative impacts of human activities on the environment

Three basic questions “what happened, why did it happen and how to do it” are answered, which fully explain the situation of the evaluation object compared with the reference standard

Applicable to regional environment, soil and water resources and agricultural protection

VSD (exposure-sensitive- adaptation) model

Vulnerability is studied from three dimensions: exposure degree, sensitivity and adaptive potential. Each indicator is refined with circle-level data, and evaluated effectively and clearly by “aspect layer—index layer—parameter layer”

It fails to clarify which reflects the natural factors, and which reflects the human factors

Suitable for the basic data of comprehensive regional EV measurement

Pressure sensitivity resilience model

The intensity of ecological pressure includes area-weighted average fractal dimension reciprocal and disturbance indexes. The ecological sensitivity includes soil erosion sensitivity index, terrain index and landscape fragmentation index. The ecological resilience refers to the self-resilience of an ecosystem and when it is disturbed, it is related to the stability of its internal organizational structure

The emphasis is put on the natural factors, and the proportion of human factors in the index is not high

The vulnerability of ecologically fragile areas are measured and compared

Fuzzy evaluation method

Establish the index system and weight, calculate factors for the membership of each evaluation index vector, evaluate regional vulnerability degree

Fuzzy trigonometric functions can reduce the shortcomings of subjective effects, which has certain objectivity, but the index of significance is not obvious, and is a heavy workload

Suitable for a specific areas or multiple regions

Analytic hierarchy process (AHP)

Establish the evaluation index, score and weight the index, multiplied by the score value and weight, which are added up to obtain the total score to determine the degree of ecological vulnerability

It provides a clearer idea and logic for selection of related indices. The index selection is subjective

Suitable for the analysis of regional and internal evaluation units

Principal component method

Data standardization, set up the correlation coefficient matrix, calculate eigenvalues and eigenvectors and cumulative contribution rates, and obtain the main ingredients of vulnerability analysis

Variable selection of dimensions is not restricted, but it particularly focuses on the main ingredients, which causes some information to be missed, and fails to fully reflect the index of all information

Suitable for regional analysis of the internal evaluation unit

BP neural network method

Set the objective function of the calculated index, and the weight between the input and output layers of the index can be adjusted and modified with the gradient descent method

Intervention processing, compatibility

Deals with the measurement of regional ecological vulnerability with some complex states

Cause-result evaluation method

Establish the corresponding index system according to the characteristics and causes of EV, and the entropy weight method is usually used to assign the weight to each index

Relatively simple, and difficult to deal with complex state

Used for the comparison of vulnerability degrees between regions for a rough analysis

Set pair analysis

Establish coefficient of difference degree and correlation degree, weights set, scheme set and evaluation set, the standard deviation classification method is used to measure and classify the EV

The calculation is complicated, and the analysis result has some intuitiveness

Can be used to measure and analyze the EV of regional units

Landscape ecology model

Computer simulation data is used to characterize the dynamic characteristics of EV, which is combined with GIS, remote sensing data and other system analysis data

This model focuses on local spatial analysis and ignores the influence of human factors

Analysis of EV from the perspectives of regional space and spatial heterogeneity

Grey relational degree

The reference sequence of ecological vulnerability characterization and the comparative sequence of influencing system behavior are determined, the data are processed dimensionless, the grey correlation coefficients of the reference sequence and the comparative sequence are calculated, and the correlation degree is sorted

The degree of correlation between vulnerability factors is emphasized

Used for comparative analysis between regions

Matter-element extension model

The classical domain, node domain and object element to be evaluated are determined, the index weight is set, the correlation degree is calculated, and finally the vulnerability degree of the member to be evaluated is obtained

This method is suitable for multi-factor analysis, which uses formal language to deal with the characteristics of ecological vulnerability

Used for delicate analysis of fragility between regions

Qualitative prediction

Quantitative prediction

Method

Evaluation

Method

Evaluation

Deduction method

Based on the past and current data, the future trends of EV can be deduced. At the macro level, the prediction has good applicability in the case of low precision and missing data

Life zone model

Select a specific model according to the vulnerability data, input parameters and constraints, and simulate the evolutionary trend of the ecological vulnerability. It has the limitation of a high requirement of data quality, which requires strict discrimination, otherwise significant errors may exist in trend judgment

Scenario analysis model

It can be used for scenario prediction of greenhouse gas emission and concentrations, and prediction of the change trend of climate vulnerability. It has the limitation that it can only conduct simulations and predictions based on the natural ecosystem, and rarely involves the social and economic ecosystems

Logistic regression method

According to the law of succession, the trend characteristics of data are used for prediction. It has the shortcoming of excessive dependency on the subjective evaluation of the principles, which does not have a high requirement for the quality and prediction of data

Neural network model

With certain accuracy, it can efficiently process noisy and incomplete data and non-linear complex systems. However, this model does not have strong interpretability for simple ecosystems

Method

Content and evaluation

Risk assessment index method

Based on the subjective evaluation method, the severity and possibility of EV are determined according to the evaluation purpose, and an expert questionnaire is compiled for scoring, so as to determine the possibility and risk level of adverse effects. It has certain subjectivity, which is applicable to fragile ecosystems affected by both human and natural factors. The model has the limitation that the subjective factor is too strong, so it cannot be evaluated in an objective and effective way

Risk synthesis index method

The relationship between the landscape structure and the regional EV risk is established, and the vulnerability risk index is determined by the area proportions of landscape components and landscape loss index. It is mainly used to analyze complex ecosystems that are more strongly influenced by natural factors than by human factors. The shortcoming is that it can only analyze simple ecosystems under the influence of natural factors, while it does not work for complicated ecosystems with less interference of human factors

Risk causal chain model

Three-dimensional models are established by identifying the risk receptors, exposure - response processes and ecological endpoints. Risk = risk probability ×vulnerability ×degree of loss. It includes the explicit characterization of the system exposure response process, the sensitivity of loss as a loss correction factor, and a more comprehensive reflection of the vulnerability pattern of the ecosystem. The main limitation is that it is too objective, and so it lacks a certain subjective judgment

“Probability-loss” two-dimensional model

Determines the probability and consequences of ecological vulnerability events. Risk = probability of risk × outcome. The method is simple, but it does not consider whether the ecosystem is affected by the risk source or its sensitivity

Yajun WANG, Lifang ZHONG. Research Framework for Ecosystem Vulnerability: Measurement, Prediction, and Risk Assessment[J]. Journal of Resources and Ecology, 2020, 11(5): 499

Table 1.
Literature review of EV measurement and prediction indicators

Table 1.
Literature review of EV measurement and prediction indicators

Table 2.
The EV index system

Table 2.
The EV index system

Table 3.
Brief evaluation of EV measurement models

Table 3.
Brief evaluation of EV measurement models

Table 4.
Prediction methods and evaluation of EV

Table 4.
Prediction methods and evaluation of EV

Table 5.
Risk assessment methods and evaluation of EV

Table 5.
Risk assessment methods and evaluation of EV

Table 1. Literature review of EV measurement and prediction indicators

Table 1. Literature review of EV measurement and prediction indicators

Table 2. The EV index system

Table 2. The EV index system

Table 3. Brief evaluation of EV measurement models

Table 3. Brief evaluation of EV measurement models

Table 4. Prediction methods and evaluation of EV

Table 4. Prediction methods and evaluation of EV

Table 5. Risk assessment methods and evaluation of EV

Table 5. Risk assessment methods and evaluation of EV

Table 1.
Literature review of EV measurement and prediction indicators

Table 1.
Literature review of EV measurement and prediction indicators

Table 2.
The EV index system

Table 2.
The EV index system

Table 3.
Brief evaluation of EV measurement models

Table 3.
Brief evaluation of EV measurement models

Table 4.
Prediction methods and evaluation of EV

Table 4.
Prediction methods and evaluation of EV

Table 5.
Risk assessment methods and evaluation of EV

Table 5.
Risk assessment methods and evaluation of EV