Keywords

1 Introduction

When ecosystem is intervened by outside, it will make sensitive reactions and self-recovery reactions, and this is ecological vulnerability [1]. China’s farming–pastoral zone is located in the monsoon climate zone of Southeast Asia, and with large precipitation fluctuations, has characteristic of fragile ecology. Yulin City is an area where ecology is particularly fragile. The ecological environment research of Yulin and the whole farming–pastoral zone has attracted extensive attention from scholars. However, the main method for ecological assessment of Yulin is single-factor evaluation method [2,3,4,5,6]. And a few methods for ecological vulnerability evaluation still have problems of large correlation between indicators and strong subjectivity of weight determination [7,8,9]. These years, remote sensing is used in ecological research, and it makes data acquisition simpler and data types more abundant [10,11,12]. This paper uses remote sensing, meteorological and statistical yearbook data, principal component analysis (PCA) and analytic hierarchy process (AHP) to obtain ecological vulnerability index and analyzes the overall status, spatial distribution and causes of ecological vulnerability of Yulin.

2 Study Area

Yulin is located in the northernmost part of Shanxi Province, between N 36°57′ –N 39°35′ and E 107°28′–E 111°15′. It has typical semiarid and arid climate, and the average annual precipitation is close to 400 mm. There are three landforms: windblown grass beach, loess hilly and gully, and beam-shaped low hills. The mineral resource is diverse and rich in total. It is an important energy export city in China. Special climate, topography, soil texture and high-intensity human activities make it a typical ecologically fragile area in the northern farming–pastoral zone.

3 Methods and Data

3.1 Construction of Ecological Vulnerability Indicator System

This paper constructs an evaluation indicator system from three aspects: ecological sensitivity, ecological restoration and ecological stress [13, 14]. Taking into account the availability of data, 13 indicators were chosen to constitute the indicator system. The indicator system is shown in Fig. 1.

Fig. 1
figure 1

Ecological vulnerability assessment indicator system

Full size image

3.2 Evaluation Indicator Selection and Weight Determination

3.2.1 Standardization of Primary Selection Indicators

We standardize the primary selection indicators to make them be unified to 0–1, so that the values are closer to 1 indicating that the ecology is more fragile [4]. Eight indicators of temperature, elevation, slope, aspect, soil erodibility factor, soil erosion intensity, GDP density and population density are positively correlated with ecological vulnerability; landscape diversity index, vegetation coverage, precipitation, relative humidity and GPP are negatively correlated with ecological vulnerability. The standardized formulas are shown in Eqs. 1 and 2, respectively.

$${\text{posively}}\,{\text{correlated:}}\quad P^{{ * {\text{i}}}} = \left( {P_{i} - P_{{\text{min}}} } \right)/\left( {P_{{\text{max}}} - P_{{\text{min}}} } \right)$$
(1)
$${\text{negatively}}\,{\text{correlated:}}\quad P^{{ * {\text{i}}}} = 1 - \left( {P_{i} - P_{\text{min}} } \right)/\left( {P_{\text{max} } - P_{\text{min} } } \right)$$
(2)

\(P^{*i}\) represents the value after the normalization of the ith evaluation indicator, \(P^{i}\) represents the value before the ith evaluation indicator is normalized, \(P_{{\text{max}}}\) represents the maximum value of the ith evaluation indicator, and \(P_{{\text{min}}}\) represents the minimum value of the ith evaluation indicator [4].

3.2.2 Selection of Evaluation Indicators

We use the PCA to analyze 13 standardized primary evaluation indicators [15,16,17,18,19]. Table 1 shows the results of sorting the principal components according to the contribution value. The cumulative contribution rate of the first five is 86.81%, which is more than 85%, indicating that the first five principal components can cover most information of the primary selection indicators, so the first five principal components are selected as evaluation indicators. Table 2 shows eigenvectors corresponding to each principal component.

Table 1 Principal component contribution rate and cumulative contribution rate
Full size table
Table 2 Eigenvectors corresponding to each principal component
Full size table

3.2.3 Determination of Indicator Weights

We use AHP to determine indicator weights. The assignment principle of the pairwise comparison matrix is shown in Table 3. The larger the value, the greater the influence of the ith element on the previous indicator layer compared to the jth element.

Table 3 Assignment of pairwise comparison matrix
Full size table

This paper constructs a pairwise comparison matrix based on the contribution rates of the first five principal components. Equation 3 is proposed to calculate relative importance scale, \(A_{ij}\) represents the relative importance scale of elements i and j, and \(\alpha_{i}\) and \(\alpha_{j}\) represent the contribution rates of elements i and j. A is the pairwise comparison matrix constructed. After calculation, the weights of the evaluation indicators are: 0.5350, 0.2868, 0.0853, 0.0555, 0.0374.

$$A_{ij} = \frac{{(\alpha_{i} - \alpha_{j} )}}{3.4086} + 1,(i < j)$$
(3)
$$A = \left| {\begin{array}{*{20}c} {\begin{array}{*{20}c} {\begin{array}{*{20}c} 1 \\ {1/3} \\ \end{array} } \\ {1/7} \\ {\begin{array}{*{20}c} {1/8} \\ {1/9} \\ \end{array} } \\ \end{array} } & {\begin{array}{*{20}c} {\begin{array}{*{20}c} {\begin{array}{*{20}c} 3 \\ 1 \\ \end{array} } \\ {1/5} \\ {\begin{array}{*{20}c} {1/6} \\ {1/7} \\ \end{array} } \\ \end{array} } & {\begin{array}{*{20}c} {\begin{array}{*{20}c} {\begin{array}{*{20}c} 7 \\ 5 \\ \end{array} } \\ 1 \\ {\begin{array}{*{20}c} {1/2} \\ {1/3} \\ \end{array} } \\ \end{array} } & {\begin{array}{*{20}c} {\begin{array}{*{20}c} {\begin{array}{*{20}c} 8 \\ 6 \\ \end{array} } \\ 2 \\ {\begin{array}{*{20}c} 1 \\ {1/2} \\ \end{array} } \\ \end{array} } & {\begin{array}{*{20}c} {\begin{array}{*{20}c} 9 \\ 7 \\ \end{array} } \\ 3 \\ {\begin{array}{*{20}c} 2 \\ 1 \\ \end{array} } \\ \end{array} } \\ \end{array} } \\ \end{array} } \\ \end{array} } \\ \end{array} } \right|$$

3.3 Evaluation Model

The five evaluation indicators are graded by the natural discontinuity grading method shown in Table 4. The higher the level, the higher the ecological vulnerability.

Table 4 Indicators’ grading results
Full size table

In Eq. 4, EVI represents the ecological vulnerability index; \(w_{i}\) represents the weight of the ith evaluation indicator; \(f_{i}\) represents the level of the ith evaluation indicator [4].

$${\text{EVI}} = \mathop \sum \limits_{i = 1}^{n} w_{i} f_{i}$$
(4)

3.4 Data Sources and Preprocessing

The soil erosion intensity is calculated according to RUSLE model (modified general soil loss model) based on the above data [20,21,22]. Thirteen indicators are unified into 1 km resolution, UTM_Zone_49N (Table 5).

Table 5 Data sources
Full size table

4 Results and Analysis

4.1 Overall State of Ecological Vulnerability

The EVI is classified as shown in Table 6. Yulin has the largest area with severe vulnerability, accounting for 25.7% of the city’s area. Moderate and above vulnerability account for nearly 70% of the city’s area, and only 20% of the city’s area is micro-vulnerability.

Table 6 Division of ecological vulnerability levels in Yulin
Full size table

4.2 Spatial Distribution of Ecological Vulnerability

The distribution of ecological vulnerability levels is shown in Fig. 2. The proportion of vulnerability levels in various districts and counties is shown in Fig. 3. The ecological vulnerability level of Yulin is characterized by high north and low south. Fugu, Shenmu, Yuyang and Dingbian are most fragile. These four counties have 20–45% extremely vulnerable areas. Especially in Shenmu, extremely vulnerable areas account for 45% of the county’s area.

Fig. 2
figure 2

Spatial distribution of ecological vulnerability levels in Yulin City

Full size image
Fig. 3
figure 3

Statistics on ecological vulnerability levels of various districts and counties in Yulin City

Full size image

4.3 Analysis of the Causes of Ecological Vulnerability

Figure 4 depicts the standardized evaluation indicators of various districts and counties in Yulin. The closer the evaluation indicator is to 1, the greater the vulnerability of the county. The main causes of ecological vulnerability in Dingbian are single landscape, low vegetation coverage and low precipitation. The reason for the ecological vulnerability of Fugu and Shenmu is human production activities and single landscape types. These two counties develop coal industry vigorously, and this leads to a decline in landscape diversity and increase in ecological vulnerability. The main reason for the ecological vulnerability of Yuyang is the landscape type is not rich enough and coupled with the pressure of human activities on the environment, and vegetation coverage is low. The main reason for the ecological fragility in Jingbian is that the landscape type is not rich and the precipitation is small. The 5/6 population in Jingbian is agricultural population, and less precipitation in arid areas will restrict the growth of crops and reduce the stability of the agricultural and livestock ecosystem. Vegetation coverage and aspect are the main reasons for the ecological fragility of Hengshan. There is a large area of half-sunny slope in Hengshan, and water evaporation on sunny slope is fast, which is not conducive to vegetation growth. One of the most obvious features of the six southern counties is the large K value and high vegetation coverage, and the vegetation has a tightening effect on the soil thus decreasing EVI.

Fig. 4
figure 4

Evaluation indicators after standardization in various districts and counties of Yulin City

Full size image

5 Summary

This paper processes remote sensing, and meteorological and socioeconomic data, and uses PCA and AHP to calculate EVI. The main research results were as follows:

  1. 1.

    The method of combining PCA and AHP is used to construct the EVI model, which reduces the correlation between indicators and improves the objectivity of weight determination.

  2. 2.

    Obtain the distribution map of vulnerability level. It is found that the ecological of six northern counties is more fragile than six southern counties. The extremely fragile areas are Fugu, Shenmu, Yuyang and Dingbian.

  3. 3.

    The main causes for the ecological vulnerability of the farming–pastoral zone in Yulin were economics, landscape diversity, annual precipitation and vegetation coverage.