Ecological performance assessment of ecologically fragile areas: a perspective of spatiotemporal analysis

Abstract

Sustainable development in ecologically fragile areas (EFAs) has faced significant challenges in recent years, but the traditional analytical approaches fail to provide an ideal assessment for ecological performance due to spatiotemporal variability in EFAs. This paper evaluates the ecological performance of EFAs based on a modified ecological footprint model, and ecological footprint intensity (EFI) is considered an essential indicator to measure ecological performance, especially for EFAs. Empirically, taking the Π-shaped Curve Area in the Yellow River basin of China as the study area, the spatiotemporal heterogeneity of EFI of 17 cities in the area is analyzed. Then, the extended STIRPAT and geographically and temporally weighted regression (GTWR) models are employed to explore the spatiotemporal heterogeneity of the factors driving EFI. The results show that from 2006 to 2019, the overall level of EFI in the area has decreased; EFI of the area offers a significant spatial agglomeration effect; results of the GTWR model suggest that factors driving EFI have spatiotemporal heterogeneity; the impact of population size, openness, marketization, technology, industrial structure rationalization, and information communication level on EFI was two-sided, while that of affluence, government scale, environmental regulation, and industrial structure advancement show inhibitory impact with the intensity of inhibition varying across periods and cities.

Appendices

Appendix 1 Details of the EF account

Table

Table 5 Information about the EF account

5

It should be noted that the EF of biomass is the ecological footprint of local biomass production, while the EF of fossil energy is its consumption footprint (Luo et al. 2018). Therefore, biomass production is used to calculate the EF of biological resources, while energy consumption is used to calculate the EF of energy. Besides, pollution discharge and energy consumption are treated separately due to the particularity of the EF accounting methods. For pollution discharge, industrial solid waste and domestic waste are converted into arable land according to 109,100 t/hm²; industrial wastewater and urban wastewater are converted into water area according to 365 t/hm²; industrial sulfur dioxide and industrial smoke dust are converted into forest land at the rate of 0.08865 t/hm² and 10.11 t/hm², respectively (Li et al. 2022). The calculation formula of EF of pollutants is as follows:

$${\mathrm{EF}}_{\mathrm{pollutant}}=\sum\nolimits_{i=1}^{3}{r}_{i}\times \frac{{Q}_{i}}{{AC}_{i}}$$

(13)

where r_i is the equivalence factor, Q_i refers to the emission of pollutants, AC_i is the ability of different lands to absorb pollutants, and i represents different pollutants. The EF of energy consumption is calculated according to the method proposed by Sun and Wang (2022), which is expressed as follows:

$${\mathrm{EF}}_{\mathrm{energy}}=\sum\nolimits_{j=1}^{4}{r}_{j}\times \frac{{R}_{j}\times 293080}{{V}_{j}}$$

(14)

where r_j is the equivalence factor, R_j is the total energy consumption (10,000 t of standard coal), V_j represents the global average energy footprint, and j =1,2,3,4 represent coal, natural gas, petroleum, and electricity, respectively.

Appendix 2 Description of driving factors

Table

Table 6 Description of driving factors

6

Appendix 3 Calculation of the spatiotemporal weight matrix

Generally, the Gaussian kernel function is used to define the spatial weight matrix, and the formula is as follows:

$${w}_{ij}=\mathrm{exp}(-{(\frac{{d}_{ij}}{h})}^{2}$$

(15)

where ${d}_{ij}$ represents the spatial distance between regions i and j and $h$ represents the bandwidth, which refers to the non-negative attenuation parameter of the functional relationship between weight and distance and is calculated by the cross-validation (CV) method according to the criterion of minimizing the sum of squared errors. The formula of the CV method is as follows:

$$\mathrm{CV}=\frac{1}{n}\sum\nolimits_{i=1}^{n}{[{y}_{i}-{\widehat{y}}_{\ne i}(h)]}^{2}$$

(16)

where ${y}_{i}$ is the predicted value and ${\widehat{y}}_{\ne i}(h)$ represents the function of bandwidth $h$, and the bandwidth $h$ takes the corresponding value of the minimum CV value.

For the measurement of spatiotemporal distance (d^ST), spatial parameter $\tau$ and temporal parameter $\mu$ need to be introduced to balance dimensional differences in time and space. The formula of d^ST is as follows:

$${d}_{ij}^{\mathrm{ST}}=\sqrt{\tau \left[{\left({u}_{i}-{u}_{j}\right)}^{2}+{\left({v}_{i}-{v}_{j}\right)}^{2}\right]+\mu {({t}_{i}-{t}_{j})}^{2}}$$

(17)

Combining Eqs. (13) to (15), ${w}_{ij}$ in the spatiotemporal weight matrix can be obtained:

$${w}_{ij}=\mathrm{exp}(-{\left(\frac{{d}_{ij}^{\mathrm{ST}}}{h}\right)}^{2}=\mathrm{exp}\left[-\left(\frac{\tau \left[{\left({u}_{i}-{u}_{j}\right)}^{2}+{\left({v}_{i}-{v}_{j}\right)}^{2}\right]}{{h}^{2}}+\frac{\mu {({t}_{i}-{t}_{j})}^{2}}{{h}^{2}}\right)\right]$$

(18)

Appendix 4 Descriptive statistics of relevant data

Table

Table 7 Descriptive statistics of relevant data

7

Appendix 5 Temporal and spatial pattern of EFI of four subcategories

Figure

Fig. 8

a–d Temporal and spatial pattern of EFI of four subcategories

8

Appendix 6 Descriptive statistics of regression parameters of the GTWR model

Table

Table 8 Descriptive statistics of regression parameters of the GTWR model

8