The study area
Qingdao with the area of 11293km2, the urban area (Shinan, Shibei, Licang, Laoshan, West Coast New Area, Chengyang and Jimo districts) is 5226km2, and the Jiaozhou, Pingdu and Laixi with the total area of 6067km2. The elevation ranged from 0 to 1090m (Figure 1). From 2000 to 2020, the GDP increased from 118.31 to 1240.06 billion yuan, the proportion of secondary industry decreased from 46.1% to 35.2%, and tertiary industry increased by 19%. The urban built-up area and population increased by 639.1km2 and 2939.2 thousand people, respectively (QingDao Municipal Bureau of Statistics 2021). The carbon emissions per unit of GDP reduced by 45% from 2005 to 2020 due to the low-carbon development policy was implemented from 2006 year. However, the total amount of carbon emissions is increasing and the spatio-temporal heterogeneity is significant with the socio-economic development.
Data source and processing
A time series raster data with resolution of 1km×1km from 2000 to 2020 in Qingdao attained as following: the carbon emissions data of fossil fuels attained from ODIAC fossil fuel emission dataset(https://db.cger.nies.go.jp/dataset/ODIAC/). Annual average temperature data was from National Earth System Science Data Center(http://www.geodata.cn/). Land use and GDP density data were from Resource and Environmental Science and Data Center(https://www.resdc.cn/DOI/doi.aspx?DOIid=33). Population density data was from the World population dataset(https://www.worldpop.org/geodata/listing?id=76). The digital elevation model with resolution of 30m × 30m was from Geospatial data cloud(http://www.gscloud.cn/sources/). Building coverage ratio and floor area ratio were calculated from buildings’ 3D information extracted from high resolution satellite images.
The population density (PD, thousand people/km2), Gross Domestic Product density (GDPD, million yuan/km2), Building coverage ratio (BCR, %), Floor area ratio (FAR), Annual average temperature (T, ℃), Elevation (E, m) and Slope (S, °) are divided into 9 categories by natural breakpoint method, and the land use types (LST) is divided into 8 categories (Table 1). The data of carbon emissions, natural environment and socio-economic factors were extracted by the 10742 grids of 1km×1km, in Qingdao City.
Table 1 Classification of environmental and socioeconomic factors
Types
|
Environmental factors
|
Socioeconomic factors
|
T (℃)
|
S (°)
|
E(m)
|
LST
|
PD (thousand people/km2)
|
GDPD (million yuan/km2)
|
BCR (%)
|
FAR
|
1
|
8.9-10.4
|
0-2.23
|
0-26
|
CL
|
0-0.1
|
1.52-21.03
|
0-2.74
|
0-0.05
|
2
|
10.4-10.9
|
2.23-4.97
|
26-56
|
F
|
0.1-0.3
|
21.03-30.28
|
2.74-8.23
|
0.05-0.16
|
3
|
10.9-11.5
|
4.97-8.20
|
56-94
|
W
|
0.3-0.7
|
30.28-47.10
|
8.23-15.29
|
0.16-0.31
|
4
|
11.5-12.1
|
8.20-11.92
|
94-149
|
RL
|
0.7-1.2
|
47.10-70.29
|
15.29-23.92
|
0.31-0.53
|
5
|
12.1-12.6
|
11.92-16.40
|
149-231
|
UL
|
1.2-1.9
|
70.29-113.77
|
23.92-34.50
|
0.53-0.85
|
6
|
12.6-13.0
|
16.40-21.62
|
231-342
|
GL
|
1.9-3.0
|
113.77-258.17
|
34.50-47.45
|
0.85-1.40
|
7
|
13.0-13.3
|
21.62-27.58
|
342-486
|
UC
|
3.0-4.2
|
258.17-554.58
|
47.45-61.17
|
1.40-2.19
|
8
|
13.3-13.6
|
27.58-35.28
|
486-690
|
R
|
4.2-5.9
|
554.58-1066.55
|
61.17-75.29
|
2.19-3.22
|
9
|
13.6-14.2
|
35.28-63.37
|
690-1088
|
|
5.9-8.3
|
1066.55-9337.56
|
75.29-100
|
3.22-4.82
|
Note: Cultivated land (CL); Forest land (F); Water body (W); Rural construction land (RL); Unused land (UL); Grassland (GL); Urban construction land (UC); Roads (R).
Method
The coefficient of variation, Sen trend analysis and Mann Kendal test based on Python are used to analyze the change stability and variation trend of carbon emissions, in Qingdao City.
The coefficient of variation (Cv) was used for the estimation of the relative variation in carbon emissions for the period 2000–2020, it can be calculated, as shown in Equation (1):
where, σ is the standard deviation of carbon emissions, unit (t); μ is the average value of carbon emissions. When Cv≤0.1 means weak variation, Cv≥1 means strong variation, 0.1<Cv<1 means moderate variation.
The Sen’s slope method was used for the estimation of the change in carbon emissions for the period 2000–2020. It is a non-parametric method which has been found highly reliable for estimation of change over time. The Sen’s slope (β) is calculated as the median of all the slopes estimated between all the successive data points of carbon emissions time series as:
where, CEi and CEj are the carbon emissions in i and j years, 1<i<j<n.β>0 andβ<0 indicated the carbon emissions are on the rise and on the decline, respectively.
The Mann–Kendall test is the most widely used test, which is recommended by the World Meteorological Organization (WMO) often used as because it does consider the data distribution, and it can cope with the outliers. The significance of the change in carbon emissions was estimated using the MK test. The standardized test static for the Mann–Kendall test (Z) can be calculated, as shown in Equation (3):
where, S is the test statistic for carbon emissions can be calculated using Equation in literature (Ali et al. 2019).
The sign of Z indicates the direction of the trend. The negative value of Z indicates a decreasing trend and vice versa. If the absolute value of Z is higher than 1.64 means it passed the significance test with 90% confidence, and it is significant at 10% significance level; If the absolute value of Z is higher than 1.96 means it passed the significance test with 95% confidence, and it is significant at 5% significance level. And when the significance level (P): 0.05<P<0.1 means significant increase or decrease, P<0.05 means extremely significant increase or decrease, others are no significant.
Geodetector is a set of statistical methods to detect spatial differentiation and reveal its driving force (Wang et al. 2017) Factor detector, interaction detector and ecological detector are used to analyze the main factors and interactions of carbon emissions in Qingdao. Factor detector used to explore the influence of different factors on the spatial differentiation of carbon emissions. The factor detector q-statistic can be calculated, as shown in Equation (4):
where h is the number of independent variable layers, h∈[1, L]; L is the number of variable or factor layers; N is the number of samples in the study area; is the variance of dependent variable.
The value of q is strictly within [0, 1], q=1 indicates that the spatial distribution of carbon emissions is completely determined by the factor (X); q=0 indicates that there is no relationship between the factors and the change of carbon emissions. The larger value of q is the greater influence or stronger explanatory power on carbon emissions and vice versa.
The interaction detector reveals whether the factors X1 and X2 (and more X) have an interactive influence on a response variable Y. That is, to evaluate whether the interaction of factors will weak or enhance the explanatory power of carbon emissions, or these factors are independent on the carbon emissions. The interaction relationship is defined in a coordinate axis, and it is determined by the q-statistic in the 5 intervals (Table 2).
Table 2 Interaction between explanatory variables (Xs)
The ecological detector identifies the difference of the impacts between two explanatory variables X1 and X2 on the spatial distribution of carbon emissions. The ecological detector is measured by the F-statistic, and can be calculated in Equation (5):
where NX1 and NX2 are the sample size, SSWX1 and SSWX2 are the sum of variance in the strata, L1 and L2 are the number of strata, of variables X1 and X2, respectively.
At the significance level of 0.05, If Y(X1) was significantly bigger than Y(X2), the associated value is “Y”, means the statistically significant differences between two factors on the distribution of carbon emissions, while “N” expresses the opposite meaning.