*3.1. Spatial and Temporal Patterns of GTCL*

### 3.1.1. Dynamic Evolution Characteristics of Regional Differences

As shown in Figures 3 and 4, the overall GTCL index in all China's provinces, municipalities and autonomous regions shows a "W"-shaped fluctuating uptrend, with the average values of the comprehensive transition index in 2000, 2005, 2010, 2015 and 2020 being 0.202, 0.137, 0.206, 0.147 and 0.237, respectively. The provinces, municipalities and autonomous regions above the average in 2000 were mainly Heilongjiang, Inner Mongolia, Anhui, Henan and Ningxia; in 2005, they were concentrated in Hebei, Jiangsu, Anhui, Shandong, Henan and Chongqing; in 2010, they were mainly in Inner Mongolia, Jilin, Heilongjiang, Anhui, Henan and Ningxia; in 2015, they were concentrated in Heilongjiang, Anhui, Jiangxi, Shandong and Henan; in 2020, they were concentrated in Inner Mongolia, Jilin, Heilongjiang, Jiangsu, Shandong and Henan. From 2000 to 2005, Inner Mongolia had the largest transition rate at −48.23%, Anhui the smallest at −19.29%; from 2005 to 2010, Heilongjiang had the largest transition rate at 169.07% and Beijing the smallest at 17.72%; from 2010 to 2015, Ningxia had the largest transition rate at −46.00% and Shanghai the smallest at 4.44%; from 2015–2020, Inner Mongolia had the largest transition rate at −51.54%, and Shanghai the smallest at −4.23%. In addition, the GTCL in provincial areas is more balanced. The extreme difference of GTCL index is 0.157 in 2000, 0.095 in 2005, 0.301 in 2010, 0.158 in 2015, and 0.673 in 2020, which indicates that the difference of GTCL between provincial areas shows an overall increasing trend.

**Figure 3.** Evolution of spatial pattern of GTCL in China during 2000–2020.

**Figure 4.** Change rate of GTCL in China during 2000–2020.

3.1.2. Global Characteristics of the Evolution of Spatial Pattern

Using the trend analysis tools in ArcGIS 10.2 software, this paper makes a threedimensional intervisibility analysis on the overall trend of the GTCL in 31 provinces, municipalities and autonomous regions of China from 2000 to 2020 as research units. Taking the GTCL as the Z axis and the X and Y axes as the due east and due north directions, respectively, the spatial visualization results are obtained (Figure 5). The results show that there are significant spatial differences in the distribution of GTCL in China from 2000 to 2020. The overall distribution of cultivated land is basically the same from east to west, high in the north and low in the south.

**Figure 5.** Trend analysis of GTCL in China during 2000–2020.

In GeoDa, a global spatial autocorrelation analysis was conducted for GTCL using Rook's criterion to calculate the Global Moran's I index (Figure 6).

**Figure 6.** Moran's *I* index value of GTCL.

The global autocorrelation coefficients of Moran's I were all positive: 0.262, 0.188, 0.395, 0.205, and 0.396 from 2000 to 2020, respectively, and were divided into three categories according to the relative magnitude of each year: strong (absolute value ≥ 0.5), relatively weak (0.3 ≤ absolute value < 0.5), and weak (0 ≤ absolute value < 0.3). Overall, GTCL in each province, autonomous region and municipality showed a significant positive correlation between 2000 and 2020, and there were obvious regional clustering characteristics in space. From 2000 to 2005, the Moran's I value decreased, indicating a weak clustering distribution pattern among provinces, municipalities and autonomous regions. Compared with 2005, the Moran's I value increased significantly in 2010, which showed a weak clustering distribution pattern among provinces, municipalities and autonomous regions. Compared with 2015, Moran's I value in 2020 showed a significant increase, but it was still larger than that in 2010, indicating that the correlation of spatial distribution of GTCL among provinces, municipalities and autonomous regions increased in the period of 2015–2020, but the overall spatial differences showed an increasing trend.

### *3.2. Spatial and Temporal Patterns of "Water, Land, Food and Carbon" Changes*

The spatial and temporal patterns of "water, land, food, and carbon" changes in each province, autonomous region and municipality were analyzed based on the evaluation indexes W (w), L (l), F (f), and C (c) (Figures 7 and 8).

**Figure 7.** Temporal and spatial distribution of "Water-Land -Food-Carbon".

The water system W (w) evaluation index ranges from 0.0048 to 0.210, with a large fluctuation trend, mainly influenced by the virtual water self-sufficiency rate, effective irrigation area, water consumption per unit area of grain yield and water consumption per unit of cultivated land (Figure 8a). From 2000 to 2005, the highest rates of change in water were in Tibet, Ningxia and Hainan, with −66.82%, −66.37% and −62.34%, respectively, and the lowest was in Anhui, with a reduction of 4.44%, probably related to the reduced contribution of local water resources to the population's food consumption. From 2005 to 2010, the highest rate of water change was in Northwest China, with the largest increases in Qinghai and Ningxia, followed by Heilongjiang and Jilin in Northeast China, and the lowest was in Beijing, with an increase of 16.17%, probably related to the increase in the effective irrigated area. From 2010 to 2015, the trend was decreasing, with the highest rate of change in Qinghai and Ningxia, at 80.54% and 63.05%, followed by Jilin, at −56.63%, and Beijing, at −7.33%. From 2015 to 2020, the largest increases were in Qinghai, Heilongjiang

and Inner Mongolia, at 318.96%, 183.63% and 132.31%, with the smallest decrease in Beijing, at 1.29%, which is related to the increase of water consumption per unit of cultivated land.

**Figure 8.** Evolution of spatial pattern of "Water–Land–Food–Carbon".

The L (l) evaluation index of the cultivated land system ranges from 0.027 to 0.278, showing a "W"-shaped variation trend, but with small fluctuations (Figure 8b). Moreover, 2000–2005 is the period of decreasing fluctuation, with an average annual change rate of 7.08%, which is related to the decrease of per capita cultivated land area; 2005–2010 is the period of increasing fluctuation, which is related to the increase of multiple crop index, among which the more typical ones are Inner Mongolia, Xinjiang and Heilongjiang, where the multiple crop index increased from 0.758, 0.936 and 0.857 to 0.980, 1.154 and 1.028; the fluctuating-decreasing phase from 2010 to 2015 is related to the increase of land reclamation rate. For example, the land reclamation rate in Shandong, Henan, Jiangsu and Anhui increased from 0.478, 0.475, 0.464 and 0.410 to 0.482, 0.485, 0.427, and 0.420, respectively; in the fluctuation-increasing phase from 2015 to 2020, Inner Mongolia, Xinjiang, Heilongjiang, and Gansu had the largest change rates, with average annual growth rates of 62.86%, 53.52%, 38.44%, and 35.37%, respectively, while Beijing, Shanghai, and Fujian had the smallest change rates. This is related to the per capita cultivated land area and the intensity of non-point source pollution.

The evaluation index of food system F (f) ranged from 0.0042 to 0.282, with a fluctuation growth trend and a more stable fluctuation trend (Figure 8c). The period of 2000–2005 was a decreasing phase, which was closely related to the instability of grain sown area and grain yield; the period of 2005–2010 was a growing phase, especially in Xinjiang, Heilongjiang, Inner Mongolia and Qinghai with annual average growth rates of 99.63%, 72.79%, 68.57% and 63.53%, respectively, mainly influenced by the per capita grain yield, the proportion of sown area of grain crops and the ratio of food crops to cash crops. From 2010–2015 was a fluctuation-decreasing phase, with a relatively stable decrease, averaging 10% per year; 2015–2020 was a fluctuation-increasing trend, mainly influenced by the average grain yield.

The evaluation index of carbon system C (c) ranged from 0.0098 to 0.063, with less fluctuation than the values of the water, cultivated land and food systems, and the increasing trend of the evaluation index was not obvious, basically in a slow growth state, with a mean value of 0.045, which was probably related to the slow increase of carbon emissions from pesticides, fertilizers, agricultural films, tillage, total power of agricultural machinery and irrigation during the use of cultivated land (Figure 8d).
