The Impact of Future Land Use Change on Carbon Emission and Its Optimization Strategy

Abstract

Rapidly changing climate issues and increasingly severe carbon emissions are great challenges to the carbon peaking and carbon neutrality strategy. Analyzing the impact of future land use changes on carbon emissions can provide an important basis and reference for scientifically constructing a low-carbon and sustainable territorial spatial planning, as well as realizing the goal of the dual-carbon strategy. Based on land use data, agricultural production activity data, and energy consumption statistics, this study simulated the land use changes of the Yangtze River Delta region (YRDR) from 2030 to 2060 under the natural development (ND) scenario and sustainable development (SD) scenario by using the Patch-generating Land Use Simulation (PLUS) model and analyzed the impacts of future land use changes on carbon emissions. The results showed that: (1) The land use simulation results obtained by using the PLUS model under the sustainable development scenario were highly consistent with the actual land use with an OA value of 97.0%, a Kappa coefficient of 0.952, and a FoM coefficient of 0.403; (2) Based on the simulated land use under the SD scenario from 2030 to 2060, the quantity of construction land was effectively controlled, and the spatial distributions of cropland and forests were found to dominate in the north and south of the Yangtze River, respectively; (3) Anhui Province was the major contributor (accounted for 49.5%) to the net carbon absorption by cropland while Zhejiang Province was the major contributor (accounted for 63.3%) to the net carbon absorption by forest in the YRDR during the period 2020–2060 under the SD scenario; (4) Carbon emissions from construction land were the main source of carbon emissions from land use in the YRDR during the period 2020–2060 with proportions higher than 99% under both the ND and SD development scenarios. These findings underscore the urgent need for the government to take measures to balance the relationships between cropland and ecological protection and economic development, which provides a reference for the optimization of land use structure and policy formulation in the future.

1. Introduction

The 2023 report of the United Nations Intergovernmental Panel on Climate Change (IPCC) showed that the global surface temperature was 1.1 °C higher in 2011–2020 compared with 1850–1900 [1,2,3]. Climate change is a global problem faced by all human beings, and it is urgent to reduce greenhouse gas emissions and curb global warming. China committed in 2020 to achieve carbon peaking by 2030 and carbon neutrality by 2060. Previous studies showed that land use carbon emissions had become one of the major sources of greenhouse gas emissions, which directly affected the regional carbon cycle system and changed the level of regional carbon emissions [2,4,5]. Therefore, the study of future land use change and its impact on carbon emission is of great value for improving land use efficiency, rationally adjusting the land use structure, and realizing the goal of the dual-carbon strategy.

Previous studies on carbon emissions from land use mainly focused on the spatial and temporal characteristics of carbon emissions [6,7,8,9], the effects of carbon emissions [10,11,12,13], and the predictions of carbon emissions in multi-scenarios [14,15,16,17]. Multi-scenario land-use carbon emission prediction is mainly used to simulate land use patterns under different development scenarios in the future, which combines the land use simulation results with direct or indirect carbon emission methods to realize the estimation of carbon emissions from different land use types. Among them, the commonly used models for future land use simulations include the cellular automaton (CA) model, GeoSOS model, CLUE-S model, FLUS model, etc. [18,19,20,21]. Transformation rule mining strategy is one of the core components of future land use simulation models, which extracts the transition rules for each land use type to predict the future evolution trend of land use based on existing land use data. However, the most widely used CA model has limitations in terms of transfer rule mining strategy which can not simulate the patch growth of each natural land use type at high spatial resolution and has deficiencies in actual urban planning and policy formulation [22,23]. To improve the simulation accuracy, Liang Xun et al. [23] proposed a patch-generating land use simulation (PLUS) model integrating a land expansion analysis strategy and a CA model based on multiple types of stochastic seeds, which can simulate the generation and evolution of stochastic patch seeds regardless of spatial and temporal constraints, and excavate the mechanism of land use change in the simulation process. Since the PLUS model was proposed in 2021, it has been widely used in land use simulation studies, but mostly at the provincial and municipal levels [24,25]. There are few reports on future land use simulation studies at the regional or national scale and related land use carbon emissions studies.

In terms of the land use carbon emission assessment method, it mainly has two ways, direct and indirect, the former refers to the carbon emission caused by land use while the latter refers to the carbon emission caused by human activities. Since the construction land carries a large number of human activities, it belongs to the indirect carbon emission and plays the role of carbon source, while the forests, grassland, water, and barren land belong to the direct carbon emission and play the role of carbon sink. It is worth noting that the land use type of cropland is very special because it plays the role of carbon sink as well as carbon source in land use carbon emissions [26,27].

Given the above background, this study aims to conduct land use simulations in the YRDR during the period 2020–2060 under the ND and SD scenarios by using the PLUS model, and then to estimate the carbon emissions from land use changes and reveal the impacts of future land use changes on the carbon emissions. The results could provide a reference for the optimization of land use structure and policy formulation for the dual-carbon strategy in the future.

2. Materials and Methods

2.1. Data Sources

Three epochs of 30 m spatial resolution land use data in raster format for 2010, 2015, and 2020 used in this study were obtained from the Google Earth Engine remote sensing cloud platform, produced by the Institute of Geography and Resources of the Chinese Academy of Sciences (IGRS, Beijing, China) using human–computer interaction interpretation method based on Landsat imagery. The data adopt a two-level classification system with the first level classified into six types: cropland, forest, grassland, water, construction land (refers to impervious surfaces throughout the study area which covers both urban and rural areas), and barren land. The overall accuracy of each epoch of land use data was higher than 90% [28]. Restricted by the single-computer processing capacity, this study resampled the three-epoch land-use data to 1 km spatial resolution land-use simulation and prediction.

According to previous studies [29,30], six natural environmental factors and six socio-economic factors were selected as driving factors. To simulate the future land use changes, four limiting factors, namely, water-restricted area, terrain-restricted area, nature reserve, and cropland-restricted area, were selected, and a multi-scenario land use simulation was established through the superposition of the limiting factors.

In addition, to evaluate the carbon emission (absorption) of each land use type, relevant energy consumption data, agricultural production activity data, etc., were also collected. A detailed description of the data is given in

Table 1. Data sources and descriptions.

Data Type	Data Subcategory	Data Description	Data Sources
Land use	Land use data	Three epochs in 2010, 2015, and 2020 with 30 m resolution	https://earthengine.google.com/ (accessed on 15 March 2024)
Natural environmental factors	Distance to water	Distance to water bodies such as rivers, lakes, reservoirs, etc.	Taken from 2020 land use data
	DEM	1 km resolution raster data	https://www.resdc.cn/ (accessed on 15 March 2024)
	Slope	Derived from DEM
	Soil type	1 km resolution raster data
	Average annual temperature	Average temperature in 2015
	Average annual precipitation	Average precipitation in 2015
Socio-economic factors	Population	Spatialized expression of population density in 2015
	GDP	Spatialized expression of GDP value in 2015
	Distance to railway	Distance to railway	OpenStreetMap
	Distance to highway	Distance to highway
	Distance to primary roads	Distance to primary roads in 2021
	Distance to secondary roads	Distance to secondary roads in 2021
limiting factors	Water-restricted area	First-class rivers and watersheds with areas > 10 km2
	Terrain-restricted area	Elevation > 950 m
	Nature reserve	Vector data in 2021
	Cropland-restricted area	Reference to the relevant provisions on the delineation of permanent basic farmland
Statistical data	Energy consumption data	Provincial consumption of major energy sources	China National Bureau of Statistics
	Agricultural production activity data	Economic production of major crops and other relevant data

.

2.2. Methods

The general framework proposed in this study consists of three main components, including the land use simulation process, carbon emission calculation and projection, and the analysis of the response relationship between land use change and carbon emission (Figure 1). First, land use simulations for the period 2020–2060 were carried out using the PLUS model, in which the land use demand under the SD scenario was determined by the related SD scenario settings; second, carbon emissions from individual land use types were obtained based on the simulation results using the direct or indirect carbon emission methods; finally, the impacts of future land use changes on carbon emissions were analyzed.

2.2.1. PLUS Model

Compared with the widely used CA models and other models, the PLUS model can simulate the generation and evolution of random patch seeds regardless of spatial and temporal constraints, and excavate the land use change mechanism during the simulation process [23]. The PLUS model integrates a rule mining framework based on the land expansion analysis strategy (LEAS) model and a CA model based on multi-type random patch seeds (CARS) [31]. The LEAS module utilizes the random forest algorithm to calculate the impact of each selected driving factor on the expansion of each land use type and then estimates the development probability of each land use type [32]. The CARS module couples the CA model with a random seed generation and lowering threshold mechanism to make the quantity of land use meet the future demand under the constraints of adaptive coefficients, development probabilities, neighborhood effects, etc.

2.2.2. Scenario Settings

Previous studies usually used natural development scenarios, ecological protection scenarios, and economic development scenarios for multi-scenario simulations [33,34], however, the balanced development among economic development, cropland protection, and ecological protection was rarely reported. This study developed the SD scenario for achieving balanced development among economic development, cropland protection, and ecological protection by controlling the restriction area and the transfer matrix. The restriction area for the SD scenario includes the water restriction area, terrain restriction area, nature reserve, and cropland restriction area. For the transfer matrix, the SD scenario restricts the conversion of cropland, forest, and water to other land types, allows the conversion of grassland to forest and water, and prohibits the conversion of construction land to other land types.

2.2.3. Model Validation and Accuracy Evaluation

The land use data in 2010 and 2015 and the PLUS model were used to simulate the land use in 2020, and the simulation results were compared with the actual land use in 2020 for model validation and accuracy evaluation. The model parameters were determined by the trial-and-error method with about two hundred model iterations and were used to predict future land use for each decade during the period 2030–2060 based on the two periods of actual land use data in 2010 and 2020.

Five indexes including overall accuracy (OA), producer accuracy (PA), user accuracy (UA), FoM, and Kappa coefficient were utilized to evaluate the accuracy of the land use simulation results. The formulas of OA, Kappa, PA, and UA are shown in Equations (1)–(4), respectively:

$$OA={\displaystyle \frac{1}{N}}·{\sum }_{i=1}^k{x}_{ii}\times 100\%$$

(1)

$$Kappa={\displaystyle \frac{N{\sum }_{i=1}^k{x}_{ii}-{\sum }_{i=1}^k{x}_{ii}(x_{i+}·x_{+i})}{N^2-{\sum }_{i=1}^k{x}_{ii}(x_{i+}·x_{+i})}}$$

(2)

$${PA}_i={\displaystyle \frac{x_{ii}}{x_{+i}}}$$

(3)

$${UA}_i={\displaystyle \frac{x_{ii}}{x_{i+}}}$$

(4)

2.2.4. Calculation of Land Use Carbon Absorption

Among the six land use types, cropland is special which is both a carbon source and a carbon sink. According to previous studies [35,36], carbon absorption for cropland was calculated by Equation (5):

$${CI}_{crop}={\sum }_i{CI}_{crop-i}={\sum }_i{C}_{crop-i}\times (1-P_{water-i})\times {\displaystyle \frac{Y_{eco-i}}{H_{crop-i}}}$$

(5)

where CI_crop is the photosynthetic carbon absorption during crop growth period; CI_crop-i is the carbon absorption of the ith crop; C_crop-i is the carbon absorption rate of the ith crop in synthesizing units of organic matter (dry weight) through photosynthesis; P_water-i is the water content of the ith crop; Y_eco-i is the economic yield of the ith crop; and H_crop-i is the economic coefficient of the ith crop.

In this study, carbon absorption for cropland was mainly calculated for rice, wheat, corn, beans, potato, cotton, sorghum, tobacco, and oilseed. The related coefficients for individual crops are shown in

Table 2. The related coefficients for individual crops to calculate carbon absorption for cropland.

Crop Type	Crop Carbon Uptake Rate	Water Content Rate	Crop Economic Coefficient
Rice	0.45	0.12	0.40
Wheat	0.48	0.12	0.40
Corn	0.47	0.13	0.40
Beans	0.45	0.13	0.34
Potatoes	0.42	0.70	0.70
Cotton	0.45	0.08	0.10
Sorghum	0.45	0.12	0.35
Tobacco	0.45	0.85	0.55
Oil seed	0.45	0.10	0.30

.

Different from cropland, the land use types of forest, grassland, water, and barren land are the main carbon sinks. The carbon absorption coefficient method was utilized to calculate the carbon absorption of these four land use types [36]. The calculation formula is shown in Equation (6):

$$E_k=\sum e_i=\sum T_i\times {\delta }_i$$

(6)

where E_k is the total carbon absorption; e_i is the carbon absorption of the ith land-use type; T_i is the area of the ith land-use type; and δ_i is the carbon absorption coefficient of the ith land-use type. The carbon absorption coefficients of these four land use types are shown in

Table 3. The carbon absorption coefficients of individual land use types.

Land Use Type	Carbon Absorption Factor (t/hm2)	References
Forest	0.644	[37]
Grassland	0.021	[9,27]
Water	0.253	[38,39]
Barren Land	0.005	[38,39]

.

2.2.5. Calculation of Land Use Carbon Emissions

Carbon emissions from cropland usually comprise four types of sources, including organic carbon losses due to tillage, fossil fuels consumed by agricultural machinery, the use of fertilizers, pesticides, and agricultural films, and electricity consumption for agricultural irrigation. The formula used to calculate carbon emissions from cropland is shown in Equation (7):

$$E=\sum E_i=\sum G_i\times {\epsilon }_i$$

(7)

where E is the total carbon emissions from cropland, E_i is the carbon emissions from the ith source; G_i is the consumption of the ith source; and ε_i is the coefficient of the ith source.

The coefficients of individual sources of carbon emissions from cropland were adopted according to previous studies [40,41], which are shown in

Table 4. The coefficients of major sources of carbon emissions from cropland.

	Tillage	Agricultural Machinery	Fertilizer	Pesticide	Agricultural Film	Irrigation
Coefficient	312.6	0.18	0.8956	4.9341	5.18	25
Unit	kg·km−2	kg·kw−1	kg·kg−1	kg·kg−1	kg·kg−1	kg·hm−2

.

Construction land carries a lot of human activities and consumes a lot of energy, which is another important source of carbon emissions. The indirect carbon emission estimation method was used to calculate carbon emissions from construction land. Nine types of energy, such as raw coal and coke, were selected for the calculation. The calculation formula is shown in Equation (8):

$$E_p=\sum E_j=\sum e_j\times {\theta }_j\times {\beta }_j$$

(8)

where E_p is the total carbon emissions from construction land; E_j is the carbon emissions from each energy source of the jth type; e_j is the carbon emission coefficient of the jth type; θ_j is the standard coal coefficient of the jth energy source; and β_j is the consumption of the jth energy source.

The coefficients of individual energy sources were taken from the 2019 Refinement to the 2006 IPCC Guidelines for National Greenhouse Gas Inventories [41] and previous studies [26,27], which are shown in

Table 5. Coefficients of individual energy sources for calculating carbon emissions from construction land.

Types of Energy	Carbon Emission Coefficient	Standard Coal Coefficient
Coal	0.7559	0.7143 tce/t
Crude oil	0.5857	1.4286 tce/t
Coke	0.8550	0.9714 tce/t
Fuel oil	0.6185	1.4286 tce/t
Gasoline	0.5538	1.4714 tce/t
Kerosene	0.5714	1.4714 tce/t
Diesel oil	0.5921	1.4571 tce/t
Natural gas	0.4483	1.3300 tce/103 m3
Electricity	0.7935	0.4040 tce/103 kw·h

.

2.2.6. Land Use Carbon Absorption/Emission Prediction

The carbon emission coefficient X_i per unit area of each land use type was calculated by using the carbon absorption/emission and the area of each land use type in 2020. The formula is shown in Equation (9):

$$X_i={\displaystyle \frac{C_i}{A_i}}$$

(9)

where X_i is the carbon emission coefficient per unit area of each land use type; C_i is the emission absorption/emission of each land use type in 2020; A_i is the area of each land use type in 2020.

Finally, carbon absorption/emissions were projected for each land use for the period 2030–2060 using the carbon emission coefficients X_i per unit area of each land use type and the Markov projections of the area of each land use type for each epoch.

3. Results

3.1. Accuracy Assessment

The land use patterns in 2020 under the ND and SD scenarios were simulated using the PLUS model based on the actual land use data in 2010 and 2015. The simulation results under the two scenarios were compared with the actual land use data in 2020 to assess model performance, which is shown in

Table 6. Accuracy assessments of land use simulations under the ND and SD scenarios in 2020.

Scenario	Simulation Accuracy for All Land Use Types			Simulation Accuracy for Cropland		Simulation Accuracy for Forest		Simulation Accuracy for Construction Land
	FoM	Kappa	OA (%)	PA (%)	UA (%)	PA (%)	UA (%)	PA (%)	UA (%)
SD	0.403	0.952	97.0	97.5	97.1	98.6	97.5	92.2	99.7
ND	0.360	0.932	95.8	96.4	96.8	97.2	97.4	91.6	90.3

and Figure 2.

The results showed that under both the ND and SD scenarios, high model performance was obtained with OA values greater than 95%, the Kappa coefficients greater than 0.93, and the FoM coefficients greater than 0.36. Compared with the ND scenario, higher model performance was obtained for the SD scenario with the OA improved by 1.2%, the Kappa coefficient improved by 0.020, and the FoM coefficients improved by 0.037. For the accuracy assessment of individual land use types, the producer accuracy and user accuracy of cropland and forest were both higher than 97%, while the PA and UA of construction land were higher than 92%.

The results indicate that the simulation outputs under the SD scenario are more consistent with the actual land use, and the PLUS model has stable and accurate simulation capability for individual land use types, which is important for simulating future land use changes.

3.2. Spatiotemporal Dynamics of Future Land Use during 2030–2060

3.2.1. Changes in the Quantity of Future Land Use during 2030–2060

Statistics on the area of each land use type during 2030–2060 under the two development scenarios are shown in

Table 7. The simulated area statistics of individual land use types during 2030–2060 under the two development scenarios (Unit: km²).

	2030		2040		2050		2060
	ND	SD	ND	SD	ND	SD	ND	SD
Cropland	178,604	182,515	175,814	186,280	172,955	188,713	172,116	184,755
Forest	101,905	105,256	99,014	101,913	96,827	99,812	95,555	110,051
Grassland	39	10	32	7	28	7	64	177
Water	19,344	17,961	17,840	17,643	17,280	17,017	16,460	20,764
Barren land	1	0	1	1	1	1	2	2
Construction land	48,505	42,656	55,697	42,554	61,307	42,848	64,024	32,472

. The results showed that the most significant differences between the two scenarios were mainly reflected in the variation trends of construction land and cropland. Under the ND scenario, the area of cropland showed a decreasing trend while the area of construction land showed an increasing trend with an average annual growth rate of 0.69% during 2030–2060. In contrast, under the SD scenario, the area of cropland showed an increasing trend while the area of construction land showed a decreasing trend with an average annual growth rate of −0.68% during 2030–2060. This indicates that under the SD scenario, the protection of cropland could be strengthened, the expansion of construction land could be restricted with its spatial distribution more concentrated, and the land utilization efficiency could be greatly improved.

The transfer trajectories of various land use types under the two development scenarios during 2020–2060 are shown in Figure 3 and Figure 4. Under the ND scenario, cropland and construction land were mainly transferred to each other, which was mainly subject to constraint mechanisms such as linking new construction land with illegally occupied cropland. Under the SD scenario, cropland and forest were mainly transferred to each other, which was subject to the regulation of the “in-and-out balance” cropland protection system with the area changes of the two land use types being relatively smooth. In addition, the total outflow area of construction land during the period 2050–2060 was predicted to be 10,376 km², which was mainly changed to cropland with an outflow area of 8922 km² (accounting for 86.8%). This is probably due to China’s increasingly intensive land use policies and the measures encouraging the reclamation of construction land into cropland, especially in rural areas. The effectiveness of these policies and measures is expected to be significantly improved in the long-term process of urbanization and the anticipation of continued population decline in the future. However, the interconversions of grassland, water, and barren land rarely occurred under both scenarios.

3.2.2. Changes in the Spatial Patterns of Future Land Use during 2030–2060

The model parameters in the PLUS model, which were determined in the simulations for land use in 2020, were used to predict the land use patterns during the period 2030–2060 under the two scenarios with the actual land use data in 2010 and 2020 (Figure 5(a1–a4,c1–c4)). The spatial distributions of the expansions of individual land use types in each epoch during the period 2020–2060 were shown in Figure 5(b1–b4,d1–d4).

Compared with the ND scenario, the expansion of construction land was effectively controlled under the SD scenario with the spatial distributions of construction land gradually stabilized in Shanghai, southern Jiangsu, and eastern Zhejiang during 2030–2060. This indicates that the cropland protection-related policy constraints played a certain role and the utilization efficiency of construction land greatly improved under the SD scenario. The spatial distributions of cropland and forest in the YRDR were mainly bounded by the Yangtze River with major cropland distributed in the north and forest mainly distributed in the south of the Yangtze River. The reason for such a phenomenon is related to the special distribution characteristics of watershed and topography in the YRDR, which is dominated by paddy fields with large watersheds mainly located in the north YRDR such as Taihu Lake, Hongze Lake, Gaoyou Lake in Jiangsu Province and Chaohu Lake in Anhui Province. In addition, the topography of the southern YRDR is hilly and mountainous, which has a lot of constraints on the distribution and expansion of cropland.

3.3. Analysis of Future Land Use Carbon Absorption/Emissions

Carbon emission/absorption of each land use under the two development scenarios during the period 2020–2060 are shown in Figure 6. The simulated cropland area under the SD scenario was larger than that under the ND scenario for each epoch, leading to both higher carbon emission and higher carbon absorption of cropland under the SD scenario. The total carbon absorption of cropland was larger than the total carbon emission of cropland under both development scenarios, resulting in the larger net carbon absorption of cropland under the SD scenario compared with that under the ND scenario. According to the National Plan for Sustainable Agricultural Development (2015–2030), the YRDR belongs to the optimized development area, of which Anhui and Jiangsu provinces are important agricultural provinces in China. The average net carbon absorption of cropland in Anhui and Jiangsu provinces accounted for 50.2% and 43.2% of the total net carbon absorption of cropland in the YRDR during 2020–2060 under the ND scenario, while the average total net carbon absorption of cropland accounted for 49.5% and 43.5% of the total net carbon absorption of cropland in the YRDR during 2020–2060 under the SD scenario, respectively. The average net carbon absorption of forests in Zhejiang and Anhui provinces accounted for 63.3% and 35.1% of the total net carbon absorption of forests in the YRDR during 2020–2060 under the SD scenario, respectively.

Under the two Scenarios, carbon emissions from construction land in the YRDR showed two typical trends. Specifically, the carbon emissions from construction land in the YRDR under the ND scenario showed an upward trend while the carbon emissions from construction land in the YRDR under the SD scenario showed a downward trend. This suggests that strengthening the government’s regulation and control role, strengthening cropland and ecological protection, controlling the continuous expansion of construction land, and improving the efficiency of construction land use could play a great role in controlling carbon emissions from land use and realizing the dual-carbon goal in the future.

The net carbon emissions from land use in the YRDR under the two development scenarios during the period 2020–2060 are shown in

Table 8. Net carbon emissions from land use under two development scenarios in the YRDR from 2030 to 2060 (Unit: Mt).

	ND Scenario					SD Scenario
	Shanghai	Jiangsu	Zhejiang	Anhui	YRDR	Shanghai	Jiangsu	Zhejiang	Anhui	YRDR
2020 *	125.39	411.94	272.63	152.24	962.20	125.39	411.94	272.63	152.24	962.20
2030	99.46	442.93	343.76	173.44	1059.59	114.92	407.18	253.35	152.92	928.40
2040	129.20	504.53	444.41	192.88	1271.02	114.82	405.66	251.46	152.94	924.89
2050	137.70	548.54	520.42	210.96	1417.62	115.02	410.56	253.96	152.70	932.25
2060	142.51	542.06	622.15	206.17	1512.88	99.46	292.93	179.81	112.04	684.23

* Net carbon emissions from land use in 2020 were calculated by using the actual land use data in 2020.

. The net carbon emissions in the YRDR were all positive values, indicating that the YRDR was a carbon source area. The net carbon emissions from land use in the YRDR under the ND scenario showed an upward trend with the net carbon emissions increasing from 962.20 million tons in 2020 to 1512.88 million tons in 2060, while, the net carbon emissions from land use in the YRDR under the SD scenario showed a downward trend with the net carbon emissions decreased from 962.20 million tons in 2020 to 684.23 million tons in 2060. Among the six land use types (Figure 6), carbon emissions from construction land were the main source of land use in the YRDR, and the average carbon emissions from construction land during 2020–2060 under the two development scenarios were higher than 99% of the total carbon emissions of all land use in the YRDR.

4. Discussion

4.1. Policy Implications

To alleviate conflicts over land use and food security and ecological environment protection, China has promulgated a series of policy documents since 1994, including the Regulations on the Protection of Basic Cropland, the Outline of the National Overall Land Utilization Plan, and the Fourteenth Five-Year Plan for the Development of Building Energy Efficiency and Green Buildings. These policies point out that rational utilization of land and effective protection of cropland is China’s basic state policy. It needs to correctly deal with the relationships between economic development and cropland protection, actively promote the transformation of land utilization from rough to intensive, and improve the land utilization rate to ensure the dynamic balance of the total amount of cropland and the rational utilization of land.

Although economic development is certainly important, food security and ecological protection can not be ignored. The results indicate that under the SD scenario, the protection of cropland could be strengthened and the expansion of construction land could be restricted with its spatial distribution more concentrated, and the utilization efficiency could be greatly improved, which can be attributed to the policy regulation of the balanced development among economic development, cropland protection, and ecological protection. This finding highlights the important role of sustainable development scenario modeling for future land use, cropland protection, and improvement in construction land utilization efficiency.

To further balance the relationships between cropland and ecological protection and economic development for sustainable green development, this study puts forward three recommendations for government decision-making. First, it is recommended to strengthen green science and technology innovation and to promote the application of advanced green technology, including the use of advanced agricultural technology and management measures to improve agricultural production efficiency, the intensification of manufacturing production, the optimization of industrial structure for green and low-carbon development, the development of green and low-carbon industries and supply chain, and the design of a green, low-carbon, and recycling economic system. Second, relevant departments should strengthen high-frequency and accurate monitoring of carbon emissions, establish an effective carbon emission monitoring system, and regularly assess the carbon emissions from land use changes. Third, the results show that within the YRDR Economic Circle, the carbon absorption of cropland in Anhui and Jiangsu provinces contributes greatly to reducing the total net carbon emissions in the YRDR. Therefore, it is suggested that the central and provincial governments should scientifically arrange the main functions of different cities according to different natural and humanistic conditions to achieve the minimization of carbon emissions from a regional integration perspective.

4.2. Advantages, Limitations, and Perspectives

Taking the YRDR as the study area, this study analyzes the impact of future land use changes on carbon emissions by simulating the land use patterns under different development scenarios, which provides a scientific basis for constructing a low-carbon land use system. The strengths of this study lie in the following two aspects: (1) The SD scenario is developed for achieving the balanced development among economic development, cropland protection, and ecological protection by controlling the restriction area and the transfer matrix. According to the accuracy of the simulation results, it is found that the PLUS model has a more refined and accurate land use simulation capability under the SD scenario compared with the ND scenario; (2) The impacts of future land use changes on carbon emission/absorption are determined by analyzing the carbon emission/absorption of individual land use types under the two development scenarios during the period 2020–2060, and the related policy implications are proposed for controlling carbon emissions from land use and realizing the dual-carbon goal in the future.

Although this study has made some progress as mentioned above, there are several shortcomings: (1) The carbon emission coefficients of several land use types (e.g., forest and grassland) used in this study were derived from the study of Fang Jingyun et al. on the estimation of carbon sinks of terrestrial vegetation in China during 1981–2000 [26]. However, both the research period and the spatial distributions of individual land use types of the study of Fang Jingyun et al. could significantly change after two decades of development. Therefore, it needs to revise the carbon emission coefficients regionally and epochally. (2) Although the results of this study show that the proposed SD scenario is effective in simulating future land use patterns, there is no objective standard for the setting of the SD scenario. The Sustainable Development Goals Report 2020 mentions 17 sustainable development goals which contain many aspects of food security, economic growth, sustainable urban development, environmentally friendly development, climate change, terrestrial ecology, and marine ecology [42]. Subsequent studies could develop different scenarios by balancing the interactions of different SDGs and setting different thresholds according to the differences between the current actual stage of the regional or national development levels and the SDG goals. (3) The quantity of future land use demand is one of the important parameters in scenario simulations. In the PLUS model, the prediction of future land use demand is accomplished by Markov chain prediction or linear prediction, but both prediction methods may have the problem of subjectivity. Some previous studies [21,43] used the System Dynamics model (SDM) to predict the future land use demand under different scenarios from subsystems such as economy, climate, population, and land use, which provides a reference method for optimizing the prediction of future land use demand. In addition, the model parameters for simulating future land use during the period 2030–2060 were derived from the simulation model using the actual land use data in 2010 and 2015, future research work should consider the variability of the model parameters over time. (4) The level of scientific and technological development increases rapidly. Although the significant impact of energy consumption (e.g., fuel oil and gasoline) on carbon emissions is clear in the calculations of carbon emissions from construction land in this study, how the growing advancement and the adaptation of science and technology can impact the results is ignored, especially the electrical vehicular technology. Therefore, the great influence of scientific and technological development should be considered in future research work.

5. Conclusions

This study adopted the PLUS model to simulate the land use patterns under two scenarios in the YRDR from 2020 to 2060 and analyzed the impact of future land use changes on carbon emissions. The results showed that the PLUS model can accurately simulate future land use patterns. The land use patterns during the period 2020–2060 under the SD scenario showed that the expansion of construction land was effectively controlled while forest and cropland were found to dominate in the south and north of the Yangtze River, respectively. From the statistical analysis results of future carbon emissions, Anhui Province and Zhejiang Province were the major contributors to the net carbon absorption in the YRDR during the period 2020–2060 under the SD scenario, which accounted for 49.5% of the net carbon absorption by cropland and 63.3% of the net carbon absorption by forest, respectively. Carbon emissions from construction land were the main source of carbon emissions from land use in the YRDR during the period 2020–2060, accounting for greater than 99%.

The findings underscore the important role of the sustainable development scenario in simulating future land use patterns which aims to achieve a balanced development among economic development, cropland protection, and ecological protection. Additionally, this study highlights that the total net carbon emissions in the YRDR could be reduced by optimizing land use through the government’s regulation measures. Specifically, strengthening cropland and ecological protection, controlling the continuous expansion of construction land, and improving the use efficiency of construction land could play an important role in controlling carbon emissions from land use. Moreover, this study emphasizes the important demand for improving management practices to achieve the dual-carbon goal, such as strengthening green science and technology innovation, improving high-frequency and accurate monitoring of carbon emissions, and scientifically arranging the main functions of different cities to achieve the minimization of carbon emissions from a regional integration perspective. This research will become increasingly important considering the urgent demand and the severe pressure on achieving the dual-carbon goal.