Volume 32 Issue 3
Jun.  2022
Turn off MathJax
Article Contents

GUO Rong, WU Tong, WU Xiaochen, LUIGI Stendardo, WANG Yueqin, 2022. Simulation of Urban Land Expansion Under Ecological Constraints in Harbin-Changchun Urban Agglomeration, China. Chinese Geographical Science, 32(3): 438−455 doi:  10.1007/s11769-022-1277-1
Citation: GUO Rong, WU Tong, WU Xiaochen, LUIGI Stendardo, WANG Yueqin, 2022. Simulation of Urban Land Expansion Under Ecological Constraints in Harbin-Changchun Urban Agglomeration, China. Chinese Geographical Science, 32(3): 438−455 doi:  10.1007/s11769-022-1277-1

Simulation of Urban Land Expansion Under Ecological Constraints in Harbin-Changchun Urban Agglomeration, China

doi: 10.1007/s11769-022-1277-1
Funds:  Under the auspices of National Key R&D Program of China (No. 2018YFC0704705)
More Information
  • Corresponding author: WU Tong. E-mail: wutong_hit@163.com
  • Received Date: 2021-08-10
  • Accepted Date: 2021-10-29
  • Available Online: 2022-05-26
  • Publish Date: 2022-05-05
  • Under the demand of urban expansion and the constraints of China’s ‘National Main Functional Area Planning’ policy, urban agglomerations are facing with a huge contradiction between land utilization and ecological protection, especially for Harbin-Changchun urban agglomeration who owns a large number of land used for the protection of agricultural production and ecological function. To alleviate this contradiction and provide insight into future land use patterns under different ecological constraints’ scenarios, we introduced the patch-based land use simulation (PLUS) model and simulated urban expansion of the Harbin-Changchun urban agglomeration. After verifying the accuracy of the simulation result in 2018, we predicted future urban expansion under the constraints of three different ecological scenarios in 2026. The morphological spatial pattern analysis (MSPA) method and minimum cumulative resistance (MCR) model were also introduced to identify different levels of ecological security pattern (ESP) as ecological constraints. The predicted result of the optimal protection (OP) scenario showed less proportion of water and forest than those of natural expansion (NE) and basic protection (BP) scenarios in 2026. The conclusions are that the PLUS model can improve the simulation accuracy at urban agglomeration scale compared with other cellular automata (CA) models, and the future urban expansion under OP scenario has the least threat to the ecosystem, while the expansion under the natural expansion (NE) scenario poses the greatest threat to the ecosystem. Combined with the MSPA and MCR methods, PLUS model can also be used in other spatial simulations of urban agglomerations under ecological constraints.
  • 加载中
  • [1] Beatriz A, Daunt P, Sanna T et al., 2021. Urban expansion and forest reserves: drivers of change and persistence on the coast of São Paulo State (Brazil). Land Use Policy, 101: 105189. doi:  10.1016/j.landusepol.2020.105189
    [2] Bennett M M, Smith L C, 2017. Advances in using multitemporal night-time lights satellite imagery to detect, estimate, and monitor socioeconomic dynamics. Remote Sensing of Environment, 192: 176–197. doi:  10.1016/j.rse.2017.01.005
    [3] Cengiz S, Görmüş S, Oğuz D, 2022. Analysis of the urban growth pattern through spatial metrics; Ankara City. Land Use Policy, 112: 105812. doi:  10.1016/j.landusepol.2021.105812
    [4] Chen B, 2015. Integrated ecological modelling for sustainable urban metabolism and management. Ecological Modelling, 318: 1–4. doi:  10.1016/j.ecolmodel.2015.10.009
    [5] Chen Y M, Li X, Liu X P et al., 2014. Modeling urban land-use dynamics in a fast developing city using the modified logistic cellular automaton with a patch-based simulation strategy. International Journal of Geographical Information Science, 28(2): 234–255. doi:  10.1080/13658816.2013.831868
    [6] Chen Y M, Li X, Liu X P et al., 2019. Simulating urban growth boundaries using a patch-based cellular automaton with economic and ecological constraints. International Journal of Geographical Information Science, 33(1): 55–80. doi:  10.1080/13658816.2018.1514119
    [7] Chettry V, Surawar M, 2021. Delineating urban growth boundary using remote sensing, ANN-MLP and CA model: a case study of Thiruvananthapuram urban agglomeration, India. Journal of the Indian Society of Remote Sensing, 49(10): 2437–2450. doi:  10.1007/s12524-021-01401-x
    [8] Deng X Z, Huang J K, Rozelle S et al., 2008. Growth, population and industrialization, and urban land expansion of China. Journal of Urban Economics, 63(1): 96–115. doi:  10.1016/j.jue.2006.12.006
    [9] Fan F L, Wang Y P, Qiu M H et al., 2009. Evaluating the temporal and spatial urban expansion patterns of Guangzhou from 1979 to 2003 by remote sensing and GIS methods. International Journal of Geographical Information Science, 23(11): 1371–1388. doi:  10.1080/13658810802443432
    [10] Fang Chuanglin, Song Jitao, Zhang Qiang et al., 2005. The formation, development and spatial heterogeneity patterns for the structures system of urban agglomerations in China. Acta Geographica Sinica, 60(5): 827–840. (in Chinese)
    [11] Geshkov M V, DeSalvo J S, 2012. The effect of land-use controls on the spatial size of U. S. urbanized areas. Journal of Regional Science, 52(4): 648–675. doi:  10.1111/j.1467-9787.2012.00763.x
    [12] Gong J Z, Liu Y S, Xia B C et al., 2009. Urban ecological security assessment and forecasting, based on a cellular automata model: a case study of Guangzhou, China. Ecological Modelling, 220(24): 3612–3620. doi:  10.1016/j.ecolmodel.2009.10.018
    [13] Gu Chaolin, 2011. Study on urban agglomeration: progress and prospects. Geographical Research, 30(5): 771–784. (in Chinese)
    [14] Guo R, Bai Y J, 2019. Simulation of an urban-rural spatial structure on the basis of green infrastructure assessment: the case of Harbin, China. Land, 8(12): 196. doi:  10.3390/land8120196
    [15] Guo R, Wu T, Liu M R et al., 2019. The construction and optimization of ecological security pattern in the Harbin-Changchun urban agglomeration, China. International Journal of Environmental Research and Public Health, 16(7): 1190. doi:  10.3390/ijerph16071190
    [16] He Jianhua, Shi Xuan, Gong Jian et al., 2016. Modeling the spatial expansion of urban agglomeration considering their spatial interaction: a case study of Wuhan Metropolitan Area. Geomatics and Information Science of Wuhan University, 41(4): 462–467. (in Chinese)
    [17] He J L, Li X, Yao Y et al., 2018. Mining transition rules of cellular automata for simulating urban expansion by using the deep learning techniques. International Journal of Geographical Information Science, 32(10): 2076–2097. doi:  10.1080/13658816.2018.1480783
    [18] Huang D Q, Huang J, Liu T, 2019. Delimiting urban growth boundaries using the CLUE-S model with village administrative boundaries. Land Use Policy, 82: 422–435. doi:  10.1016/j.landusepol.2018.12.028
    [19] Huang Jinchuan, Lin Haoxi, 2017. Spatial evolution analysis and multi-scenarios simulation of Beijing-Tianjin-Hebei Urban Agglomeration. Geographical Research, 36(3): 506–517. (in Chinese)
    [20] Inkoom J N, Nyarko B K, Antwi K B, 2017. Explicit modeling of spatial growth patterns in Shama, Ghana: an agent-based approach. Journal of Geovisualization and Spatial Analysis, 1(1−2): 7. doi:  10.1007/s41651-017-0006-2
    [21] Jawarneh R N, 2021. Modeling past, present, and future urban growth impacts on primary agricultural land in Greater Irbid Municipality, Jordan using SLEUTH (1972–2050). ISPRS International Journal of Geo-Information, 10(4): 212. doi:  10.3390/ijgi10040212
    [22] Li F, Ye Y P, Song B W et al., 2015. Evaluation of urban suitable ecological land based on the minimum cumulative resistance model: a case study from Changzhou, China. Ecological Modelling, 318: 194–203. doi:  10.1016/j.ecolmodel.2014.09.002
    [23] Li Hui, Yi Na, Yao Wenjing et al., 2011. Shangri-La county ecological land use planning based on landscape security pattern. Acta Ecologica Sinica, 31(20): 5928–5936. (in Chinese)
    [24] Li S C, Bing Z L, Jin G, 2019. Spatially explicit mapping of soil conservation service in Monetary Units due to land use/cover change for the three Gorges reservoir area, China. Remote Sensing, 11(4): 468. doi:  10.3390/rs11040468
    [25] Li X, Yeh A G O, 2001. Calibration of cellular automata by using neural networks for the simulation of complex urban systems. Environment and Planning A:Economy and Space, 33(8): 1445–1462. doi:  10.1068/a33210
    [26] Li X M, Zhou W Q, 2018. Dasymetric mapping of urban population in China based on radiance corrected DMSP-OLS nighttime light and land cover data. Science of the Total Environment, 643: 1248–1256. doi:  10.1016/j.scitotenv.2018.06.244
    [27] Liang X, Liu X P, Li X et al., 2018. Delineating multi-scenario urban growth boundaries with a CA-based FLUS model and morphological method. Landscape and Urban Planning, 177: 47–63. doi:  10.1016/j.landurbplan.2018.04.016
    [28] Liang X, Guan Q F, Clarke K C et al., 2021. Understanding the drivers of sustainable land expansion using a patch-generating land use simulation (PLUS) model: a case study in Wuhan, China. Computers, Environment and Urban Systems, 85: 101569. doi:  10.1016/j.compenvurbsys.2020.101569
    [29] Lin J, Huang B, Chen M et al., 2014. Modeling urban vertical growth using cellular automata—Guangzhou as a case study. Applied Geography, 53: 172–186. doi:  10.1016/j.apgeog.2014.06.007
    [30] Liu Cuiling, Long Ying, 2015. Urban expansion simulation and analysis in the Beijing-Tianjin-Hebei Region. Progress in Geography, 34(2): 217–228. (in Chinese)
    [31] Liu J M, Xiao B, Li Y S et al., 2021. Simulation of dynamic urban expansion under ecological constraints using a long short term memory network model and cellular Automata. Remote Sensing, 13(8): 1499. doi:  10.3390/rs13081499
    [32] Liu X P, Liang X, Li X et al., 2017. A future land use simulation model (FLUS) for simulating multiple land use scenarios by coupling human and natural effects. Landscape and Urban Planning, 168: 94–116. doi:  10.1016/j.landurbplan.2017.09.019
    [33] Liu Xiaoyang, Zeng Jian, Jia Mengyuan et al., 2020. Construction of ecological security pattern and simulation of urban sprawl in the urban agglomeration of Min Delta. Acta Ecologica Sinica, 40(21): 7873–7885. (in Chinese)
    [34] Lv J J, Wang Y F, Liang X et al., 2021. Simulating urban expansion by incorporating an integrated gravitational field model into a demand-driven random forest-cellular automata model. Cities, 109: 103044. doi:  10.1016/j.cities.2020.103044
    [35] Ouyang Xiao, Zhu Xiang, 2020. Spatio-temporal characteristics of urban land expansion in Chinese urban agglomerations. Acta Geographica Sinica, 75(3): 571–588. (in Chinese)
    [36] Pontius Jr R G, Boersma W, Castella J C et al., 2008. Comparing the input, output, and validation maps for several models of land change. The Annals of Regional Science, 42(1): 11–37. doi:  10.1007/s00168-007-0138-2
    [37] Qiu Yao, Chang Qing, Wang Jing, 2013. A MSPA-based planning of urban green infrastructure network—a case of Shenzhen. Chinese Landscape Architecture, 29(5): 104–108. (in Chinese)
    [38] Santé I, García A M, Miranda D et al., 2010. Cellular automata models for the simulation of real-world urban processes: a review and analysis. Landscape and Urban Planning, 96(2): 108–122. doi:  10.1016/j.landurbplan.2010.03.001
    [39] Saura S, Torné J, 2009. Conefor sensinode 2.2: a software package for quantifying the importance of habitat patches for landscape connectivity. Environmental Modelling & Software, 24(1): 135–139. doi:  10.1016/j.envsoft.2008.05.005
    [40] Seto K C, Fragkias M, Güneralp B et al., 2011. A meta-analysis of global urban land expansion. PLoS One, 6(8): e23777. doi:  10.1371/journal.pone.0023777
    [41] Soille P, Vogt P, 2009. Morphological segmentation of binary patterns. Pattern Recognition Letters, 30(4): 456–459. doi:  10.1016/j.patrec.2008.10.015
    [42] Su Y X, Chen X Z, Liao J S et al., 2016. Modeling the optimal ecological security pattern for guiding the urban constructed land expansions. Urban Forestry & Urban Greening, 19: 35–46. doi:  10.1016/j.ufug.2016.06.013
    [43] Tang Y, Yuan Y B, Zhong Q Y, 2021. Evaluation of land comprehensive carrying capacity and spatio-temporal analysis of the Harbin-Changchun urban agglomeration. International Journal of Environmental Research and Public Health, 18(2): 521. doi:  10.3390/ijerph18020521
    [44] Verburg P H, Overmars K P, 2009. Combining top-down and bottom-up dynamics in land use modeling: exploring the future of abandoned farmlands in Europe with the Dyna-CLUE model. Landscape Ecology, 24(9): 1167–1181. doi:  10.1007/s10980-009-9355-7
    [45] Wang Jing, Fang Chuanglin, 2011. Growth of urban construction land: progress and prospect. Progress in Geography, 30(11): 1440–1448. (in Chinese)
    [46] Wu Xinxin, Liu Xiaoping, Liang Xun et al., 2018. Multi-scenarios simulation of urban growth boundaries in Pearl River delta based on FLUS-UGB. Journal of Geo-Information Science, 20(4): 532–542. (in Chinese)
    [47] Xia C, Zhang A Q, Wang H J et al., 2019. Bidirectional urban flows in rapidly urbanizing metropolitan areas and their macro and micro impacts on urban growth: a case study of the Yangtze River middle reaches megalopolis, China. Land Use Policy, 82: 158–168. doi:  10.1016/j.landusepol.2018.12.007
    [48] Xu L, Huang Q H, Ding D D et al., 2018. Modelling urban expansion guided by land ecological suitability: a case study of Changzhou City, China. Habitat International, 75: 12–24. doi:  10.1016/j.habitatint.2018.04.002
    [49] Yang Tianrong, Kuang Wenhui, Liu Weidong et al., 2017. Optimizing the layout of eco-spatial structure in Guanzhong urban agglomeration based on the ecological security pattern. Geographical Research, 36(3): 441–452. (in Chinese)
    [50] Yao Y, Liu X P, Li X et al., 2017. Simulating urban land-use changes at a large scale by integrating dynamic land parcel subdivision and vector-based cellular automata. International Journal of Geographical Information Science, 31(12): 2452–2479. doi:  10.1080/13658816.2017.1360494
    [51] Ye H, Yang Z P, Xu X L, 2020. Ecological corridors analysis based on MSPA and MCR model—a case study of the Tomur World Natural Heritage Region. Sustainability, 12(3): 959. doi:  10.3390/su12030959
    [52] Yu Kongjian, Wang Sisi, Li Dihua, 2012. Regional Ecological Security Patterns. Beijing: China Architecture & Building Press, 56–65. (in Chinese)
    [53] Yuan Y J, Wu S H, Yu Y N et al., 2018. Spatiotemporal interaction between ecosystem services and urbanization: case study of Nanjing City, China. Ecological Indicators, 95: 917–929. doi:  10.1016/j.ecolind.2018.07.056
    [54] Zhang D C, Liu X P, Wu X Y et al., 2019. Multiple intra-urban land use simulations and driving factors analysis: a case study in Huicheng, China. Giscience & Remote Sensing, 56(2): 282–308. doi:  10.1080/15481603.2018.1507074
    [55] Zhang L Q, Peng J, Liu Y X et al., 2017. Coupling ecosystem services supply and human ecological demand to identify landscape ecological security pattern: a case study in Beijing-Tianjin-Hebei region, China. Urban Ecosystems, 20(3): 701–714. doi:  10.1007/s11252-016-0629-y
    [56] Zhang W T, Li B, 2021. Research on an analytical framework for urban spatial structural and functional optimisation: a case study of Beijing city, China. Land, 10(1): 86. doi:  10.3390/land10010086
    [57] Zhou Kan, Wu Jianxiong, Fan Jie et al., 2022. Drivers of regional environmental pollution load and zoning control: a case study of the Yangtze River economic Belt, China. Chinese Geographical Science, 32(1): 31–48. doi:  10.1007/s11769-022-1257-5
    [58] Zhu G Y, Tang Z S, Shangguan Z P et al., 2019. Factors affecting the spatial and temporal variations in soil erodibility of China. Journal of Geophysical Research, 124(3): 737–749. doi:  10.1029/2018JF00491
  • 加载中
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Figures(10)  / Tables(7)

Article Metrics

Article views(581) PDF downloads(57) Cited by()

Proportional views
Related

Simulation of Urban Land Expansion Under Ecological Constraints in Harbin-Changchun Urban Agglomeration, China

doi: 10.1007/s11769-022-1277-1
Funds:  Under the auspices of National Key R&D Program of China (No. 2018YFC0704705)

Abstract: Under the demand of urban expansion and the constraints of China’s ‘National Main Functional Area Planning’ policy, urban agglomerations are facing with a huge contradiction between land utilization and ecological protection, especially for Harbin-Changchun urban agglomeration who owns a large number of land used for the protection of agricultural production and ecological function. To alleviate this contradiction and provide insight into future land use patterns under different ecological constraints’ scenarios, we introduced the patch-based land use simulation (PLUS) model and simulated urban expansion of the Harbin-Changchun urban agglomeration. After verifying the accuracy of the simulation result in 2018, we predicted future urban expansion under the constraints of three different ecological scenarios in 2026. The morphological spatial pattern analysis (MSPA) method and minimum cumulative resistance (MCR) model were also introduced to identify different levels of ecological security pattern (ESP) as ecological constraints. The predicted result of the optimal protection (OP) scenario showed less proportion of water and forest than those of natural expansion (NE) and basic protection (BP) scenarios in 2026. The conclusions are that the PLUS model can improve the simulation accuracy at urban agglomeration scale compared with other cellular automata (CA) models, and the future urban expansion under OP scenario has the least threat to the ecosystem, while the expansion under the natural expansion (NE) scenario poses the greatest threat to the ecosystem. Combined with the MSPA and MCR methods, PLUS model can also be used in other spatial simulations of urban agglomerations under ecological constraints.

GUO Rong, WU Tong, WU Xiaochen, LUIGI Stendardo, WANG Yueqin, 2022. Simulation of Urban Land Expansion Under Ecological Constraints in Harbin-Changchun Urban Agglomeration, China. Chinese Geographical Science, 32(3): 438−455 doi:  10.1007/s11769-022-1277-1
Citation: GUO Rong, WU Tong, WU Xiaochen, LUIGI Stendardo, WANG Yueqin, 2022. Simulation of Urban Land Expansion Under Ecological Constraints in Harbin-Changchun Urban Agglomeration, China. Chinese Geographical Science, 32(3): 438−455 doi:  10.1007/s11769-022-1277-1
    • Along with the population growth and intensive social-economic activities, China has gone through rapid urbanization since its reform and opening up in 1978, leading to an accelerated flowing of productive factors and decreasing distance among cities (Wang and Fang, 2011). In this process, urban agglomerations have gradually become the most economically active and densely populated areas of China, promoting the rapid expansion of urban land (Gu, 2011). Due to the disordered distribution structures of urban land caused by excessive expansion, ecological land with high value is faced with the risk of being converted into urban land, which would exacerbate the contradictions between fast urbanization and ecological protection in urban agglomerations (Fang et al., 2005; Zhou et al., 2022). Inevitably, the continuous expansion of urban construction land without constraints will aggravate a series of ecological problems such as biodiversity loss, soil erosion, and water pollution, which would directly reduce the value of ecosystem services and ultimately damage the health of the ecosystem (Yuan et al., 2018; Li et al., 2019). Therefore, the Chinese government has issued the ecological civilization construction policy and emphasized the importance of promoting ecological protection by delineating three lines, including permanent basic farmland protection red line, ecological protection line, and urban development boundary (Chen et al., 2019; Liu et al., 2021). However, existing research has mainly focused on the simulation of urban expansion of city individuals, and there are fewer studies based on variable scenarios at the scale of urban agglomerations, which restricts decision-makers from being able to understand fully the underlying drivers of urban expansion and determine the optimal land use structures under different scenarios (Zhang et al., 2019; Zhang and Li, 2021; Cengiz et al., 2022). This study aims at maintaining and avoiding the overexploitation and destruction of the ecological resources and spaces during the process of urban expansion in urban agglomerations. Thus, to provide significant guidance for regional integration and sustainable development in China (Liu et al., 2021).

      As a result of human beings’ interactions with the natural environment, urban land expansion has gradually become a hot topic for regional development research. Many scholars have noticed the importance of spatiotemporal characteristics analysis of urban land expansion, using quantitative indexes including expansion intensity, expansion rate, expansion direction, and so on (Seto et al., 2011; Ou and Zhu, 2020; Beatriz et al., 2021). In recent years, the extensive usage of remote sensing (RS), geographic information systems (GIS), and global positioning systems (GPS) has facilitated the availability of obtaining urban land use/coverage data, thereby promoting studies on the dynamic simulation of urban expansion and helping urban planners understand the evolution of land use changes (Chen, 2015; Bennett and Smith, 2017). Cellular automata (CA) modelling is widely adopted by scholars to simulate land use patterns for city individuals due to its powerful functions in terms of spatial integration and parallel computing (Santé et al., 2010; Lin et al., 2014). Adopting a bottom-up calculation method, with the transformation rules of neighboring cells, and boundary constraints as basic elements, CA modelling has a strong advantage in terms of simplicity and flexibility of complex land use simulations (Inkoom et al., 2017; He et al., 2018). To explore the potential driving factors of land use change and improve the accuracy of the simulation results, scholars have upgraded the traditional CA models by combining them with deep learning techniques and intelligent algorithms such as agent-based models (ABM), artificial neural networks (ANNs), and recurrent neural networks (RNNs) (Li and Yeh, 2001; Chettry and Surawar, 2021). Scholars have also made an effort to upgrade traditional CA models to simulate urban land expansion and predict land use patterns, such as slope, landuse, exclusion, urban extent, transportation and hillshade (SLEUTH) model, conversion of land use and its effects (CLUE-s) model, and future land use simulation (FLUS) model, which are based on land use map data to mine distribution rules through competition mechanisms (Liang et al., 2018; Huang et al., 2019; Jawarneh, 2021). However, these improved CA models generally focus on urban construction land rather than natural land types, and are inadequate for simulating the evolution of patch-level changes of multiple land use types (Chen et al., 2014). To resolve the above issues, a patch-based land use simulation (PLUS) model was proposed to simulate spatiotemporal urban expansion by exploiting its advantages in revealing the underlying factors and their contributions to multiple land use changes (Liang et al., 2021). Combined with a series of scenarios characterized by different sets of constraints, the simulation results based on the PLUS model can better support planning policies to achieve sustainable development.

      The model parameters obtained from the CA-based models are only one aspact of the factors affecting the simulation accuracy, and can not play a decisive role in the simulation accuracy alone. As the basis of influencing the intensity of land expansion and leading to land use change, the rationality and representativeness of driving factors are equally important. At present, different scholars have excavated the traditional driving factors of land use change from the urban scale, and believe that land use change is the result of the combined action of many influencing factors such as nature, society, economy and politics (Ouyang and Zhu, 2020; Zhang and Li, 2021). However, a unified analysis framework of driving factors of urban land expansion has not been established. With the rapid increase in the status of urban agglomerations in regional development and China’s national strategy, many studies have elaborated on the factors and mechanisms of large-scale urban expansion. A few studies noted that spillover effect factor usually expressed by inter-city interaction (e.g., the flow of people, materials and information) is playing an important role in urban expansion of urban agglomerations, which can guide other cities to expand toward the core cities (He et al., 2016; Huang and Lin, 2017). In details, with the continuous strengthening of spatial interaction among cities, the production factors keep gathering in the core cities, and the spillover effect will make the functions of the core cities radiate to the surrounding areas. Then with the increasingly clear functions and positioning of surrounding urban areas and counties, the benefits of industrial division of labor and rational gradient distribution of industries will be increasingly obvious, thus promoting the expansion direction of urban land to move closer to the core city. Spillover effect factor is a unique influencing factor of urban land expansion in urban agglomeration. It is thus necessary to consider spatial interactions among cities when considering factors of urban expansion. To address this problem, the gravitational field model (GFM) has been employed by many researchers to measure the strength of spatial interactions among cities (Lv et al., 2021).

      In recent years, with the advancement of ecological construction goals, there has been an increasing number of studies on predictions of land use patterns under different ecological constraint (EC) scenarios (Geshkov and Desalvo, 2012; Xu et al., 2018; Guo and Bai, 2019). Normally, different partitions of ecological security patterns (ESP) determined by ecological sources identification and resistance surface construction are usually regarded as the basis of EC scenario design (Gong et al., 2009; Su et al., 2016). Many studies identified ecological sources by constructing a comprehensive evaluation index system, and using indicators such as ecological sensitivity, ecological fragility, and the importance of ecological services (Zhang et al., 2017). However, due to the subjective selection of the vectorized raster data, important patches that are scattered in shape are usually ignored. In recent years, a morphological spatial pattern analysis (MSPA) method, which emphasizes structural connections relying on land use data, has been widely used to identify of ecological sources (Ye et al., 2020). Therefore, the integrity and accuracy of the identification are improved. In addition, the minimum cumulative resistance surface in the ecological security pattern can be obtained from the MCR model, and different ESP levels can be used as the basis for the design of different EC scenarios (Wu et al., 2018; Liu et al., 2020).

      From the research discussed above, it can be seen that the application of the PLUS model and the construction of ES scenarios are mostly based on city individuals. However, there are few studies on urban agglomerations considering interaction among cities. As one of the developing urban agglomerations in China, although the Harbin-Changchun urban agglomeration (HCUA) is facing problems such as industrial recession and population contraction, its land urbanization is still in a rapid process (Tang et al., 2021). Thus, the HCUA is experiencing rapid urbanization and urban land expansion. In the future, with the promotion of relevant policies and plans of the Chinese government to take urban agglomeration as the main form of high-quality urbanization development. Coupled with the promotion of globalization and informatization, the rapid development of economy will further promote HCUA’s urban expansion and population aggression, and continuously promote the urbanization process with the circular cumulative effect. This process may result in a fragile ecological environment and decline service functions, and threaten the sustainable development of the area. Therefore, this study took HCUA as a case study and provided insights to: 1) analysis the spatiotemporal change of urban expansion in HCUA from 2005 to 2018; 2) verify the feasibility of the PLUS model for simulating urban expansion of urban agglomerations; 3) combine the MSPA method and MCR model for identification of ESP to design the EC scenarios; and 4) predict dynamic urban expansion to 2026 under natural expansion (NE) and EC scenarios. In short, this study can broaden the application field of the PLUS model, and provide a profound scientific analysis basis for planners to make future land use strategies under different ecological constraints’ scenarios.

    • HCUA is located in Northeast China, adjacent to eastern Russia, bordering eastern Mongolia, radiating the Manzhouli port, and connected to Beijing-Tianjin-Hebei urban agglomeration and Bohai Rim Economic Zone (Guo et al., 2019). As a transportation hub, HCUA plays an important role in promoting communication between Northeast Asia, Central Asia, and Europe, and it is also a key to the opening up of Northeast China to the rest of the world. With Harbin and Changchun as the central cities, HCUA covers 11 prefecture-level regions including Harbin, Daqing, Suihua, Qiqihar, Mudanjiang, Changchun, Jilin, Siping, Songyuan, Liaoyuan, and Yanbian (autonomous prefecture) and 67 county-level cities under its jurisdiction, with an area of 322 323 km2 (Fig. 1).

      Figure 1.  Location of the Harbin-Changchun urban agglomeration, China. DEM, digital elevation model

    • The spatial data we used in this study include: land use data (2005, 2010, 2015, 2018), basic administrative boundary data, normalized difference vegetation index (NDVI) data, DMSP-OLS nighttime light (NTL) data and digital elevation model (DEM) data. They are derived from different data centers, which are: 1) the Resource and Environment Science Data Center of the Chinese Academy of Sciences (http://www.resdc.cn/); 2) the National Basic GIS database of China (http://ngcc.sbsm.gov.cn/); 3) the National Aeronautics and Space Administration (NASA) (https://modis.gsfc.nasa.gov/); 4) the National Geophysical Data Center of China (http://www.ngdc.noaa.gov/); 5) the geospatial data cloud (http://www.gscloud.cn/). Besides, railway and highway data were derived from the website (http://www.openstreetmap.org/), and all statistical related data, including GDP, population, urbanization etc., were derived from or calculated based on the yearbooks of national, provincial and local cities. In order to unify the scale of the grid, we reclassified these data to 250 m × 250 m. As urban expansion is affected by socio-economic factors, natural environmental factors, and location factors, 13 spatial variables were selected as the driving factors of the urban expansion simulation, including the DEM, slope, NDVI, distance to core cities (Harbin and Changchun), distance to railway, distance to highway, distance to water, GDP, growth rate of GDP, population, level of urbanization, proportion of industrial output value in total value and the spatial interaction capacity of each city (Ouyang and Zhu, 2020; Zhang and Li, 2021). It is worth noting that we adopted the gravitational field model to calculate the spatial interaction capacity, which was obtained through calculating the square root of multiplying the population by GDP of every two cities, and dividing this square root by the traffic distance (Lv et al., 2021). Fig. 2 shows the driving factor dataset of spatial simulation in the HCUA.

      Figure 2.  Driving factors dataset of spatial simulation in the Harbin-Changchun urban agglomeration, China: a. DEM (digital elevation model); b. slope; c. NDVI (normalized difference vegetation index); d. distance to core cities; e. distance to railway; f. distance to highway; g. distance to water; h. GDP; i, Growth rate of GDP; j. population; k. level of urbanization; l. proportion of industrial output value; m. the spatial interaction capacity of each city

    • As shown in Fig. 3, the methodology framework of this study included three main parts: 1) the transition rules mining for expansion of each land use type based on the land expansion analysis strategy (LEAS) module; 2) simulation and prediction by cellular automata model based on multiple random seeds (CARS) module; and 3) the setting of the EC scenarios as geographic constraints based on MSPA and MCR methods.

      Figure 3.  The methodology framework of the study. T, the time interval; RF, the random forest algorithm; PLUS, the patch-based land use simulation (PLUS) model; ANN-CA, the cellular automata with artificial neural network model; RNN-CA, the cellular automata with recurrent neural network model; MSPA, morphological spatial pattern analysis method; MCR, minimum cumulative resistance method

    • Combined the LEAS module and the CARS module, the PLUS model provides new visions for exploring the underlying factors of land use change within a specific time interval and simulating the simultaneous evolution of multiple types of land patches. In order to verify the accuracy of the simulation results of the PLUS model, we also used the cellular automata model combined with artificial neural network (ANN-CA), and the cellular automata model combined with recurrent neural network (RNN-CA) to simulate the spatial expansion, thus the above three simulation results were compared. The application method of ANN-CA and RNN-CA refer to the existing research (Li and Yeh, 2001; Chettry and Surawar, 2021)

    • A land expansion analysis strategy (LEAS) was adopted to mine the transition rules, which was based on the random forest (RF) algorithm. The RF algorithm contained a large number of decision trees, which were constructed based on sub-datasets drawn from the original training dataset using random sampling methods (Yao et al., 2017). Based on these decision trees, the growth probability was obtained, and the formula is:

      $$ P_{ij,k}^d = \dfrac{{\displaystyle\sum\limits_{n = 1}^M {I\left( {{h_n}\left( x \right) = d} \right)} }}{M} $$ (1)

      where $P_{ij,k}^d$ is the growing probability of cell (i, j) being classified under land use type k; I(∙) is the indicative function of the decision tree set; x is a high dimensional vector that consists of multiple driving factors; hn(x) is the prediction type of the nth decision tree for x, and M represents the total number of decision trees; d indicates whether the land use type k has changed from another land use type (d = 1) or not (d = 0).

    • In the PLUS model, the CARS module plays a key role in simulating and predicting the patch evolution of multiple land use types (Liu and Long, 2015; Liu et al., 2017). The total probability of each cell can be calculated as follows:

      $$ TP_{ij,k}^{d = 1,t} = P_{ij,k}^{d = 1} \times {{\Omega }}_{ij,k}^t \times D_k^t \times {\text{Con}}\left( {S_{ij}^t} \right) $$ (2)

      where $TP_{ij,k}^{d = 1,t}$ is the total growing probability of cell (i, j); $P_{ij,k}^d$is obtained through the LEAS module; ${{\Omega }}_{ij,k}^t$ denotes the neighborhood weights of cell (i, j), which reflects the interaction between cell (i, j) and its surrounding cells of land use type k; $D_k^t$ is the self-adaptive driving coefficient used to adjust the current amount of land use k at time iteration t; ${{{\rm{Con}}\; (S}}_{ij}^t)$ denotes geographic constraints (0 or 1). Based on the Moore neighborhood mode, the neighborhood effects can be calculated as follows:

      $$ {{\Omega }}_{ij,k}^t = \frac{{{\text{Con}}\left( {S_{ij}^t = k} \right)}}{{3 \times 3 - 1}} \times {w_k} $$ (3)

      where ${{{\rm{Con}}\;(S}}_{ij}^{t - 1} = k)$ represents the state of cell (i, j) classified under land use type k at the last iteration t; and ${w_k}$ is the weighting of the different land use types. The self-adaptive driving coefficient can be calculated as follows:

      $$ D_k^t = \left\{ \begin{gathered} \;\;\;\;\;D_k^{t - 1}\;\;\;\;\;\;\;if\;\left| {G_k^{t - 1}} \right| \le \left| {G_k^{t - 2}} \right| \hfill \\ D_k^{t - 1} \times \frac{{G_k^{t - 2}}}{{G_k^{t - 1}}}\;\;\;if\;G_k^{t - 1} < G_k^{t - 2} < 0 \hfill \\ D_k^{t - 1} \times \frac{{G_k^{t - 1}}}{{G_k^{t - 2}}}\;\;\;if\;0 < G_k^{t - 2} < G_k^{t - 1} \hfill \\ \end{gathered} \right. $$ (4)

      where $G_k^{t - 1}$ and $G_k^{t - 2}$ represent the difference between the current amount and the future demand of land use category k in the iterative operation at t–1 and t–2, respectively. When the neighborhood effect of a certain type of land use k is equal to 0, change seeds are generated using the Monte Carlo approach, and the formula is:

      $$ P_{ij,k}^{d = 1,t} = \left\{ \begin{gathered} P_{ij,k}^{d = 1} \times \left( {\gamma \times {\mu _k}} \right) \times D_k^t \hfill \\ if\Omega _{ij,k}^t = 0\;and\;\gamma < P_{ij,k}^{d = 1} \hfill \\ P_{ij,k}^{d = 1} \times {{\Omega }}_{ij,k}^t \times D_k^t \; \;all\; others \hfill \\ \end{gathered} \right. $$ (5)

      where $P_{ij,k}^{d = 1,t}$ is the growing probability of cell (i, j) without the neighborhood effect; μk is the threshold for generating new land use patches for k; $\gamma $ is used to assess the status of candidate land use which is selected by the roulette wheel.

      $$ I f \sum_{k=1}^{N}\left|G_{c}^{t-1}\right|-\sum_{k=1}^{N}\left|G_{c}^{t}\right|<{Step}\; Then, q=q+1 $$ (6)
      $$ \left\{ \begin{gathered} Change\;P_{ij,c}^{d = 1} > \Gamma and\;T{M_{k,c}} = 1 \hfill \\ No\;change\;P_{ij,c}^{d = 1} \le {{\Gamma }}\;T{M_{k,c}} = 0 \hfill \\ \end{gathered} \right.\;\;\;\;{{\Gamma }} = \delta \times \gamma 1 $$ (7)

      ine-formula$G_c^t$ and $G_c^{t - 1}$ represent the difference between the current amount and the future demand of land use category c in the iterative operation at t and t–1; N is the total number of new land use patches and q is the number of decay steps; $P_{ij,c}^{d = 1}$ is the growing probability of cell (i, j) of candidate land use c which was selected by the roulette wheel; Step is the step size to estimate the land use demand; $\delta $ is the decay factor of decreasing threshold; $\gamma1$ is a normally distributed stochastic value with a mean of 1, ranging from 0 to 2; and l is the number of attenuation steps. $T{M_{k,c}}$ is a conversion matrix that defines whether the land use type k is allowed to be converted to type c (Verburg and Overmars, 2009).

    • In this study, we used the Kappa coefficient and FOM to verify the effectiveness of the PLUS model (Pontius et al., 2008). The Kappa coefficient was calculated by the transfer matrix obtained by spatially superimposing maps of land use types. It is a common method used to compare the consistency of remote sensing images. Generally, when Kappa ≥ 0.75, the consistency is high; when 0.4 ≤ Kappa ≤ 0.75, the consistency is general; when Kappa ≤ 0.4, the consistency is poor. Compared with the Kappa coefficient, FOM has higher accuracy in reflecting the consistency of complex geographic systems. The calculation formula is:

      $$ FOM = \frac{B}{{A + B + C + D}} $$ (8)

      where FOM presents the figure of merit; A is the number of cells that have been converted in the real world, but have not been converted in the simulation; B is the number of cells that have been converted in both the real world and the simulation; C is the number of cells that were converted in both the real world and the simulation, but the number of error cells are not the same in reality; D is the number of error cells that have not been converted in reality but were converted in the simulation.

    • An ecological security pattern indicates important areas to ensure the ecological service functions and species migration. Based on the PLUS model, taking different levels of ESP as ES scenarios can obtain future land use distribution patterns under different development goals. In this study, the identification of sources, correction of resistance surface, and extraction of minimum cumulative resistance surface were important steps for the ESP construction.

    • Morphological spatial pattern analysis (MSPA) is an image processing method for measuring, identifying, and segmenting the space of raster images based on mathematical morphology principles such as corrosion, expansion, opening operation, and closing operation (Soille and Vogt, 2008). It can help better identify important ecological patches in the research area. Then relevant landscape pattern indexes were used to choose the ecological sources. The calculation formulas are:

      $$ IIC = \dfrac{{\displaystyle\sum\limits_{i = 1}^n {\displaystyle\sum\limits_{j = 1}^n {\frac{{{a_i} \times {a_j}}}{{1 + n{s_{ij}}}}} } }}{{A_L^2}} $$ (9)
      $$ PC = \dfrac{{\displaystyle\sum\limits_{i = 1}^n {\displaystyle\sum\limits_{j = 1}^n {p_{ij}^ * \times {a_i} \times {a_j}} } }}{{A_L^2}} $$ (10)

      where IIC is the integral index of connectivity and PC is the probability of connectivity; n represents the total number of patches in the landscape; ai and aj are the contribution values of patches i and j; $n{s_{ij}}$ refers to the number of connections between i and j; AL represents the area of the entire landscape; and $p_{ij}^ *$ is the maximum probability of diffusion between i and j, which is set to 0.5 in this study (Zhu et al., 2019).

    • Traditionally, the value of the basic resistance surface is assigned according to different land use types evaluated by experts. This method has been criticized for relying too much on experts’ personal experiences. The NTL data coefficient of DMSP-OLS were often used to modify this value. Based on this, we assigned basic resistance values to different landscapes, and adopted night light data to correct the foundation resistance surface. The formula is:

      $$ {R_i} = \frac{{TL{I_i}}}{{TL{I_a}}} \times R $$ (11)

      where ${R_i}$ is the modified resistance coefficient of grid i; $TL{I_i}$ is the night light index of grid i; $TL{I_a}$ is the average night light index corresponding to grid i, and R is the basic resistance coefficient of grid i.

    • According to the theory of ESP, the spatial movement of organisms and the maintenance of habitats need to overcome the resistance of the landscape in the horizontal ecological process. The minimum cumulative resistance (MCR) can refer to the cost or work done to overcome the resistance from the ‘source’ through the landscape with different resistances, and reflect the cumulative cost distance from the target patch to the nearest source patch. The MCR surface was extracted by the MCR model, together with the basic resistance surface, form an overall segmented and partially continuous spatial distribution pattern. By identifying the mutation point of the MCR value and the corresponding number of grids, an automatic division of different levels of ESPs was realized (Qiu et al., 2013). The calculation formula is:

      $$ MCR = f\min \sum\limits_{j = n}^{i = m} {{D_{ij}} \times {R_i}} $$ (12)

      where MCR is the minimum cumulative resistance value; f is the positive correlation function, ${D_{ij}}$ shows the spatial distance between grid unit i and source j. ${R_i}$ presents the ecological resistance coefficient of grid unit i to the movement process.

    • From 2005 to 2018, the area of built-up land in the HCUA showed an upward trend and phase characteristics were obvious (Table 1). The expansion intensity indexes of the HCUA in the three time intervals (2005–2010, 2010–2015, and 2015–2018) were 0.22%, 0.08%, and 0.32% respectively, which were calculated by dividing the area of the expanded built-up land by the total area of the region (322 323km2). During 2005–2010, with the support of national control policies, including a series of strategies to revitalize the old industrial base in Northeast China, the planning outline for opening up to northeast Asia, and the construction of international cooperation demonstration zones for port cities, the built-up land of the HCUA experienced a rapid expansion. During 2010–2015, a series of farmland protection policies promulgated by the State Council of China pointed out that the HCUA was located in an important grain production area of China, which restricted the rapid expansion of built-up land. During 2015–2018, the advancement of the coordinated development planning strategies of the HCUA accelerated the flow of productive factors, which promoted the highest expansion intensity of built-up land.

      Land use type20052010201520182005–2018
      Area / km2Proportion / %Area / km2Proportion / %Area / km2Proportion / %Area / km2Proportion / %Changed area / km2Changed proportion / %
      Cultivated15086846.8114977246.4714995146.5215436547.8934971.08
      Forest11541935.8111544735.8211542235.8111416135.42–1258–0.39
      Grass170755.3180255.59178975.55140894.37–2986–0.93
      Water113773.5392982.8892752.8877682.41–3609–1.12
      Unused161385.01176155.47173745.39184915.7423530.73
      Build-up114463.55121663.77124043.85134494.1720030.62

      Table 1.  Area and proportion of different land use types in the Harbin-Changchun urban agglomeration (HCUA). China from 2005 to 2018

      The spatial pattern evolution of built-up land expansion presented an obvious ‘center-periphery’ characteristic (Fig. 4). The ranking of the contribution rate of different cities to the expansion during 2005–2018 was as follows: Changchun (42.83%) > Harbin (20.34%) >Daqing (10.02%) > Jilin (9.10%) > Mudanjiang (5.24%) > Songyuan (4.02%) > Siping (3.12%) > Suihua (3.03) > Qiqihar (0.85) > Liaoyuan (0.83) > Yanji (0.62%). More than 60% of the expanded built-up land was distributed in Harbin and Changchun with the municipal district as the centers. Due to the high concentration of economic, population and other productive factors to the two cities, which promoted the urban spatial expansion, these two cities were in the center of urban land expansion pattern of HCUA. However, affected by the division and obstruction of administrative boundaries, insufficient ability of regional division of labor and cooperation, weak support of transportation network and policies, these two core cities had not formed a continuous integrated area to drive the development of other cities in HCUA. Qiqihar, Suihua, Liaoyuan and Yanbian have become vulnerable areas under the influence of urban internal shrinkage and weak external radiation, which led to a low-speed expansion. Therefore, these cities were on the periphery of urban land expansion pattern of HCUA.

      Figure 4.  The spatial pattern evolution of built-up land expansion of the Harbin-Changchun urban agglomeration (HCUA), China in 2005‒2018

      Furthermore, other land use types were analyzed (Table 1). The area of forest and grass land showed a downward trend after increasing. From 2005 to 2018, the area of forest land decreased by 0.39%, and the area of grass decreased by 0.93%. Contrary to the trend of change of forest and grass land, from 2005 to 2018, the area of cultivated land increased by 1.08%, which was mainly affected by the Chinese government’s requirements for the delimitation of permanent basic farmland. The area of water decreased each year. From 2005 to 2018, the water area reduced by 1.12%, which was the largest among all land use types. The area of unused land showed an unstable and fluctuating trend. From 2005 to 2018, the total area of unused land increased by 0.73%. The decrease in the area of forest, grass, and water areas, along with the increase in the area of unused land brought a series of ‘urban diseases’ such as soil erodibility, wetlands loss, and air pollution, which have become obstacles to the sustainable development of the urban agglomeration (Li et al., 2015; Li and Zhou, 2018).

    • Based on the land expansion data obtained by overlapping land use maps of 2010 and 2018, 5% of cells were randomly selected as a sample and their corresponding spatial variable data were also extracted. Thus, the original training dataset established and used to train the RF model. A 3 × 3 moore neighborhood was also used to quantify the neighborhood effect of the PLUS model. We set the expansion coefficient (μk) to 0.2, the patch generation coefficient (δ) to 0.8 (Chen et al., 2019).

      Fig. 5 showed the global and partial enlarged view of simulation results by RNN-CA, ANN-CA and PLUS models, and the observed map of the Harbin-Changchun urban agglomeration in 2018. From the global view, we found that the simulation results obtained by the PLUS model were closer to the observed land use than those obtained by the traditional CA models. From the partial view, the traditional CA models showed an obvious cell cluster phenomenon, while the simulation results of the PLUS model avoided this and were more in line with the distribution pattern of observed land use patches.

      Figure 5.  Global and partial enlarged view of simulation results and observed map of the Harbin-Changchun urban agglomeration, China in 2018: a. global and partial enlarged view of the simulation result by RNN-CA ( recurrent neural network-cellular automata ); b. global and partial enlarged view of the simulation result by ANN (artificial neural network) -CA; c. global and partial enlarged view of the simulation result by PLUS ( patch-based land use simulation); d. global and partial enlarged view of the observed map

      To verify the accuracy of the model, the PLUS model was compared with the RNN-CA and ANN-CA models from on both the cell and patch scales. On the cell scale, the FOM and Kappa coefficient methods were used to compare the simulation results of the PLUS, ANN-CA, and RNN-CA models. To make the results comparable, we set the same parameters in the three models. Accuracy comparison among results simulated by the patch-based land use simulation (PLUS) model and other traditional CA models in 2018 were shown in Table 2. We found that the FOM and Kappa coefficients of the PLUS model were higher than those of the RNN-CA and ANN-CA models. On the patch scale, we selected four landscape metrics to quantify their landscape similarity: 1) Number of patches (NP); 2) The largest-patch index (LPI); 3) Euclidean nearest-neighbor distance metrics including the mean (ENN_MN); and 4) Mean Perimeter-Area Ratio (PARA_MN). We found that the four metrics obtained by the PLUS model simulation result were the closest to the metrics of the observed pattern in 2018, which substantiated the earlier finding that the land use pattern simulated by the PLUS model was closer to the observed land use pattern (Table 3).

      Landscape metricsPLUSRNN-CAANN-CA
      Kappa0.8950.8130.834
      Fom0.3210.2910.302
      Notes: ANN, artificial neural network-cellular automata; RNN, recurrent neural network

      Table 2.  Accuracy comparison between results simulated by the patch-based land use simulation (PLUS) model and those simulated by other traditional cellular automata (CA) models in 2018

      Landscape metricsObservedPLUSRNN-CAANN-CA
      NP22211206511425916328
      LPI67.50869.11474.86371.215
      PARA_MN1248.1251231.4261036.1241135.657
      ENN_MN216.367221.876329.554267.516
      Notes: PLUS, patch-based land use simulation; cellular automata; ANN, artificial neural network-cellular automata; RNN, recurrent neural network; NP is the number of patches; LPI is the largest-patch index; PARA_MN is the mean Perimeter-Area Ratio; ENN_MN is the Euclidean nearest-neighbor distance metrics including the mean

      Table 3.  Landscape metrics comparison among the simulation results and the observed pattern in 2018

    • Fig. 6a showed seven non-overlapping types of patches based on the MSPA method in 2018. We used the Conefor2.6 software (downloaded for free at http://www.conefor.org) in this step (Saura and Torné, 2009). Inside this software, IIC and PC were analyzed for 164 core patches larger than 10 km2 in the MSPA results, and the top 14 core patches with IIC ≥ 0.32 and PC ≥ 0.55 were selected (Table 4). Thus, 14 ecological sources were identified. As shown Fig. 6b, these sources were mainly composed of forest and water areas, which were distributed in the Changbai Mountains in the southeast of the region, the southern and eastern mountainous areas of Yanbian Prefecture, the Laoyeling and Weihuling areas in the eastern part of Jilin, the mountainous area at the junction of the east of Harbin and Mudanjiang, and the mountainous area at the edge of the northeast of Suihua.

      Figure 6.  Types of patches identified based on the morphological spatial pattern analysis (MSPA) method, and the spatial distribution of ecological sources of the Harbin-Changchun urban agglomeration, China in 2018: a. non-overlapping types of patches; b. spatial distribution of ecological sources

      SourceIICPCSourceIICPC
      197.79197.35681.1751.897
      29.59012.44790.7021.169
      31.7633.328100.3391.118
      41.8323.109110.3941.098
      51.8132.913120.3920.624
      61.6432.278130.3220.589
      71.4022.087140.3630.558

      Table 4.  Integral index of connectivity (IIC) and probability of connectivity (PC) of the top 14 core patches

      Different types of land cover have different degrees of resistance to the flow of material and energy between ecological sources. We assigned different cost values to each land use type to construct a basic resistance surface (Fig. 7a). The resistance value of water and built-up land were set to 1 and 500 respectively. The resistance values of other land types were assigned in the order of unused > cultivated > grass > forestry. Combined with the nighttime light intensity (TLI) value for each grid (Fig. 7b), the modified resistance surface was obtained (Fig. 7c). These above results were main layers used to construct the ESP of HCUA below.

      Figure 7.  Three layers used to construct the ecological security pattern (ESP) of Harbin-Changchun urban agglomeration (HCUA), China in 2018: a. the cost value of the basic resistance surface; b. nighttime light intensity (TLI); c. cost values of the modified resistance surface

      The Cost Distance analysis model in the ArcGIS 10.2 software was used to calculate the MCR values for the ecological sources. The three levels (high-level, medium-level and low-level) of ESP were divided at the mutation points where the number of grids change with the MCR values used as the thresholds. Fig. 8 showed different levels of the ESP in the HCUA. The low-level ESP covered an area of 250 219.35 km2, accounting for 77.63% of the total area, which was an insurmountable ecological bottom line in regional development. The exploration of urban land and other human activities harmful to the ecosystem should be strictly prohibited. The area of the medium-level ESP was 39 581.26 km2, accounting for 12.28% of the total area. This served as a buffer zone, surrounding the low-security level area, with rich ecological service functions. In this area, the recovery period after the ecosystem disturbance was short and the anti-disturbance ability was strong, and the development of urban construction land needs to be restricted. The area of high-level ESP was 32 522.39 km2, accounting for 10.09% of the total area, which was the transition area for the exchange and circulation of material and energy between the natural system and the urban system, guaranteeing the optimal state of the sustainable ecosystem. In this area, development and construction activities can be carried out according to specific conditions.

      Figure 8.  The ecological security pattern (ESP) of the Harbin-Changchun urban agglomeration (HCUA), China

    • Ecological civilization construction is an important policy for the Chinese government to protect the ecological environment, and ESP construction is an important means to balance ecological protection and socio-economic development in territorial planning. Therefore, we adopted the influence coefficient set of the driving factors to create a prediction module, and used the results of the ESP construction by setting excluded layer(s) that prohibit urban expansion to construct two EC scenarios: basic protection (BP) scenario, and optimal protection (OP) scenario. The importance rankings of each driving factor of six land use types were determined by the training of the RF algorithm in the PLUS model (Fig. 9).

      Figure 9.  Driving factors’ importance rankings of six land use types used in the scenario simulation: a. cultivated growth; b. forest land growth; c. grass land growth; d. built-up land growth; e. water land growth; f. unused land growth

      Table 5 showed layer(s) prohibiting urban expansion under the natural expansion (NE), basic protection (BP) and optimal protection (OP) scenarios of urban expansion in 2026. Considering the historical trend of land use change determines the land use demands in the future, we used the Markov chain to generate the transition probability matrix from 2010 to 2018 to predict the land use demands in 2026, which were regarded as the quantitative constraints for land expansion (Table 6).

      ScenariosLayer(s) prohibiting urban expansion
      NE scenarioNone
      BP scenarioArea of low-level ESP
      OP scenarioAreas of low-level and middle-level ESPs
      Notes: ESP, ecological security patten

      Table 5.  Layer(s) prohibiting urban expansion under the natural expansion (NE), basic protection (BP) and optimal protection (OP) scenarios of urban expansion of Harbin-Changchun urban agglomeration (HCUA), China in 2026

      Land use typeCultivatedForestryGrassWaterBuilt-upUnused
      Area of the land use demands /km21581781129121129266171468318632

      Table 6.  Land use demands obtained by the Markov chain of the Harbin-Changchun urban agglomeration (HCUA), China in 2026

      Fig. 10 illustrated the global and partial enlarged view of prediction results of the PLUS model under three scenarios with the land use demands as the scale constraint. Table 7 showed the comparison results of the scale of other land use patches occupied by expanded construction land under different scenarios in 2026. Which can be seen that under the NE scenario, the expanded urban construction land in 2026 will mainly be located in the low-level ESP zone, accounting for 62.32% of the total expansion, while only 27.38% and 10.30% will be located in medium-level ESP and high-level ESP zones respectively. The areas of other land use types occupied by expanded urban construction land, in descending order, will be water, cultivated land, forestry, grass, and unused land. Under the NE scenario, land use patches with important ecological service values will be largely occupied by urban construction land, which will not be conducive to the maintenance and development of the domestic ecosystem.

      Figure 10.  Global and partial enlarged view of prediction results obtained by the patch-based land use simulation (PLUS) model under different scenarios of the Harbin-Changchun urban agglomeration (HCUA), China in 2026: a. global and partial enlarged view of prediction result under the natural expansion (NE) scenario; b. global and partial enlarged view of prediction result under the basic protection (BP) scenario; c. global and partial enlarged view of prediction result under the optimal protection (OP) scenario

      Land use typeNE scenarioBP scenarioOP scenario
      Occupation scale of the ESP area /km2Occupation scale of the ESP area /km2Occupation scale of the ESP area /km2
      HighMediumLowTotalHighMediumTotalHigh
      Built-up land43.2259.31219.84322.47168.22300.12468.34524.09
      Forestry27.3651.08186.23264.6769.50167.34236.84192.41
      Grass23.2836.73105.49165.5053.11136.49189.60115.33
      Water29.66184.48220.12434.2664.32162.73227.05205.73
      Unused3.465.7336.1645.3510.6199.79111.40194.64
      Total126.90337.32767.781234365.66866.3412341234
      Notes: ESP, ecological security patten

      Table 7.  Scales of other land use types occupied by expansion of urban built-up land in the natural expansion (NE), basic protection (BP) and optimal protection (OP) scenarios of the Harbin-Changchun urban agglomeration (HCUA), China in 2026

      Under the BP scenario, the expanded urban construction land in the simulated land use pattern of 2026 will be mainly distributed in the middle-level ESP zone, accounting for 70.32% of the total expansion. Compared with the NE scenario, the proportion of new urban construction land in the high-level ESP zone will increase to 29.68%. The areas of other land use types occupied by expanded urban construction land, in descending order, are cultivated land, forest, water, grass, and unused land. In 2026, 18.43% of expanded urban construction land will be converted from the water area, which is 16.82% less than that in the NE scenario, and the forest area occupied by construction land will reduce by 2.26%, which is beneficial for reducing the impact of human activities on the ecological sources.

      Under the OP scenario, the scale of other land use types occupied by the expanded urban construction land in 2026 will be, in descending order, cultivated land, forest, water, unused land, and grass. Compared with the NE scenario and the BO scenario, the proportion of water occupied by expanded construction land will reduce by 18.55% and 1.73%, and the proportion of forest occupation will decrease by 5.87% and 3.61%. Although the scale of cultivated land occupied by expansion will be 18.15% of the total, it will be concentrated in high-level ESP which will have less impact on ecological security. The results showed that the land use pattern under the OP scenario can guarantee the stable development of the regional ecosystem.

    • According to China’s ‘National Main Functional Area Planning’ policy, the HCUA has a large number of lands used for the protection of agricultural production and ecological functions, and 52 national prohibited development areas (20 in Heilongjiang Province and 32 in Jilin Province). Among them, the main area of agricultural product accounts for 38.35% and the key area of ecological functions account for 43.90% of the total area. These restricted development areas coexist and account for too much in HCUA. Therefore, in the future development of HCUA, how to balance the relationship between land utilization and environmental protection should be carefully considered. Simulation of urban expansion under ecological constraints can help decision makers develop sound policies to alleviate contradictions between land utilization and ecological protection, which is of great importance for achieving the goal of the Chinese government’s ecological civilization construction plan.

      In addition, the ranking of the importance of the underlying drivers obtained from the LEAS module is generally in line with several other similar studies. For example, urban land expansion usually occurs in areas with large populations and high levels of economic development, among which, population is closely related to residential demands, and GDP promotes urban land expansion through industrial land and real estate construction (Deng et al., 2008). Another example is that forestry and grasslands are more likely to grow in natural areas of low-density human activities (with high NDVI, DEM and slope). At the same time, this study also explored some transition rules could not be found by the previous analysis methods. For example, the distance to core cities has a significant effect on the expansion of urban construction land, while the effects of distances to railway, highway, and water are not as obvious as expected. Core cities are the main gathering areas of productive factors and national support policies, and the expansion mainly occurred through the optimization and adjustment of existing land resources in these areas due to the scarcity of other land use types. Attracted by the concentration of productive factors in these areas, the land around core cities is more likely to change to construction land (Lv et al., 2021). In the past, accessibility factors such as the distances from rivers and traffic were considered to be the main controlling factors for the expansion of urban land (Fan et al., 2009). However, with the improvement of the urban agglomeration’s transportation network, the spatial spillover effects of information, industry, and urban land use have increased, leading to a relative weakening of the impact of these factors on urban land expansion (Xia et al., 2019).

      The existing studies on urban expansion simulation mostly focus on the improvement and accuracy verification of simulation models. There is a lack of discussion on how the simulation results can support the determination of urban planning policies. Combining the PLUS model with the construction of ESP, policymakers can obtain land use patterns and determine reasonable land polices under different scenarios. When we used the MCR model to construct and partition the ESP, the assignment of basic resistance values is the key step that affects the final result. Based on the research of Yu, K.J and other scholars, different land use types were assigned different resistance values (Li et al., 2011; Yu et al., 2012; Yang et al., 2017). Although the absolute values of the resistance coefficient need to be discussed, the assignment in this study is to express the relative value of different land use types, emphasizing the comparison of impacts of different land uses on ecological processes, rather than the absolute value. Therefore, the setting of the resistance coefficient in this study can meet the requirements of the analysis.

      We also discussed the policy implications to promote the coordinated development of HCUA. Under the NE scenario, the expansion of urban built-up land is not constrained by the ESP, and there is sufficient space for expansion. The planning policies should focus on avoiding the impact of urban built-up land expansion on the ecological environment. Ecological and natural reserves should be properly planned, and the level of intensive land use needs to be decreased to avoid the irreversible impact of the expansion of urban construction land on the ecological environment. Under the BP scenario, the expansion of urban built-up land is restricted by the ecological bottom line, which ensures the basic safety of the ecosystem, but the expansion space is limited. The stock planning of urban construction land and the economic structure transformation should be accelerated. Under the OP scenario, the expansion of urban built-up land gives priority to ecological protection, and the ecological environment is better protected, but the expansion space of urban built-up land is more limited than in the BP scenario, and the scale of cultivated land occupation is relatively high. Planning policies of ‘urban renewal’ can be formulated to strengthen the protection of cultivated land. Considering the particularity of the research area, it can be gradually achieved by improving the quality of old residential areas, transforming and renewing old industrial areas, and developing the efficient use of areas. Other strategies such as spatial regulation and the establishment of a balance mechanism for arable land occupation and compensation can also alleviate the negative effects of urban expansion on arable land protection and ecological protection

      Although this study is of great significance for enhancing the application range of the PLUS model, and helps policymakers determine reasonable land polices, it still has some limitations. Firstly, the selection of the driving factors for land use change needs to be explored further. 12 driving factors were selected from the aspects of natural environment, socio-economics, and location according to the characteristics of land use change in the study area, but the widespread competition and cooperation between cities within urban agglomerations were ignored due to the difficulty of quantifying these parameters. Therefore, models should be used in subsequent studies to consider these influencing factors, and analyze the impact of urban interactions within urban agglomerations on land use expansion. Considering the model requirements, simulation accuracy, and calculation efficiency, this study resampled the original 30 m pixels into 250 m pixels, which undoubtedly reduced the accuracy of the results. Undoubtedly, the simulation accuracy and the simulation operation speed of the PLUS model needs to be improved.

    • Taking the HCUA as the research object, an effective analytical framework was proposed to simulate land use patterns under NE and EC scenarios. The main conclusions are as follows:

      (1) The spatiotemporal evolution of urban land expansion presented an obvious “center-periphery” characteristic with the decrease of forest land and the increase of unused land. To ease the contradiction, we introduced a PLUS model to mine the rules of land use dynamics with the LEAS module and simulate the change of land use patches with the CARS module. By comparing the simulated and observed land use patterns in 2018, the PLUS model was proven to be effective in the investigation of urban expansion at the scale of urban agglomerations, with high Kappa coefficient and FOM value.

      (2) EC scenarios were designed based on the results of ESP construction using the MSPA and MCR models. It is also worthy to note that 14 core patches were mainly distributed in mountainous areas in the east, and the areas identified as low, medium, and high-level ESP each occupied a total land area of 250 219.35 km2, 39 581.26 km2, and 32 522.39 km2, respectively, accounting for 77.63%, 12.28% and 10.09% of the study area.

      (3) Based on a comparative analysis of the prediction results of the PLUS model under three scenarios with the land use demands set as the scale constraints, it was demonstrated that the distribution pattern of construction land and the scale of other land use patches occupied by construction land can maintain healthy ecosystems under the OP scenario better than under the BP scenario, while the simulation results under the NE scenario without ECs posed the greatest threat to the ecosystem.

      On the whole, the PLUS model can improve the simulation accuracy at urban agglomeration scale compared with other CA models. Combined with MSPA and MCR methods, the PLUS model can also be used in other spatial simulations of urban agglomerations under ecological constraints, which has broadened the application scope of the PLUS model. In the Harbin-Changchun urban agglomeration, planning decision-makers should avoid using the NE scenario, selectively implement the BP scenario, and achieve the OP scenario to make trade-off strategies. However, in other urban agglomerations, due to different resources and environment, the future land use scenarios under ecological constraints are also different. Whether it is suitable for planning decision-makers to adopt the same planning strategies needs to be further verified.

Reference (58)

Catalog

    /

    DownLoad:  Full-Size Img  PowerPoint
    Return
    Return