Coupling cellular automata with area partitioning and spatiotemporal convolution for dynamic land use change simulation

Multiple scenario land use simulation based on a coupled MOGA-PLUS model: a case of the Yellow River Basin

Land use changes have profoundly influenced global environmental dynamics. The Yellow River (YR), as the world's fifth-longest river, significantly contributes to regional social and economic growth due to its extensive drainage area, making it a key global player. To ensure ecological stability and coordinate land use demand, modeling the future land allocation patterns of the Yellow River Basin (YRB) will assist in striking a balance between land use functions and the optimization of its spatial design, particularly in water and sand management. In this research, we used a multi-objective genetic algorithm (MOGA) with the PLUS model to simulate several different futures for the YRB's land use between 1990 and 2020 and predict its spatial pattern in 2030. An analysis of the spatiotemporal evolution of land use changes in the YRB indicated that construction land expansion is the primary driver of landscape pattern and structure changes and ecological degradation, with climate change also contributing to the expansion of the watershed area. On the other hand, the multi-scenario simulation, constrained by specific targets, revealed that economic development was mainly reflected in land expansion for construction. At the same time, grassland and woodland were essential pillars to support the region's ecological health, and increasing the development of unused land emerged as a potential pathway towards sustainable development in the region. This study could be used as a template for the long-term growth of other large river basins by elucidating the impacts of human activities on land use and rationalizing land resource allocation under various policy constraints.

Urban expansion simulation with an explainable ensemble deep learning framework

Urban expansion simulation is of significant importance to land management and policymaking. Advances in deep learning facilitate capturing and anticipating urban land dynamics with state-of-the-art accuracy properties. In this context, a novel deep learning-based ensemble framework was proposed for urban expansion simulation at an intra-urban granular level. The ensemble framework comprises i) multiple deep learning models as encoders, using transformers for encoding multi-temporal spatial features and convolutional layers for processing single-temporal spatial features, ii) a tailored channel-wise attention module to address the challenge of limited interpretability in deep learning methods. The channel attention module enables the examination of the rationality of feature importance, thereby establishing confidence in the simulated results. The proposed method accurately anticipated urban expansion in Shenzhen, China, and it outperformed all the baseline methods in terms of both spatial accuracy and temporal consistency.

A cellular automata model coupled with partitioning CNN-LSTM and PLUS models for urban land change simulation

Urbanisation is a key aspect of land use change (LUC), and accurately modelling of urban LUC is crucial for sustainable development. Cellular automata (CA) are widely used in LUC research. However, previous studies have overlooked the significant temporal dependence and spatial heterogeneity associated with LUC. To address these gaps, this study proposes a novel model called KCLP-CA, which integrates k-means, a convolutional neural network (CNN), a long and short-term memory neural network (LSTM), and the popular patch-generation land use model (PLUS). Initially, k-means and CNN are utilised to address spatial heterogeneity, while LSTM tackles temporal dependence. The LSTM and land expansion analysis strategy (LEAS) models of PLUS are employed to obtain land use conversion probability maps. Finally, a simulation of land use dynamic change was conducted using a linear weighted fusion conversion probability map that accounts for random factors. To validate the KCLP-CA model, land use data collected from Hangzhou between 1995 and 2000 were employed. The results showed that the KCLP-CA model outperformed traditional methods, including artificial neural networks and random forest model, with the figure of merit (FoM) index increasing from 2.12% to 4.19%. Random forest analysis of drivers impacting LUC revealed that distance to water and road network density exerted the greatest influence on urban land development in Hangzhou. Incorporation of various policy planning factors affecting urban development yielded simulation results aligning more closely with reality, resulting in a FoM index increase of 1.64-1.76%. In summary, the model developed in this study combines the strengths of two sub models to deliver an accurate and effective simulation of future land use.

Future carbon storages of ecosystem based on land use change and carbon sequestration practices in a large economic belt

Assessments of ecosystem carbon storage are needed to form the scientific basis for carbon policies. Due to lack of data, there are few accurate, large-scale, and long-term predictions of ecosystem carbon storage. This study used the Distributed Land-Use Change Prediction (DLUCP) model with ten socioeconomic and two climate change scenarios for a total of 20 combinations that take into account population increase, technology innovation, climate change, and Grain for Green Project to make high-resolution predictions of land use change in the Yangtze River Economic Belt. Low and high carbon sequestration practices were considered to predict future carbon densities. Land use change data, carbon densities data, and the InVEST model were used to predict changes in ecosystem carbon storage from now to 2070. The results show a slight increase (1.88-4.17%) in carbon storage in the study area only based on land use change. Grain for Green Project has the largest impact on carbon storage among population increase, technology innovation, climate scenarios, and Grain for Green Project, which increases carbon storage by 4.17%. After the implementation of carbon sequestration practices, there is an increase in carbon storages from 28.51 to 56.77% in the study area from now to 2070, and increasing carbon storages of forest in each stream and carbon storage of cropland in downstream are efficient ways to achieve carbon neutralization.

Scales and Historical Evolution: Methods to Reveal the Relationships between Ecosystem Service Bundles and Socio-Ecological Drivers-A Case Study of Dalian City, China

Ecosystem service (ES) bundles can be defined as the temporal and spatial co-occurrence of ESs. ES bundles are jointly driven by socio-ecological factors and form at different scales. However, in recent research, a few studies have analyzed the dynamic evolution and driving mechanisms of ES bundles at different scales. Therefore, this study explored the spatial patterns of six ESs supplied in Dalian (China) from 2005 to 2015 at three spatial scales, determining the distribution and evolution patterns of ES bundles and their responses to socio-ecological driving factors. Our results are as follows: (1) We identified four ES bundles representing ecological conservation, water conservation, ecological depletion, and food supply. The developmental trajectory of each ES bundle could be attributed to the combined effects of environmental conditions and urban expansion. In particular, the water conservation bundle and food supply bundle were changed to the ecological depletion bundle. Given the ongoing urbanization, the conflict between ESs has intensified. (2) The impact of socio-ecological driving factors on ES bundles vary with scale. At three spatial scales, the digital elevation model (DEM) and normalized difference vegetation index (NDVI) had a great impact on ES bundles. Urbanization indicators also strongly explain the spatial distribution of ES bundles at the county and grid scales. The interaction factor detector shows that there is no combination of mutual weakening, indicating that the formation of ES bundles is driven by multiple factors in Dalian. Overall, this study used a more holistic approach to manage the ecosystem by studying the temporal-spatial dynamics of the multiple ESs.

An Integrated Spatial Autoregressive Model for Analyzing and Simulating Urban Spatial Growth in a Garden City, China

In the past, the research on models related to urban land-use change and prediction was greatly complicated by the high precision of models. When planning some garden cities, we should explore a more applicable, specific, and effective macro approach than the community-level one. In this study, a model consisting of spatial autoregressive (SAR), cellular automata (CA), and Markov chains is constructed. One It can well-consider the spatial autocorrelation and integrate the advantages of CA into a geographical simulation to find the driving forces behind the expansion of a garden city. This framework has been applied to the urban planning and development of Chengdu, China. The research results show that the application of the SAR model shows the development trend in the southeast region and the needs to optimize the central region and protect the western region as an ecological reserve. The descriptive statistics and the spatial autocorrelation of the residuals are reliable. The influence of spatial variables from strong to weak is distance to water, slope, population density, GDP, distance to main roads, distance to railways, and distance to the center of the county (district). Taking 2005 as the initial year, the land-use situation in 2015 was simulated and compared with the actual land-use situation. It seems that the Kappa coefficient of the construction-land simulation is 0.7634, with high accuracy. Therefore, the land use in 2025 and 2035 is further simulated, which provides a reference for garden cities to formulate a reasonable urban space development strategy.

Machine learning in modelling land-use and land cover-change (LULCC): Current status, challenges and prospects

Land-use and land-cover change (LULCC) are of importance in natural resource management, environmental modelling and assessment, and agricultural production management. However, LULCC detection and modelling is a complex, data-driven process in the remote sensing field due to the processing of massive historical and current data, real-time interaction of scenario data, and spatial environmental data. In this paper, we review principles and methods of LULCC modelling, using machine learning and beyond, such as traditional cellular automata (CA). Then, we examine the characteristics, capabilities, limitations, and perspectives of machine learning. Machine learning has not yet been dramatic in modelling LULCC, such as urbanization prediction and crop yield prediction because competition and transition between land cover types are dynamic at a local scale under varying natural drivers and human activities. Upcoming challenges of machine learning in modelling LULCC remain in the detection and prediction of LULC evolutionary processes if considering their applicability and feasibility, such as the spatio-temporal transition mechanisms to describe occurrence, transition, spreading, and spatial patterns of changes, availability of training data of all the change drivers, particularly sequence data, and identification and inclusion of local ecological, hydrological, and social-economic drivers in addressing the spectral feature change. This review points out the need for multidisciplinary research beyond image processing and pattern recognition of machine learning in accelerating and advancing studies of LULCC modelling. Despite this, we believe that machine learning has strong potentials to incorporate new exploratory variables in modelling LULCC through expanding remote sensing big data and advancing transient algorithms.

Analysis of the Driving Force of Land Use Change Based on Geographic Detection and Simulation of Future Land Use Scenarios

Land use and land cover changes (LULCC) are the result of the combined action of many influencing factors such as nature, society, economy and politics. Taking Chongqing as an example, the driving factors of urban land expansion in Chongqing from 1999 to 2019 are analyzed using a geographic detection (GD) method. Based on this analysis, a land use scenario of Chongqing in 2029 is simulated by an Artificial Neural Network-Cellular Automata model. The results of the analysis of factors affecting land use change show that five factors have a significance >0.05: population, distance from central city, school density, GDP and the distance from railway, showing that these factors have a high impact on LULCC in Chongqing. In addition, the results of risk detection analysis show that areas with a population >50/km2; the areas with a distance <200 km from the city center; areas with a school density >5/km2; areas with a high GDP; and areas with a distance <25 km from the railway have a greater impact on urban land use change than other areas. The land use scenario in 2029 also is simulated based on the land use situation in 2019. The predicted results clearly reflect a land use change trend of increasing urban land and decreasing agricultural land in the region. These land use changes are especially related to the expansion of the population, economy, roads, and schools in the process of urbanization. This analysis also shows that the GD-ANN-CA model developed in this paper is well suited to urban land use simulation.

Gray Forecast of Ecosystem Services Value and Its Driving Forces in Karst Areas of China: A Case Study in Guizhou Province, China

A sound ecosystem is the prerequisite for the sustainable development of human society, and the karst ecosystem is a key component of the global ecosystem, which is essential to human welfare and livelihood. However, there remains a gap in the literature on the changing trend and driving factors of ecosystem services value (ESV) in karst areas. In this study, Guizhou Province, a representative region of karst mountainous areas, was taken as a case to bridge the gap. ESV in the karst areas was predicted, based on the land use change data in 2009-2018, and the driving mechanisms were explored through the gray correlation analysis method. Results show that a total loss of CNY 21.47 billion ESV from 2009 to 2018 is due to the conversion of a total of 22.566% of the land in Guizhou, with forest land as the main cause of ESV change. By 2025 and 2030, the areas of garden land, water area, and construction land in Guizhou Province will continue to increase, whereas the areas of cultivated land, forest land, and garden land will decline. The total ESV shows a downward trend and will decrease to CNY 218.71 billion by 2030. Gray correlation analysis results illuminate that the total population and tertiary industry proportion are the uppermost, among all the driving factors that affect ESV change. The findings in this study have important implications for optimizing and adjusting the land use structure ecological protection and will enrich the literature on ESV in ecologically fragile areas.

Modeling the response of ecological service value to land use change through deep learning simulation in Lanzhou, China

Land use (LU) changes caused by urbanization, climate, and anthropogenic activities alter the supply of ecosystem services (ES), which affects the ecological service value (ESV) of a given region. Existing LU simulation models extract neighborhood effects with only one data time slice, which ignores long-term dependence in neighborhood interactions. Previous studies on the dynamic relationship between LU change and ES in semi-arid areas is rare than that in humid coastal areas. Here, we selected a semi-arid city, Lanzhou, in Northwest China as the study area, to simulate LU changes in 2030 under natural growth (NG), ecological protection (EP), economic development (EP), and ecological protection-economic development (EPD) scenarios, using a novel deep learning method, named CL-CA. Convolutional neural network and long short term memory (CNN-LSTM) with cellular automata (CA) were utilized to extract the spatiotemporal neighborhood features. The overall simulation performance of the proposed model was larger than 0.92, which is surpassed that of LSTM-CA, artificial neural network (ANN)-CA, and recursive neural network (RNN)-CA. Ultimately, we utilized LU and ES to quantitatively evaluate the ESV changes. The results indicated that: (1) The variable trend of ESV in arid area is different from that in coastal humid areas. (2) Forest land and water were the main factors that affect the ESV change. (3) The EPD scenario was more suitable for sustainable urban development.

Urban Expansion Simulated by Integrated Cellular Automata and Agent-Based Models; An Example of Tallinn, Estonia

From 1990 to 2018, built-up areas in Tallinn, Estonia’s capital city, increased by 25.03%, while its population decreased by −10.19%. Investigating the factors affecting urban expansion and modeling it are critical steps to detect future expansion trends and plan for a more sustainable environment. Different models have been used to investigate, predict, and simulate urban expansion in recent years. In this paper, we coupled the cellular automata, agent-based, and Markov models (CA–Agent model) in a novel manner to address the complexity of the dynamic simulation, generate heterogeneity in space, define more complicated rules, and employ the suitability analysis. In the CA–Agent model, cells are dynamic agents, and the model’s outcome emerges from cellular agents’ interactions over time using the rules of behavior and their decisions concerning the adjacent neighboring cells and probabilities of spatial changes. We performed the CA–Agent model run two times for 2018 and 2030. The first simulated results were used to validate the performance of the model. Kappa showed 0.86, indicating a relatively high model fit, so we conducted the second 12-year run up to the year 2030. The results illustrated that using these model parameters, the overall built-up areas will reach 175.24 sq. km with an increase of 30.25% in total from 1990 to 2030. Thus, implementing the CA–Agent model in the study area illustrated the temporal changes of land conversion and represented the present spatial planning results requiring regulation of urban expansion encroachment on agricultural and forest lands.

Landscape pattern change simulations in Tibet based on the combination of the SSP-RCP scenarios

Monitoring landscape pattern change can provide spatial explicit basis for future landscape management. The future socioeconomic and climate change drivers should be systematically combined in landscape pattern monitoring, while they are often regarded as independent parameters in landscape monitoring models. This study sought to project the detailed landscape pattern change based on landscape composition and configuration in Tibet by 2030, and combined the shared socioeconomic pathways (SSPs) and representative concentration pathways (RCPs). The results showed area of the unused land and forest will reduce by a minimum standard of 11.42 × 10⁴ and 9.04 × 10⁴ km² from 2010 to 2030, respectively. Other land use types will increase, and the highest increase in grassland will be 9.30 × 10⁵ km². Combined SSP1 and RCP2.6 scenario show high landscape aggregation and low edge density on cultivated land, urban land and grassland in Tibet as a whole. However, in typical cultivated and urban landscape, the abovementioned rule is appeared in the combined SSP4 and RCP6.0 scenario. These findings stress the importance of systematically modeling the socioeconomic demand and climate change in landscape pattern monitoring, and using both landscape composition and configuration indexes for scenario evaluation.