| Urban expansion caused by urbanization and human activities changes the supply of ecosystem services,and then affects the regional Ecological service value（ESV）.It is very important to evaluate and simulate the dynamic relationship between urban expansion and ecological service value objectively and comprehensively to promote regional and urban sustainable development.Most of the existing urban expansion simulation models only use a data time slice to extract the neighborhood effect,which is inconsistent with the nature of long-term spatio-temporal dependence of neighborhood interaction,resulting in limited model accuracy.Deep learning models can simultaneously extract spatial and temporal features of long time series data.Coupled deep learning algorithm and Cellular automata（CA）can reflect spatiotemporal neighborhood features more accurately and improve the simulation performance of urban expansion model.In addition,studies on the dynamic relationship between urban expansion and ecosystem services pay more attention to the coastal humid areas with rapid urban development.Lanzhou is located in the inland area of northwest China,with less precipitation and large amount of air evaporation,combined with the features of valley city and important town in northwest China,which affects the social and economic stability of the semi-arid region of northwest China.In this context,quantitative analysis of lanzhou’s urban expansion and its impact on ecosystem services is of great significance to comprehensively understand the evolution of urban ecosystem services value in the semi-arid region of northwest China.In this study,deep learning and cellular automata models were combined to simulate urban expansion in 2030 under four scenarios（Natural growth,Ecological protection,Economic development and Ecological protection-economic development）in Lanzhou,a semi-arid valley basin in northwest China.On this basis,we use the agent-based method to calculate the ecological service value of Lanzhou under different scenarios from 2000 to 2020 and from 2020 to 2030,and use different land use and ecosystem service types to represent and analyze the changes of ecological service value under different scenarios.The main conclusions are as follows:（1）Using measure index,urban expansion and landscape pattern analysis methods to analyze the quantitative structure of land use and spatial and temporal pattern evolution characteristics of urban expansion in Lanzhou.From the perspective of spatial distribution,grass land is the main land use type in Lanzhou,accounting for more than 50% of the total land area.From the perspective of time distribution,from2000 to 2020,the newly added built-up land in the study area is concentrated in the junction of Gaolan and Yongdeng,namely lanzhou New Area,and the main occupied land use type is cultivated land.From 2010 to 2020,the expansion area,expansion rate,expansion intensity and dynamic attitude of built-up land in the study area are the largest,which proves that the loss of high ecological value land use types（water,forest land）is greater in this period,and the expansion of urban built-up land is more obvious.From 2010 to 2020,there were frequent state transitions among different land types in the study area,which resulted in higher overall landscape fragmentation and damaged the connectivity established by dominant patches in the earlier period.The continuous expansion of construction land makes its shape tend to be regular and thus increases the aggregation degree of the overall landscape.（2）The ecological service value of Lanzhou from 2000 to 2020 was calculated based on the proxy method,and analyzed the impact of land use on the ecological service value.The study found that the ecological service value of Lanzhou decreased first and then increased,and the total ecological service value decreased from 5.482 billion yuan in 2000 to 5.420 billion yuan in 2020.This trend is different from the result that the value of coastal humid ecological services gradually decreases with the continuous expansion of urban built-up land.However,the ecological service value of the study area increased with the expansion of urban built-up land.Climate regulation and hydrological regulation are the most important ecosystem service types in Lanzhou city.The areas with the lowest ecological service value are mainly distributed in the main urban area,Lanzhou New Area and surrounding areas,and the spatial distribution of built-up land is basically consistent.The medium ecological service value is mainly distributed in cultivated land and forest land in Yongdeng.The water across the main urban area and the woodland and grass land in Yuzhong have the highest ecological service value.The results of profit and loss,transfer matrix,contribution rate and variation rate of ecological service value showed that forest land and water were the main contributors to the increase of ecological service value.（3）Based on the coupling model of deep learning and cellular automata to simulate urban expansion,the model can simultaneously capture the spatial correlation and long-term time dependence of land use temporal data and driving factors,and obtain more accurate simulation results of land use change than traditional models.Based on previous studies and existing literature,four different coupling models of deep learning and cellular automata,CNN-LSTM-CA,LSTM-CA,ANN-CA and RNN-CA.The overall accuracy of the four models ranked as CNN-LSTM-CA（0.9283）> LSTM-CA（0.9101）> ANN-CA（0.8837）> RNN-CA（0.8326）.CNNLSTM-CA model has the highest accuracy and the best simulation performance,and the simulation results are closest to the actual land use in spatial distribution,indicating that the comprehensive consideration of spatio-temporal neighborhood characteristics is beneficial to the improvement of the model accuracy.Therefore,the CNN-LSTMCA model was adopted in this study to simulate the future urban expansion scenario of the study area.（4）Based on a series of national policies on ecological protection and economic development,four different scenarios of natural growth,ecological protection,economic development and ecological protection-economic development were set up to predict the urban expansion and ecological service value in the study area in 2030.In2030,the total ESV under different scenarios ranked as Ecological protection（5.807 billion yuan）> Ecological protection-Economic development（5.762 billion yuan）>Natural growth（5.648 billion yuan）> Economic development（5.399 billion）.From2020 to 2030,the results of urban expansion show that the built-up land under the four scenarios continues to increase,with the most dramatic expansion under the natural growth scenario and the least expansion under the ecological protection scenario.The results of ecological service value showed that the total ecological service value of the other three scenarios increased to different degrees compared with 2020,except that the total ecological service value of the economic development scenario continued to decline,and the increase of the ecological protection scenario was the most obvious,followed by the ecological protection-economic development scenario.Through the comprehensive characterization analysis of the value of ecological services by different ecosystem service types and different land use types,it is found that land use change under ecological protection-economic development scenario can improve economic development on the basis of protecting ecological benefits,and is more conducive to the sustainable development of cities. |
PDF Full Text Request