Coupling Multi-view Spatio-temporal Network and Vector Cellular Automatafor Uban Land Use Simulation
The dynamics of urban land use change originates from spatial interactions,and fully learning the spatial roles is the focus of improving the accuracy of land use simulation.In this paper,a vector metacellular automata model based on multi-view mechanism is proposed.Taking Los Angeles County in the United States as the study area,the changes of urban land use from 2010 to 2020 are simulated.The study obtains the following conclusions:i)The cross-view mechanism of MSTGAT-VCA can effectively learn the spatio-temporal interactions of urban parcel changes.iii)The overall accuracy(OA)and quality factor(FoM)of the MSTGAT-VCA model reaches 0.9322 and 0.4193,respectively,which is significantly improved compared with that of the traditional CA model.The study shows that the MSTGAT-VCA model can better simulate land use change than the traditional model.
land use simulationvector cellular automatagraph deep learning
NETL
NSTL
万方数据
