A platform for research: civil engineering, architecture and urbanism
A platform for research: civil engineering, architecture and urbanism
LandSys II: Agent-Based Land Use–Forecast Model with Artificial Neural Networks and Multiagent Model
AbstractThis paper extends the previous LandSys I to introduce artificial neural networks (ANNs) into the framework of cellular automata (CA), multiagents, and geographic information system (GIS) to forecast land-use change at the grid cell level (50×50 m). In the model, the temporal and spatial interactions of land-use change are described by CA where transition rules are defined by ANNs to reduce the tedious work of parameter calibration in LandSys I. Compared with LandSys I, an improved multiagent model in LandSys II captures both zoning policies and human decision-making behaviors. The effect of multiple human decision-making behaviors (e.g., governments, households, developers) on land-use change has been quantified. Based on the historical GIS data for Orange County, Florida, the model has a higher predictive ability (87.7%, compared to 85.7% in LandSys I) for land-use change from Year 1990 to 2000. It is also found that either increasing hidden layers in ANNs or the use of multiagent models improves prediction accuracy. A comparison between LandSys I and II indicates that both models are viable; however, LandSys II is freely transferable and is more suitable for land-use forecasting, whereas LandSys I is more appropriate for evaluating the interconnections between land use and its affecting variables.
LandSys II: Agent-Based Land Use–Forecast Model with Artificial Neural Networks and Multiagent Model
AbstractThis paper extends the previous LandSys I to introduce artificial neural networks (ANNs) into the framework of cellular automata (CA), multiagents, and geographic information system (GIS) to forecast land-use change at the grid cell level (50×50 m). In the model, the temporal and spatial interactions of land-use change are described by CA where transition rules are defined by ANNs to reduce the tedious work of parameter calibration in LandSys I. Compared with LandSys I, an improved multiagent model in LandSys II captures both zoning policies and human decision-making behaviors. The effect of multiple human decision-making behaviors (e.g., governments, households, developers) on land-use change has been quantified. Based on the historical GIS data for Orange County, Florida, the model has a higher predictive ability (87.7%, compared to 85.7% in LandSys I) for land-use change from Year 1990 to 2000. It is also found that either increasing hidden layers in ANNs or the use of multiagent models improves prediction accuracy. A comparison between LandSys I and II indicates that both models are viable; however, LandSys II is freely transferable and is more suitable for land-use forecasting, whereas LandSys I is more appropriate for evaluating the interconnections between land use and its affecting variables.
LandSys II: Agent-Based Land Use–Forecast Model with Artificial Neural Networks and Multiagent Model
Zhao, Liyuan (author) / Peng, Zhong-Ren
2015
Article (Journal)
English
LandSys II: Agent-Based Land Use—Forecast Model with Artificial Neural Networks and Multiagent Model
| British Library Online Contents | 2015
Using neural networks and GIS to forecast land use changes: a Land Transformation Model
| Online Contents | 2002
Development of Artificial Neural Network Model for Indian Steel Consumption Forecast
| Springer Verlag | 2024