by Sophia Christine Linke
Policy makers, who are faced with decision making about spatial activities in urban areas, can be guided by means of computer based decision support systems. An example of such support systems are land use models, offering important benefits and assistance. The land use model METRONAMICA is such a decision support system. Its allocation methodology is based on a cellular automaton. The model is interactive and user-friendly, facilitates learning and employs complex and weakly-structured decision contexts.
For the thesis at hand a study area in the Netherlands, the municipality Weert, was chosen. Here, several major changes and redevelopments are planned but different opinions considering future growth perspectives exist. METRONAMICA could be used for developing them. The question is, if the model provides accurate results and usable support when being applied at a high resolution and local level.
In the course of the study METRONAMICA’s possible application is tested and evaluated. First METRONAMICA is compared to a selection of other spatial models (CLUE, LandUseScanner, UrbanSim and SLEUTH). Keeping in mind the local scale of application, the availability of suitable data then is assessed, input datasets are created and suitable variables are chosen while carrying out the model calibration. For assessing the quality of calibration, two measurements are applied: Kappa statistics and Zipf’s law. Special attention is paid to the applied cell size (25 x 25 m) and the size of the neighbourhood the cellular automaton takes into consideration (8 cell and 16 cell radius neighbourhood). In order to evaluate METRONAMICA’s possibilities for local planning support, several development scenarios are created and simulated until the year 2040.
The results show, that a local application seems to be possible, but more detailed data should be inserted into the modelling environment to be able to model more detailed land use functions (e.g. splitting up the urban land uses). Then, also much more detailed transition rules could be implemented, accounting for dynamics within the urban environment. The simulation results obtained show a remarkable difference between the 8- and 16-cell neighbourhoods. The model was calibrated with an 8-cell neighbourhood, and then the neighbourhood extended to 16 cells, resulting in very blobby simulation result. The model should hence fully be calibrated again. Changing the neighbourhood while expecting to achieve better simulation results, seems only possible when adapting also the transition rules.
For the thesis at hand a study area in the Netherlands, the municipality Weert, was chosen. Here, several major changes and redevelopments are planned but different opinions considering future growth perspectives exist. METRONAMICA could be used for developing them. The question is, if the model provides accurate results and usable support when being applied at a high resolution and local level.
In the course of the study METRONAMICA’s possible application is tested and evaluated. First METRONAMICA is compared to a selection of other spatial models (CLUE, LandUseScanner, UrbanSim and SLEUTH). Keeping in mind the local scale of application, the availability of suitable data then is assessed, input datasets are created and suitable variables are chosen while carrying out the model calibration. For assessing the quality of calibration, two measurements are applied: Kappa statistics and Zipf’s law. Special attention is paid to the applied cell size (25 x 25 m) and the size of the neighbourhood the cellular automaton takes into consideration (8 cell and 16 cell radius neighbourhood). In order to evaluate METRONAMICA’s possibilities for local planning support, several development scenarios are created and simulated until the year 2040.
The results show, that a local application seems to be possible, but more detailed data should be inserted into the modelling environment to be able to model more detailed land use functions (e.g. splitting up the urban land uses). Then, also much more detailed transition rules could be implemented, accounting for dynamics within the urban environment. The simulation results obtained show a remarkable difference between the 8- and 16-cell neighbourhoods. The model was calibrated with an 8-cell neighbourhood, and then the neighbourhood extended to 16 cells, resulting in very blobby simulation result. The model should hence fully be calibrated again. Changing the neighbourhood while expecting to achieve better simulation results, seems only possible when adapting also the transition rules.
more about land use models:
No comments:
Post a Comment