University of Twente Student Theses


Using spatial logistic regression analysis to support CA based modelling of informal development: Dar es Salaam, Tanzania

Bitalko, Demeke Ashenafi (2012) Using spatial logistic regression analysis to support CA based modelling of informal development: Dar es Salaam, Tanzania.

[img] PDF
Abstract:The application of different types of land use models has been a way to better understand the mechanisms of IS expansion, which is a manifestation of rapid urbanisation in Dar es Salaam. In tacking the challenges of IS development, modelling has been used as tool to support planning and policy making processes through accentuating proactive measures. So far, empirical logistic regression (LR) and the dynamic cellular automata (CA) modelling approaches have been used for modelling IS developments in Dar es Salaam. Their capacity in making use of explanatory variables to predict the probability of cells’ to be developed to new IS lands and showing the pattern of the expansion in a spatially explicit way has made (LRM) powerful tools in modelling land use developments. However, their limitation in showing the temporal dynamics and developing scenarios has been a limitation in manipulating the full potential of LR the models. On the other hand, CA models except for their complication in calibration and inclusion of global factors of land use change, are able to make a better representation in land use dynamics and are good in developing scenarios to see ‘what if’ conditions. Therefore, the focus of this research has been making a structured approach to integrate LR analysis with dynamic CA based modelling with a theme of providing more representative interpretable and structured information on the dynamics of IS expansion by making use of the cumulative benefits of the integrated model. The integration of the two approaches has been done by incorporating both global and local factors of IS expansion for predicting the potentials the cells to be developed in LR model, while the CA based approach uses the predictions to simulate IS expansion dynamics by constraining the quantity based on IS expansion land demand. Two different levels of integration were used to model the IS expansion dynamics in Dar es Salaam. The first level of integration was made between three LRMs, 1982-1992 (model-A), 1982-2002 (model-B), and 1992-2002 (model-C), based on the IS expansion, in order to achieve better model performance., while the second level of integration was between LRM and dynamic CA based model in order to increase the interpretability and representativeness of the models to the phenomena of IS expansion. The evaluation of the models and comparisons of the results have provided with the fact that integrated models have a better performance in simulating IS expansion in DAR. In the evaluation of the first level model integration has been very high for integrated models than the individual LRMs while the results from the overall integrated model were able to provide a more realistic IS expansion simulations which was found to be representative of the observed IS expansion in DAR. According to the results, the most important key drivers in the rule definition for calculating probability of cells were distance to minor roads, distance to existing ISs with both negative relationship with the expansion of ISs in DAR. The probabilities of the cells to be informal is hence constrained by the IS expansion land use demand observed in order to simulate the IS expansion. At every simulation the models update the overall probabilities of the cells through the updating local probabilities based on the simulation results. The logical justification and approaches of model the integration have shown a promising results and increased the interpretability of the IS expansion dynamics in DAR. Moreover, the application of these results and approaches of the study can facilitate an informed policy and decision making processes decision making processes related to IS expansions Key words: Informal settlements, LR modelling, CA modelling, model integration
Item Type:Essay (Master)
Faculty:ITC: Faculty of Geo-information Science and Earth Observation
Programme:Geoinformation Science and Earth Observation MSc (75014)
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page