Urban Spatial Scenario Design Modelling (USSDM) in Dar es Salaam: Background Information Modelling urban settlement dynamics in Dar es Salaam Revision: 2 (July 2013) Prepared by: Katja Buchta TUM team contributors Prof. Dr. Stephan Pauleit Andreas Printz Florian Renner Ardhi team contributors Prof. Wilbard J. Kombe Deusdedit Kibassa Created within the CLUVA-project (Climate change and Urban Vulnerability in Africa) www.cluva.eu Project co-funded by the European Commission within the Sixth Framework Programme (2002-2006) This document is designed to provide background information for Urban Growth spatial dynamics in Dar es Salaam. In case you have feedback or questions, please send them to buchta@wzw.tum.de
Table of Contents Glossary... 2 1 Summary... 3 2 Model aim and usage... 3 3 Model concept and scope... 3 3.1 Population driven Model... 3 3.2 GIS based model... 4 3.3 Cell/Raster based model... 4 3.4 Conceptual model... 4 3.5 Temporal scope of the model... 5 3.6 Spatial scope of the model... 5 4 Input data... 5 4.1 Population data... 6 4.2 Urban Morphology Types... 6 4.3 Land use dynamic... 6 4.4 Satellite Town Project and 20,000 Plots Project... 6 4.5 Road network... 6 4.6 Centrality... 7 5 Influencing factors... 7 6 Overlay and weighting of the influencing factors... 9 7 Modelling scenarios... 11 8 Results... 12 Glossary Influencing factors: Specific factors that have an influence on the transformation of cells into settlement cells based on previous urban dynamics studies and the local experts input. Transformability index: An index that is calculated using weighted overlay of several influencing factors raster files where cells could be then ranked based on this index. 2
1 Summary In the context of urban development in Dar es Salaam and its vulnerability and adaptation to climate change hazards, an Urban Spatial Scenario Design Model (USSDM) was developed. This model simulates different scenarios of spatial growth of settlement in Dar es Salaam according to projected population growth data. The model calculates a transformability index, based on specific influencing factors (input by user) and determines the cells that should be transformed in four iterations (2012, 2015, 2020, 2025) based on this transformability index. The number of cells being changed in each iteration and the excluded cells are defined by the user. 2 Model aim and usage The model aims to help visualising the spatial changes in Dar es Salaam due to settlement growth. The model can be used as a tool to understand and manage the different outcomes of distinctive urban development strategies on growth patterns (e.g. different population density, preventing flood prone areas from further settlement). It also helps to understand the interaction between different development strategies and climate change drivers, such as loss of green areas. 3 Model concept and scope 3.1 Population driven Model The basic driving force for the model is the land demand, which is influenced by the population development of Dar es Salaam (see Fig. 1). Population [in 1.000] 8.000 7.000 6.000 5.000 4.000 3.000 2.000 1.000 0 2025 2020 2015 2010 2005 2000 1995 1990 1985 1980 1975 1970 1965 1960 1955 1950 year population Fig. 1: Population development in Dar es Salaam, 1950 2025. Source: United Nations, Department of Economic and Social Affairs (http://esa.un.org/unpd/wup/unup/index.html). 3
3.2 GIS-based model The software that was used in this model is ArcGIS 10.0. The model is built in the Model builder environment of the software. Some operations in the model require the spatial analyst extension. To run and edit a model, you use the Model builder in ArcMap. In case you work with a ArcGIS version below 10.0, the model has to be saved in your version first. A detailed description and GIS manual of the model is given in a separate Technical user Guide document. 3.3 Cell/Raster based model The model is raster based. The whole study area is divided into cells of each 100 x 100m. The grid value of the cells can represent the state of the cell (settlement or non-settlement), the population density, and the influencing factor score (see section 6). The model calculates for each cell, how likely it is under the given assumptions that the cell will transform into settlement area in a certain period, indicated by the Transformability Index (TI). On this basis, the spatial growth of settlement is projected. 3.4 Conceptual model Fig. 2 shows a graphical representation of the conceptual model. Each iteration starts with the exclusion of areas (cells) that are considered to not transform into settlement areas. The next step is to calculate the transformability index based on the influencing factors; afterwards the cells are sorted accordingly. The required land demand is calculated in each iteration on basis of the population growth and the assumed population density. Next, the required number of cells is selected, starting with the highest transformability index value. At the end of each iteration, the influencing factor neighbourhood is updated as well as the population density in each cell. This process is repeated for each time step. 4
Fig. 2: Visual overview of the general structure of the model. 3.5 Temporal scope of the model Starting from 2008 on, spatial development scenarios were built for 2012, 2015, 2020 and 2025. Further time steps can be included; the time span can be extended by your GIS expert. 3.6 Spatial scope of the model With growing population, it is very likely that parts of the population will also settle outside the current administrative boundaries of Dar es Salaam. However, there was no detailed data (e.g. UMT map) of the greater area available. To still include this process into the model and to avoid an overloading of the administrative area, it was assumed that an estimated 10% of the additional population would settle outside the boundaries of Dar es Salaam, and was hence excluded from modelling the spatial allocation of land demands within Dar es Salaam. 4 Input data The outcome of the model and its accuracy depend strongly on the data input. Following data was available and included into the model: 5
4.1 Population data For the population growth rates and population density, data from the United Nations Department of Economic and Social Affairs UNDESA Population Division were used for the administrative area of Dar es Salaam (http://esa.un.org/unpd/wup/unup/index.html). 4.2 Urban Morphology Types As part of the task 2.7 in the CLUVA project, Urban Morphology Types (UMT) maps were prepared. The UMTs were selected when compared to land cover or land use maps as they combine both urban form and function (CLUVA D2.7). For the years 2002 and 2008 UMT maps were provided by Ardhi University, Technical University of Munich and the University of Manchester. UMT maps were used as input data for land use dynamic. Based on the UMT class it was classified whether an area is likely to be transformed into settlement area (= dynamic) or not (= stable). 4.3 Land use dynamic Based on the comparison analysis of the 2002 and 2008 UMT maps, the dynamism of each UMT class was classified (CLUVA D2.8). 4.4 Satellite Town Project and 20,000 Plots Project Data considering the Satellite Town project and the 20,000 Plots project was derived from the UN Habitat Citywide Action Plan for Upgrading Unplanned and Unserviced Settlements in Dar es Salaam (2010, available at the internet). 4.5 Road network The road network of Dar es Salaam was partially provided by the Ardhi University, partially derived from the UMT map, and partially extracted from road maps. The road network was classified based on importance into three classes. The classes are district roads, major roads (primary roads, primary links, secondary roads, tertiary roads, trunk roads, regional roads), and small roads (including all streets and roads that have lower hierarchy than the other two classes). The road network of Dar es Salaam is shown in Fig. 3. 6
Fig. 3: Road Network of Dar es Salaam.( Source: CLUVA 2012) 4.6 Centrality The sub-centres of Dar es Salaam are according to the Draft Master Plan of December 2012. Fig. 4 shows all centrality types included in the Master Plan. For the modelling process, only Local centralities and New metropolitan centralities were included. Fig. 4: Dar es Salaam Centrality (source: CLUVA 2012) 5 Influencing factors Not all areas in Dar es Salaam are probable to transform into settlement areas at the same level; different conditions make certain areas more attractive for people to settle in. For example, people prefer to live close to a road network, or in neighbourhood of other residential areas. In the model, this attractiveness was expressed by different influencing factors. The following influencing factors were included into the model: Land use dynamic: Based on the 2002 2008 comparison, the land use change of each UMT class was calculated. UMT classes that showed high changes (absolute 7
and/ or relative) in the past and/ or that played an important role in the process of settlement growth were scored higher than UMT classes with lower changes and/ or less importance. Centrality: Distance to planned areas of centrality of Dar es Salaam ( Local centralities, New metropolitan centralities ) according to the Draft Master Plan (December 2012). Road network: Distance to small roads, major roads and district roads. For each road type the distance was calculated and proximity scores were set. As the road types are of different attractiveness, the combined road network factor includes a hierarchy where more attractive roads were weighted more heavily: Small roads 16.7%, major roads 33.3%, district roads 50.0%. Although it can be assumed that the road network will develop dynamically together with settlement growth, the model did not take into account future changes of the road network due to missing data. Neighbourhood: The surrounding areas of existing settlement areas (>= 15 persons/ ha in the year 2002). The surrounding area is defined as a rectangle of 5 cells height and 5 cells width (500m x 500m). For each cell the sum of the surrounding cells (indicating whether it is an existing settlement cell or not) is calculated. (Note: The full scale of the neighbourhood score values is only visible when zoomed in, as the intermediate values between 0 and 100 only cover a small area at the edge to the settlement areas). Planning project satellite cities: Designated areas of the satellite cities project. Planning project 20,000 plots: Designated areas of the 20,000 plots project- Each factor has a score range from 0 100: 0 = least probable to transform to settlements, 100 = most probable to transform to settlements. Fig. 5 shows the influencing factors and their score range. 8
Fig. 5: The six influencing factors for urban growth in Dar es Salaam. 6 Overlay and weighting of the influencing factors To calculate the Transformability Index, the influencing factors are overlaid and the values of the influencing factor scores are summed up. The higher the sum of a cell the higher the cell s potential to be transformed into settlement area. Fig. 6 shows the summing up equation, Fig. 7 visualises the process graphically. Land use dynamic score + Sub centres proximity score + Road network proximity score + Neighbourhood score + Satellite cities project score + 20,000 plots project score = Transformability Index Fig. 6: Transformability index equation 9
Fig. 7: Overlay of influencing factors to get the transformability index Weighting of Influencing factors One of the features of the model is the possibility to give different weight to the influencing factors. This could be used to stress the importance of one (or more) influencing factors, or to exclude an influencing factor from the model by setting its weight factor to 0. Fig. 8 shows how the transformability index value changes, if the influencing factors Distance to road network and Distance to sub-centres are each weighted double. 10
Fig. 8: Overlay of influencing factors to get the transformability index with different weighting. 7 Modelling scenarios Based on different planning strategies, four different scenario types were considered. Two parameters of planning strategies were chosen: population densification and the inclusion/exclusion of highly flood prone areas from further settlement activities. By these two parameters the vulnerability of the population can be influenced considerably. The population density mainly influences the land demand for settlement areas, which is strongly correlated to the loss of areas providing ecosystem services. The inclusion/ exclusion of settlement activities in flood prone areas is also strongly related to the vulnerability of the population. In flood prone areas usually informal settlements develop, which have only limited adaptation capabilities. Four different scenario types were considered: Inclusion of high flood risk areas from further settlement Exclusion of high flood risk areas from further settlement Low densification (150 persons/ ha) For each scenario type, a different model is prepared. S1 S3 High densification (350 persons/ ha) S2 S4 11
8 Results Scenario S1: 2012 2025 Scenario S2: Scenario S3: Scenario S4: Fig. 9: Selected results: Population density in Background 2012 and 2025 information: according Urban to S1, growth S2, S3 scenario and S4. modelling for Dar es Salaam 12