Adapting to the consequences of climate change in urbanised regions are of increasing importance in most parts of the world. This PhD study contributes with methods and knowledge regarding how planning of urban areas can be used as a tool to reduce impacts from transportation. The research focus on the development of land use and transportation models, that will be used to illustrate how different paths of future urban development can reduce transportation demands and thereby the related impacts on the climate and environment.
The demands for transportation are generated from a many different human activities. Modelling where people locate themselves in urban areas in relation to the placement of their opportunities (works, shopping etc.), is important part of describing how the impacts from transportation will evolve. The work conducted focus on three future pathways from the PASHMINA project, where different future development strategies are proposed. The methodologies applied vary from different GIS analysis, to land use change simulations conducted with in the ‘Land Use Change Impact Analysis’ (LUCIA) Cellular automata model, highlighting the consequences of future planning strategies
The results from this PhD study shows of how spatial planning can utilize advanced GIS techniques to gain valuable insight into the outcomes of different planning strategies. The results illustrate the potential improvements on sustainability that could be achieved, if planning practice where to change from the current patterns. Amongst others, factors such as urban proximity, public transportation accessibility and job proximity have been evaluated in the work with the scenarios. Furthermore well as methodologies for traffic impact analysis and population density measures that have been suggested, all demonstrating that the potential gains from rethinking the urban landscape through planning strategy.
First i will shortly present the two projects that financed this phd study, to present how the tasks from the two projects shaped the study. Second i will present the main hypothesis ond objectives from the study, to frame the work and results that i am going to present. Third and fourt i will present the methodologies that i have used, and the results obtained from the different tasks within the study. Finally fitht, i will present the main conclusions of the study.
Adapting to the concequences of climate change is of increasing importance in the urbanized region
This PhD study contributes with methods and knowledge regarding how planning of urban areas can be used as a tool to reduce negative concequences for the environment from land use change
The aim of this study was to show how spatial planning can utilize advanced GIS techniques to gain valuable insight into the outcomes of different planning strategies
The results illustrate the potential improvements on sustainability that could be achieved, if planning practice where to change from the current patterns.
My Phd was constructed as a combination of two EU projects:
From AU Pashmina project From AAU the BLAST project
In practice i was part time at both universities.
From Pashmina a series of scenarios for future development was created – which are the centrepoint of this study. From BLAST the traffic prediction mapping was required. Therefor the task was combined with the scenarios, to analyse traffic development from the scenarios.
The project was organised in three phases:
Configuration and calibration of the LUCIA model to create the land use predictions. (main part) Creation of supplemnatary models and tools to be used to analyse the scenario results. Finally the scenarios are analysed for sustainability based on both the core LUCIA results, as well as the supplementary tools.
The main hypothesis that was the overall framework of the work conducted was : Spatial analysis and modelling of land-use change can be applied to strengthen spatial planning by creating insight into the expected outcome of plans and thereby contribute to sustainable development Focus will therefor be on the knowledge that can be gained from applying spatial modelling techniques in planning. The outcome of this application, should be able to analyse the results in relation to sustainability, so that the planning can be conducted to ensure sustainable development.
Under the main hypothesis, a series of research questions was formed.
Analyse drivers, and calibrate the LUCIA model Translate the scenarios from the PASHMINA project into scenarios in the model
Produce result maps highlight the concequences by analysing the results, using different measures of sustainability.
In the PASHMINA project, four scenarios was proposed, based on an analysis of scenarios from other projects, and a survey amongst different experts.
The four scenarios was the pear, the apple, the orange and the potatoe.
Pear… Apple… Orange…
The potatoe scenario was in many of the modelling tasks in the project discarded, since it was agreed, that the models would not be able to make meaningfull predictions in this collaps situation.
These three three scenarios was the backbone of the work conducted in this study. For all of the scenarios, a comprehensive narrative was written, that describe the world as it would evolve in the future under their circumstances.
For the study, a case region comprising of Sealand and Lolland falster was selected - This was the case region that was written into the PASHMINA proposal, so i had limited influence on this choice.
The case region comprise of the metropolitan area of copenhagen, and the hinterland.
The case region are on an overall level expected to have a population increase in the comming decades, however with great variations on municipality level.
In order to make a toolbox of spatial data, a study of litterature was conducted, analysing the most commenly applied. These data would potentially be used to impliment the scenarios. Based on this study, a list of potential datasets was created, and based on the availability of data, the datasets was then created. The list of datasets that was created include measures of urban density, proximity to motorway junctions, urban areas, stations etc., accessibility measures, and a property value indicators.
To illustrate public transportation accessibility, a raster based calculation was proposed.
Uses the networks from service providers, with information of travel speed.
The concept was, that in vector based model, the netowork can be accessed at all points – our network only on stations.
Therefor a system was created, that applied constrints to the network, so that the stations was only entry to the network.
Based on a cost path analysis for centres, the traveltimes are combined, highlighting where we have high and low accessibility.
Indicator based on the ‘acceptable commuting time’ concept – how longe are peopple willing to travel to work. A centre with few jobs have smaller attraction than major centres. Based on data from DST 175 centres was classified with an time attribute. Using service area calculations in a network anaylsis, the interland for all centres was established. The results from each centres was assigned the value 1, converted to raster and summarized. The cell value depictes the number of centres reacheable from each cell within acceptable commuting time.
The datasets was analysed for sensitivity in the GIS analysis – testing parameter values – for the car asseccibility for instance, different values was assigned to the centres to test the output result, and se how the results vary.
The indicators was also tested against observed land use changes from 1990 – 2002., to analyse if there are a general trend in the placement of urban areas in relation to the data values. If the observed urban development was spread evenly over the data, then the dataset have limited use. If hovever there are a clear trend in the placement, the dataset can be used in a prediction of new urban areas.
Finally alle the indicators was analysed statistically to highlight the differneces in mean, std. Deviations etc.
Here we have the main elements from the narrative that was used as the foundation for the scenarios. Since the scenarios originally had a main focus on economy and development, only few elements could be realted to planning.
GWOL(Pear): New reserves of fossil fules – continued car dependent sprawl – increased land consumption GWIL(Apple): network of transit oriented centres, infrastructure investments, improved efficiency of viechles BG(orange): compact cities, travelsmart, slow town, reduced travel needs, reducing work time
These are the main statements from the narratives that was selected to form the basis of the work with the scenarios in the case region
Here are the overview on the implimentation of the interpretations of the scenarios.
Pear – no zones, road accessibility, proximity low Apple – developement on the fingerplan concepts, public transportation accessibility, proximmity is increased Orange – recreational accessibility, urban density and proximity is the key parameters.
These setups where then to be incorporated into the LUCIA model – which i will just shortly present.
The LUCIA model is a so called cellular automata model. Cellular automata is a mathematical methodology for cell based representation of data, where each individual cell is evaluated based on its value, and according to its neighbours.
The LUCIA models basic operations can be seen on this figure (go through it)
The factors included in the model are used in a multicriteria expression, assigning weights to factors in the calculation. Based on this calculation, the model predicts the most suitable candidates for land use change, and assignes them to urban in t+1
The model is then capable of producing future land use maps – year by year, determining how the urban landscape is going to develop based on the input data..
In the work with the scenarios, some basis assumtions was made, in order to frame the work with the scenarios:
No modifications made to the spatial divisions of admin boundrys. The prediction of population development are assumed to be able to handle all urban/rural movement, population decline etc. Therefor there is no need to modify units (Cph/rural) or relocate population No landscape protection measures are inforced – we urbanize forests, urban green space etc. The prediction of land use demand from LUCIA will be the guidline for allocation.
In this figure we see the configuration of drivers and factor weights assigned to each scenario in LUCIA. As it can be seen for instance the usage of the proximity and suiteability factor is quite different between the scenarios.
The way that these scenarios was constructed, was in itterative process. First a number of datasets was selected, that was belived to be able to construct the development from the narrative. Secondly the scenarios was modelled using one factor at a time, with different values, and thereafter visually analyzing the results. If a driver had an influence that seemed to strong, then the setups was modified and re-run. The scope was to create a plausible result for each scenario, that would still have some degree of fredom for development, but still to high degree honour the narrative.
In order to make sure that we can trust our predictions, the model must be valiadet. In this way, it is ensured, that the model can recreate the observed reality before we can use it to create predictions.
A periode from 1990 to 2002 with observed data was used as validation period.
First, the Kappa measures was used to compare the results 2002 map with the observed map.
Secondly, the results was tested for fragmentation pattern. The result should be able to replicate the fragemented pattern from the observation period.
Here we have the results from each of the scenarios.
(1-3)It might be difficult to see the differnces in the maps, but if we look at Slagelse, we can se that there is in fact some major differences in the results. (4) If we combine the three results op, the overall pattern can be seen.
The scenarios was first analysed in relation to the land use conversions that they predict. The results was compared to the CORINE 2006 map from EEA, for the landscape classifications. I have highlighted a few elements here, which are First the green urban area conversion which is quite high in the the orange scenario – more than double the apple. Secondly it can be seen, that the apple scenario occupies more land from the category principally occupied by agriculture. Finally it can be seen, that the apple scenario has the highest single class conversion of existing forrest, but combined, the orange scenario covers the most forrest.
Looking at the other results, it can be seen, that the orange scenarios alternative placements in relation to recreation can be seen in some of the other classes, occupying grassland and shrub
Then the results from the scenarios was tested against a series of measures: First the actual scenario output was analysed And second the scenarios results was used as input to other model tools, to create extended results for the scenario predictions.
For the three scenarios, the fragementation analysis was conducted for the 2040 result maps, analysing the areas that was urbanized It can be seen, that both the apple and orange scenario have a great effect on reducing fragmentation, making the urban landscape less scattered. It is commonly agreed, that if the the urban landscape is more compact and less scattered, the transportation work will be reduced – which benefits the environment.
We dont know how the public trasnportation landscape is going to develope towards 2040, so it is hard to predict how the accessibility is going to develop Therefor we can only conduct analysis of this parameter against the current accessibility
It can be seen, how the effect of the fingerplan and the closeness to stations priciples that was used in the creation of the apple scenario have a possitive effect on the potential accessibility to public transportation – which ofcourse was expected due to the driver configuration.
The Orange scenario which uses the recreational indicator as one of the main drivers produce a poor result here, since the public transportation was not in focus, due to the nature of the scenarios dictating new forms of work and travel habbits.
If the public transportation accessbility is analysed for each municipality, the improvements from the pear to the apple scenario can be highlighted.4
It can be seen, that in especially the southern and wester parts of the region, inrpovements can be achived, in creating an urban surface with closer linking to the public transportation network.
Comparing the three scenarios to the job proximity indicators, it can be seen, how the two alternative scenarioes perform in locating urban development in areas which has a better local connection to jobs. By locating the development in areas with higher job densities should theoretically result in lower transportation amounts
It can be seen, that the increas in job proximity is quite good for both the alternative scenarios, both showing an increase above 10% Therefor on this parameter as well, the two alternative scenarios would theoretically produce positive effects in relation to travel to and from work
A series of further modelling tasks was conducted, where three was included in the thesis.
Based on the scenarios a series of hotspot maps was calculated As addons to LUCIA two models was sugessted, calculating traffic and population density
For the hotspot predictions the three rasters was combined into one, with the cell values idicating how many scenarios they where urbanized in (0-3)
Then a hotspot analysis was conducted with different scale, to highlight where the urbanization is going to happen across the scenarios. This identifyes the areas where high degree of urbanization is going to happen – no matter what the planning focus is going to be.
The first proposed LUCIA addon was a population distribution measure.
Based on a dassymetric mapping approach where the population density in the different land use classes in 2006 are extrapolated to 2004 population maps are created for the three scenarios.
The model uses regional estimation of weights, menaning that it is determined for each municiplaity – as the startpoint for the analysis. However due to time constriaints, the model was never coded to be self balancing, to produce the best population prediction based on the 2040 population, but was only based on the weights determined initially.
The results here from køge shows how the pear scenrio comapres to the apple and orange. It can be seen, that the distribution of population changes alot between the scenarios, which directly could be used for estimating capacity planning.
The second proposed addon was a traffic prediction addon, designed for the BLAST project
Here a raster based model was proposed, that based on the celles location in space (in relation to its nearby centres), estimates a traffic spread from the cell. Each centre is wighted in the calculation, to determine which centres the new cell is in relation to.
Then based on the statistics from the church district and municipality that the cell is located in, an equation estimates the traffic generation from the cell, both in terms of magnitude and coverage.
The results from the model shows, how the traffic development is going to evolve year by year based on the outputs from LUCIA.
Here examples from Køge shows how differently the traffic development is going to evolve in the pear and apple scenario.
It can be seen, that the traffic is much more spread out in the pear scenarios, where as it is more concentrated in the apple.
For three case municipalites, these are the overall findings with the addon model.
In each case, the pear scenario is considered basline at index 100, and the results show percent changes.
The apple scenario produce reductions for all three case municiaplites, both in therms of area and amount. The orange scenario are difficult to analyse with this model – since its location of urban areas in this analysis in many of the cases will perform wors that the pear scenario
This is due to the fact, that the model was constructed with urban closenes and distance to centres as perameters – which the orange scenarios does not honour.
Based on the criteria for implementation and validation that can be found in the literature, the model has been successfully validated for the observation period. The goal was not a pixel by pixel exact replication of the observed changes, but more a pattern that resembles in terms of fragmentation, and overall urban placement.
The main spatial drivers of the observed land-use change in the case region are a combination of accessibility to urban centres and main roads, combined with property value. These where the drivers that was used to create the validation of the baseline – meaning that with this set of drivers, the urban development can be discribed as it was observed from 1990 - 2002
The scenarios was created in subjective process – changes in the drivers affect the results. The focus was primarily on the story that was told by the scenarios, and the interpretations of these.
It was dissifult to find examples of a scientific method for interpreting the scenarios into the model – wherefor the observation based approach was selected.
Focus on protection and preservation might have been included – runs with more strict factors would perhaps have been usefull in the concequence analysis.
The results show how different the planning paths are by creating very different urban landscapes. This produces significant new knowledge about the results of adjustments in planning practice, which can be used to analyse consequences of urban distribution under different planning strategies.
It has been shown that the impacts from the alternative planning paths can provide positive results in relation to urban sustainability, by creatiing denser urban surfaces, combatting sprawl, and thereby reducing the dependency of car transportation by enforcing these patterns
Hard to finde studys to compare the sustainablity analysis to As with this one, other studys are based on al long range of assumtions and case specific parameters, that there was little that could actually be compared.
Work with the population and traffic prediction addons could and should be continued – but both models demonstrates the potential in the application, which was the scope here. It can produce important knowledge to consider traffic generated and population distribution in the planning process, which is what have been demonstrated here.
With the increase in computational power and advances in spatial data and models, municipality planners will be able to use scientific models
The knowledge requirements may not be met presently, but with strong candidates educating from the universities, the skill levels of the GIS departments are rising.
Therefor models such as the LUCIA CA model will definitely become possible tools for the planners, to create visualisations of their plans to ensure as high a degree of sustainable development as possible.
Supporting sustainable urban development
through spatial planning by cell-based modelling
Ph.d. defence by Morten Fuglsang, April 29 2014
1. Brief introduction
2. The projects financing the phd study
3. The hypothesis and objectives
• Adapting towards more sustainable
development in urbanised regions are of
• The aim of this study was to show how spatial
planning can utilize advanced GIS techniques
to gain valuable insight into the outcomes of
different planning strategies
• The project was financed by two EU projects:
1. The PASHMINA project (Paradigm Shifts And
2. The BLAST project (Bringing Land And Sea
Overall structure of the work
• Three phases of the
1. The modelling of the
scenarios towards 2040
2. Development of
3. Analysis of sustainability
The main hypothesis
• Spatial analysis and modelling of land-use
change can be applied to strengthen spatial
planning by creating insight into the expected
outcome of plans and thereby contribute to
• Analysing and describing drivers of land-
use change in the case region, aiming at
calibration and validation of the land-use
simulations with the LUCIA model.
• Analysing how the narratives from the
PASHMINA project can be translated into
land-use change projections expressing
the structural changes from the
• Producing a set of spatial maps predicting the land-
use change by 2040 under the drivers described in
the scenario narratives.
• Highlighting the consequences of the PASHMINA
scenarios in terms of sustainability through the
development of GIS analysis and discussing the
outcome in relation to spatial planning.
• For the study, the
region of Zealand and
Lolland Falster was
used as case region
• Based on a literature study, the
common datasets were analyzed.
• A series of these datasets were
created to potentially being used in
the model including:
1. Urban density
Public transportation accessibility
• Based on a ‘cell-
• Added constraints to
the network to use
Car accessibility indicator
• Uses a classification
of 175 centers
• Using service areas,
the hinterland of
each center was
Analysis of indicators
• For each indicator, an analysis of sensitivity
was conducted on key parameters
• Furthermore, the distance relationship to
urban development was analyzed
The LUCIA model
• Calculates land use
demand from factors
• Uses land use maps and
development as main
• Use existing administrative
• Population estimates from DST are
depicting shifts from rural to urban
• No landscape protection measures
• LUCIA’s prediction of land use
requirements are adequate
• Based on the CORINE classification, the results
were analyzed for land-use conversions.
Analysis of scenario
• The scenarios were tested against a
series of sustainability-measures
• Furthermore, the scenario-results
were used as inputs to other
models to further access the
implications of the proposed
• The alternative scenarios produce a
more compact urban structure
The effect of the public transportation
accessibility used in the apple
scenario is evident
Both alternative scenarios locate
urban development closer to the jobs
Three further analysis tasks were
• A hotspot prediction mapping
• Two add-ons for the LUCIA model –
one for traffic and one for
population were suggested
• Based on a combination
of the scenario results,
hotspots were calculated
• Highlighted areas that
are likely to urbanize no
matter what planning
practices are used
• Using a dasymetric mapping
approach, a model was
• Uses regional parameters
for distribution functions
• Not, however, a self-
Traffic prediction model
• A raster based traffic
distribution model was
• Uses location in space
combined with statistics to
predict traffic consequences
of new urban developments
• The model produces result maps showing how
the traffic load is going to evolve
• The Apple scenario shows improvements in
traffic generated for all three case areas
• Used to analyze the scenarios, it is clear that
the model have difficulties with the orange
• The LUCIA model has been satisfactory
validated for the observation period
• The main spatial drivers of the observed land-
use change in the case region are a
combination of accessibility to urban centres
and main roads, combined with property
• The scenarios were created in a
subjective process – changes in the
drivers affect the results
• Focus on protection and
preservation might have been
Conclusions - continued
• The results show how different the planning
paths are from creations of very different
• It has been shown that the impacts from the
alternative planning paths can provide positive
results in relation to urban sustainability
• Hard to find studies to compare the
sustainability analysis to
• Work with the population and
traffic prediction add-ons could and
should be continued
Conclusions - continued
• With the increase in computational power and
advances in spatial data and models, municipality
planners will be able to use scientific models, in order
to predict the outcome of their plans.
• The knowledge requirements may not be met
presently, but with strong candidates educating from
the universities, the skill levels of the GIS
departments are rising
Conclusions - final
• Therefore, models such as the LUCIA CA model
will definitely become possible tools for the
planners to create visualisations of their plans
to ensure as high a degree of sustainable
development as possible for the future…