A methodological approach to estimate changes on population density in a LUCC model: A case study
in Bogota, Colombia
Globally, urbanized land area is expanding faster than population. However, urban expansion is often
poorly planned and not well integrated with land use and transportation networks. Rapid urban
population growth could lead to sprawl that is poorly connected to urban activity centers, exacerbating
congestion forces (Kaw & Roberts, 2016). To enhance livability in cities and improve opportunities for
prosperity, planners, government policy makers, and stakeholders need to better manage the spatial city
structure and give access to land in a sustainable way.
Bogota is one of the most important metropolitan areas in South America, and is the most important
economic and industrial center of Colombia and the largest and most populous city in the country. It is a
dense and compact city with a rapidly growing population (Bocarejo et al., 2013). Urban and transport
planning must play a crucial role in developing a sustainable city, that seeks to supply the needs for
housing, land, and transport infrastructure. At this moment, the city is suffering land scarcity, particularly
on the urban core, which might induce urban sprawl. According to Buxton & Tieman (2005), the debate
over the costs and benefits of urban consolidation has perhaps never been more important given the rise
in the importance of sustainability principles as a basis of planning, the growth of mega cities and
development pressure on urban fringes.
In 2015, Bogota had 7.9 million inhabitants spread on an urban area of 365 km2, resulting in an average
population density close to 21,600 inhabitants per square kilometers. During the last two decades,
population migration has encouraged a fast and uncontrolled growth of the city´s boundaries, and illegal
settlements on the urban fringes became the most common way for developing low-income housing.
Regarding the urban built environment as a whole, Bogota is characterized as a low-rise, reduced-greenspace urban area. In average, the number of floors is around 2.05 and for predominantly residential
constructions it goes up to 2.3 (Bocarejo et al., 2014). Despite this low-rise structure, the average
population density remains high due to the poor average indoor space per capita (dwelling space) of 25
m2/inhab. This means that the population density in low-income zones is around 23,100 inhab/km2,
12,700 inhab/km2 in medium-income zones and 7,500 inhab/km2 in the wealthiest zones (Bocarejo et al.,
2014).
Bogota´s population growth has been significant for the past 20 years. It has moved from around 5 million
in 1990 to almost 7.2 million by 2010. Population projections by the National Administrative Department
of Statistics (DANE) estimates that Bogota will grow at an average rate of 1.1% per year in the next 25
years. The city is divided into 18 urban communities which are in turn divided into 112 planning zones.
These zones are named Unidades de Planeación Zonal (UPZ) and determine by the planning department
of Bogota. The purpose of the UPZ is to define and regulate urban land-use and management at a detailed
scale (Bocarejo et al., 2013). Today urban growth trends in Bogota show a concentration of low-income
households in the south and in the west of the city. The high-income households are located in the north
and in the center of the city (Bocarejo et al. 2013). Lack of available land to expand and high land prices
have expelled low-income population to peripheral zones of the city (Oviedo Hernandez & Dávila, 2016).
As available land within the Bogota’s jurisdiction runs out, the majority of the city´s urban growth is likely
to occur outside the municipality of Bogota. The Soacha municipality in the south of Bogota is notorious
for rapid growth of informal settlements on lower-priced land resulting from the expansion of Bogota
(Oviedo Hernandez & Dávila, 2016).
BoLD (Bogota Land Development model), is a LUCC model conducted for the city of Bogota, Colombia,
conceived to address the need to understand global impacts on the urban development of transport
infrastructure projects. The modelling was done using the Cellular Automata (CA) software Metronamica,
developed by the Research Institute for Knowledge Systems (RIKS). The model was calibrated towards
two recent complete dataset of land-use cover for the baseline year (2005) and the calibration year
(2014). Although the time lapse is just below what is recommended in the literature (9 years), the rapid
growth experienced in Bogota during those years has resulted in enough amount of LUCC as to properly
calibrate the model (Paez & Escobar, 2016). The model was then used for predicting the urban patterns
between 2014 and 2040, during four scenarios of transportation infrastructure – highway and train-based
development with restricted and unrestricted natural reserve.
In general terms, a LUCC model based on CA requires inputs in the following areas: 1) Future land
demands. 2) Current and future land zoning changes and suitability conditions. 3) Neighboring
relationships between land-uses, and 4) Accessibility analysis based on transport infrastructure (Paez &
Escobar, 2016). Regarding the estimation of future land demands for residential land-use, three major
assumptions were made: 1) High-income residential land-use will always correspond to 5% of the total
cell of residential land-use. 2) Poverty will decrease 5% in 9 years and the medium income will increase at
the same rate during that period. 3) Population densities remain constant over time (Paez & Escobar,
2016). The latter assumption represents a modeling limitation and an opportunity to enhance the BoLD
model. In addition, there is a potential for better land-use demand information, due to population growth
estimations can be improve using several techniques, such as system dynamics.
Based on the information mentioned above, the main objective of this research is to develop a
methodological approach to estimate changes on population residential density and incorporate them to
the residential future land demands. The fundamental hypothesis that supports the investigation is that
the results obtained with the actual BoLD model show a major urban sprawl that might not happen if
density is taken into account when future land demands are calculated. Urban growth is a management
problem all over the world. In most big cities planning or policies are unable to control growth (Lahti,
2008). This leads to uncontrolled urban sprawl, characterized by a low-density expansion. The causes of
sprawl are typically connected to growth in population and economy, but also preferences in living,
changes in demography, price of land and cultural traditions (Batty, et al., 1999).
The essential motivation of this research is to develop a methodological approach for the estimation of
future residential land demands incorporating population density changes. The first step, is a regression
model using econometric software STATA to find out statistically significant drivers or variables that allow
to explain density patterns over Bogota’s urban territory. For the purpose of this study, each of the 112
UPZ were adopted as the unit of observation. In order to achieve this, a vast data-base was compiled using
more than 50 variables including mobility, transport, cadastral and urban parameters such as; number of
trips generated and attracted, number of urban amenities, accessibility, floor area ratio, land and property
values, average travel times, among other variables. Once a well-specified model is obtained in STATA a
GWR is run in ArcGIS using the same variables in order to contemplate spatial relationships and a local
model for each UPZ. Combining the density models segregated by income and the population growth
model using systems dynamics software Vensim, future residential land demands are calculated for BoLD
model.
A regression model was used to identify and quantify the drivers of population density for high, middle,
and low income residential segments and a model using systems dynamic techniques was developed to
generate a demographic growth model. Once density and population growth is determined for each
income segment, it is possible to calculate the future residential land demands for BoLD model. The
regression models segregated by income show the influence of urban and mobility variables over density
patterns over Bogota’s urban territory.
Taking into account density changes in future land demands is a vital step to enhance BoLD model in order
to make CA modelling more robust and permit decision and policy makers address in an accurate way
changes in urban patterns and land-use. These regression model and population growth model using
systems dynamics techniques may allow urban planners and land developers to test several hypotheses,
such as if density increases due to economies agglomeration, improving access to services, and generating
affordable housing.
Future research is suggested in the notion of the compact city, which is often referred as the city of short
distance, contrasts the car-oriented urban sprawl of many modern cities and can be characterized by
combining efficient, multifunctional and multi-modal transport systems whilst fostering relatively high
population densities (De Vries & et al., 2016). This city morphology is desirable, due to urban sprawl force
a high demand for a limited supply of land which at the same time compels to informal land development
(Masum & et al, 2016). Besides, inadequate housing in informal settlements (or slums) is known to be an
obstacle to economic growth in cities (Weldesilassie & et al., 2016).
References
Batty, M., Xie, Y., & Sun, Z. (1999). Modeling urban dynamics through GIS-based cellular automata.
Computers, Environment and Urban Systems, 23(3), 205–233. http://doi.org/10.1016/S01989715(99)00015-0
Bocarejo, J. P., Guzman, L. A., Portilla, I., Melendez, D., Gomez, A. M., & Rivera, C. I. (2014). The Access
as determinant variable on residential location choice of low-income households in Bogota, 0(July
2016), 1–23.
Bocarejo, J. P., Portilla, I., & Pérez, M. A. (2013). Impact of Transmilenio on density, land use, and land
value in Bogotá. Research in Transportation Economics, 40(1), 78–86.
http://doi.org/10.1016/j.retrec.2012.06.030
De Vries, W., & et al. (2016). SCALING UP COMPACT CITY DEVELOPMENT STRATEGIES – A SOLUTION TO
COPE WITH URBAN SPRAWL IN DEVELOPING COUNTRIES? . Land and Poverty Conference 2016:
Scaling up Responsible Land Governance. Washington: The World Bank.
Lahti, J. (2008). Modelling Urban Growth Using Cellular Automata: A case study of Sydney, Australia.
Masum, F., & et al. (2016). THE LIMITATIONS OF FORMAL LAND DELIVERY SYSTEM: NEED FOR A
PROPOOR URBAN LAND DEVELOPMENT POLICY IN DHAKA, BANGLADESH. Land and Poverty
Conference 2016: Scaling up Responsible Land Governance. Washington: The World Bank.
Oviedo Hernandez, D., & Dávila, J. D. (2016). Transport, urban development and the peripheral poor in
Colombia — Placing splintering urbanism in the context of transport networks. Journal of Transport
Geography, 51, 180–192. http://doi.org/10.1016/j.jtrangeo.2016.01.003
Paez, D., & Escobar, F. (2016). 12 Urban transportation scenarios in a LUCC model : a case study in
Bogota Colombia, 1–41.
Weldesilassie, A., Gebrehiwot, B., & Franklin, S. (2016). MANAGING URBAN LAND FOR LOW COST
HOUSING FOR AFRICA’S CITIES: THE IMPACT OF THE GOVERNMENT CONDOMINIUM SCHEME IN
ETHIOPIA. Land and Poverty Conference 2016: Scaling up Responsible Land Governance.
Washington: The World Bank.