Monitoring and Modelling of Informal Settlements – a Review on Recent Developments and Challenges Peter Hofmann
Hannes Taubenböck
Christian Werthmann
Department of Geoinformatics University of Salzburg A-5020 Salzburg, Austria
[email protected]
German Aerospace Center (DLR) German Remote Sensing Data Center (DFD) Oberpfaffenhofen, Germany
[email protected]
Institute of Landscape Architecture University of Hannover D-30419 Hannover, Germany
[email protected]
Thus, adequate methods and tools to detect informal settlements and to estimate their future developments are of importance. At this background we investigate recent methods for monitoring and modelling informal settlements and the methods’ potential to foster the understanding of slum genesis and development in general and at slum-individual level.
Abstract— The increasing development of informal settlements and slums in the years to come is a humanitarian challenge we are facing worldwide. The reasons for this development are mainly given by an urbanization of poverty and the lack of affordable housing in urban areas. Consequently, the majority of slum dwellers are forced to live in housing conditions which are simply unacceptable. In order to fight urban poverty and develop strategies to reduce the number of slum dwellers it is essential to inventory and monitor slums and to understand the underlying mechanisms of slum genesis. However, systematic spatial knowledge about these mechanisms is still absent. This paper addresses the capabilities of mapping and structural analysis of informal settlements using remote sensing data and modelling their genesis based on agent based approaches.
I.
II.
DETECTING AND MONITORING INFORMAL SETTLEMENTS
In order to detect and monitor informal settlements, different and mutually non-exclusive methodological ways of information gathering can be applied, ranging from volunteered mapping by the settlements inhabitants [4], to the analysis of remote sensing data. A. Detecting informal settlements in remote sensing data Due to the lack of a definite definition of slums or a general theory on such locations [5] it is hard to describe general mapping rules for slums. Additionally, some informal settlements can be hardly distinguished from formal settlements just by their physiognomy. For these cases rather in-depth knowledge about the individual history and actual situation is indispensable [6]. This lack of a definition or theory paired with the high variability in appearance across different contexts [7] turns the (semi-) automatic spatial detection and delineation of slums in remote sensing data to a complex problem. In 2001 Hofmann [8] stated that relatively little systematic research exists about the physical appearance of slums. Although this statement has been made more than a decade ago global spatial knowledge on the location, dimension, structure and morphology of informal settlements or slums is still absent. In 2008 an expert group meeting took place [9] which led to the preliminary notion of qualitative and quantitative indicators for slum identification from VHR remote sensing data (Tab. 1). However, since that time, in remote sensing various earth observation data has been tested to increase spatial knowledge on slums. Kohli et al. [7] developed a general ontology of slums with respect to image based classification. In any case, VHR optical data have been widely applied to detect and analyze slums. Few studies even used radar data to detect informal settlements [10]. Conceptually, most studies address the block level aiming to
INTRODUCTION
Per UN-Habitat definition, a slum dweller fulfills at least one of five conditions: lack of tenure, lack of access to safe water, lack of access to improved sanitation non-durable housing and overcrowding. The term “informal settlements” is used for urbanization originating outside of municipal planning efforts (if they exist). Thus, informal settlements are not synonymous with “slums”, but most of them fulfill at least one of the UN Habitat’s conditions for slums. Recently, more than 1 billion people are living in slums worldwide [1]. Estimates expect this number to rise up to 1.5 billion in 2020 [2] and to three billion in 2050 [3]. Although informal settlements are a world-wide phenomenon, mainly observable in less developed or developing countries, yet little is known about the mechanisms of their genesis and development. In many cases they are the result of a continued rural-urban migration and a lack of affordable formal housing offers. Often, informal settlements occur on marginal land of urban regions as a result of informal urbanization. The process of informal urbanization ranges from large scale, organized land occupation by undercover developers to small scale, sporadic land occupation by individual families. Consequently the morphology of informal settlements ranges from regular, gridded layouts to irregular, organic layouts. If not removed, the development of informal settlements can have very different outcomes. Respectively, informal settlements can be very different in their dynamic of physical shape, which aggravates their detection for inventory or planning issues.
1
TABLE I.
not yet allow for an automatic spatial separation of individual structures with sufficient accuracy. Thus, studies on structural analysis at individual building level [19] are still rare (Fig. 2).
INDICATORS FOR DETECTING INFORMAL SETTLEMENTS IN VHR REMOTE SENSING DATA (SOURCE: [9] P. 6-7)
Qualitative indicator Haphazard, high density building pattern Durability of housing/material for roofing (corrugated tin, plastic sheeting/tarps, cloth/grass) Proximity to natural and technological hazards Structure, size and condition of road network
Quantitative indicators Proportions of land cover (% area covered by structures, % area of open spaces, % of vegetation) Spatial metrics (lacunarity, textural contrast, variability of building size & orientation) Morphology of building heights (derived from shadows or lidar) Characteristic scale (relative size of housing units, relative size of road networks, plot ratio, floor area ratio)
Proximity to public services (health, education, open space, public transport) Fig. 2: 3-D city model derived from QuickBird data and field work from Bharat Nagar in Mumbai, India for differentiation of formal and informal settlement structues (Source: [19]).
delineate informal settlements within the urban fabric. In this context, Taubenböck and Kraff [11] exemplarily proved that morphological and structural characteristics of informal settlements can show significant differences to the surrounding formal urban areas. Object-based algorithms [12] (Fig. 1), snakes [13], radial casting algorithms [14] as well as visual classifications [15] have been applied to take advantage of the morphological difference of slum appearance in remote sensing data. Nevertheless, the variety of spatial features possibly classifying informal settlements is large [16]. Other authors use direct spatial features derived from individual buildings such as size, shape or height together with their density and orientation [17], [18]. At the level of individual buildings, approaches are mostly restricted to visual interpretations as the high density of buildings, spatial overlapping of roofs and their heterogeneous composition do
B. Monitoring informal settlements in remote sensing data Although many studies demonstrate successful identification of informal settlement structures and slums, only few studies have addressed the multi-temporal evolution of slums [20], [21], [22], [23], all of them more or less mapping the general change of land use and land cover but not explicitly the areas covered by informal settlement structures. Recently, Kit and Lüdeke [24] used multi-temporal VHR data in order to automatically detect the change of slum areas in Hyderabad. Gruebner et al. [25] used visual interpretation of VHR data to map spatial extents and evolution of slum areas in Dhaka from 2006 to 2010. With it they allow monitoring demolition and new developments of informal areas. III.
MODELLING AND SIMULATING INFORMAL SETTLEMENTS
While detecting and monitoring informal settlements aims at describing and delineating their physiognomy, location and extent (over a past period of time), the modelling and simulation of informal settlements is rather explanatory and predictive. Several methods for spatial modelling and simulating are available. All of them are based on mathematical models formalizing the underlying development hypothesis of slum and informal settlement development. A. Available methods Simulation and modelling is very popular in landscape ecology, in social and economic sciences, in political sciences or in urban studies. The mathematical methods used for process description are manifold, ranging from Genetic Algorithms (GA) [26], via Markov-Random-Fields (MRF) [27], Regression Models (RM) [28], Multi Criteria Evaluation (MCE) [29], Artificial Neural Networks (ANN) [30] to Cellular Automata (CA) [31], Agent-Based Modelling (ABM) [32] and Multiple Point Statistics (MPS) [33]. Combinations of these methods are possible. For modelling and simulating informal settlements, recently ABM and CA are very widespread. Since both methods are close relatives, CA and ABM are very often used and described synonymously. In
Fig. 1: Segmentation and classification results for an IKONOS scene in Cape Town (top) and a QuickBird scene in Rio de Janeiro (bottom). Left: top segmentation level with respective classes. Right: base segmentation level with respective classes. Blue rectangular in the left images indicates the location of the right images (Source: [18]).
2
Recently, we can just hypothesize these mechanisms and create appropriate simulation models, which we than compare with the observable reality. Therefore, the better the surveyed information the better the models can be evaluated and calibrated, which in turn improves our understanding of the mechanisms. In this context, detection and modelling must be capable to operate in 3D and intrinsically consider time to enlarge information and knowledge on environmental indicators such as their location and evolution in the larger urban context. Within the slums, indicators such as the building density, the buildings appearance and orientation in 2D and 3D, the layout of street networks and the terrain situation must be detected and considered. Accordingly, the spatial and temporal resolution of data and models must be appropriate. Methods for information extraction from remote sensing data must allow analyzing VHR remote sensing data together with accurate elevation data (Fig. 2) at multiple dates and scales. Although highly desirable, identifying general rules of genesis for informal settlements is recently unfeasible. However, international comparative studies of informal settlements and their histories could potentially uncover mechanisms yet unknown. But comparing informal settlements and slums means to identify and develop comparable features and provide standardized methods of comparison.
contrast, some authors [34] clearly distinguish between CA and ABM. CA are organized as regular 2D grids wherein each grid cell follows pre-defined rules of state transition. Thus, spatially each cell of a CA is static, but during simulation it changes its state depending on the underlying transition rules. Eventually, the agglomerate of all CA in a simulation produces space-temporal patterns which are more or less similar with those observable in the real world. In ABM agents are not necessarily bound to a specific position [34]. By definition [35] each agent is capable to sense the environment it is acting in - including other agents - and intends to achieve its pre-defined goals. All agents are autonomous in their decisions how to achieve their goals. Thus, they can behave proactive and collaborative and can have different roles within the ABM. Individual agents can emerge to higher level aggregates which can behave dynamic, too [36]. However, CA and ABM are not necessarily mutually exclusive. Systems such as VecGCA [37] or hybrid models [34] allow polygons in GIS to behave like CA and thus to change their shape, that is, acting like dynamic patches [38]. Hofmann et al. [39] made a first attempt to conceptually marry image segments with the ABM paradigm. B. ABM and CA to simulate informal settlements Roy et al. [40] recently compiled a review on contemporary slum simulation models. All of them are either agent-based or based on CA but follow particularly different intentions. The review roughly analyzes six slum simulation systems (ISGM [41], Peripherization [42], ISGPM [43], MSGM [44], Slumulation [45] and Economic disparity [46]) comparing their socio-economic and environmental aspects, their spatial and temporal resolution, their level of abstraction and the procedures to evaluate and calibrate them. MSGM intends to simulate the inner physical settlement structure while Economic disparity and ISGM focus on slum boundaries and the socio-economic structure. Slumulation is rather focusing on the location of potential slums within Hyderabad. Aside the reviewed models, Crooks et al [32] developed an ABM to investigate segregation patterns of informal settlements under varying environmental conditions and under varying behavior rules. Interestingly, none of the models takes into account information gathered from remote sensing data or considers the 3rd dimension. IV.
V.
CONCLUSIONS
Recent methods of detecting slums and informal settlements are mainly focusing on detecting the status quo. Methods based on remote sensing data predominantly intend to delineate informal settlement structures from other urban fabric. For this purpose expert knowledge in image analysis and about principle mechanisms of the settlement-forming mechanisms is necessary. Multi-temporal remote sensing data recently is just used to document changes of land cover in urban areas but it has no impact on the modelling in terms of calibrating the models. 3D information is neither used for object detection within slums nor for modelling and simulation. However, in order to gain a deeper understanding of the mechanisms of informal settlement genesis, it is necessary to monitor individual informal settlements and their structures in-situ and ex-post and in all spatial dimensions. With this knowledge at hand, models simulating the development of informal settlements can be developed and improved, which in turn can improve the methods for detecting and delineating informal settlements from remote sensing data and foster our understanding of their genesis. Thus, from our point of view future developments should allow to more closely link methods of remote sensing data analysis with those of modeling and simulation and consider time and space in all their facets.
CHALLENGES FOR MONITORING AND MODELLING INFORMAL SETTLEMENTS
Whatever instruments and actions are applied to improve the housing situations of slum dwellers, appropriate situation awareness of the acting protagonists is necessary, which includes appropriate geo-information. Since slums and informal settlements are unplanned and dynamic by nature, only methods allowing an in-situ and sometimes even an invivo (e.g. volunteered mapping) information retrieval are appropriate. For estimating future developments CA and ABM are certainly appropriate and state-of-the-art. However, the creation of plausible scenarios of informal settlement and slum development is only feasible if the models are fed with profound knowledge about the genesis steering mechanisms.
References [1] UN. Habitat, "State of the World’s Cities 2010/2011: Bridging the Urban Divide. March 18, 2010," UN Human Settlements Programme: Nairobi, Kenya, 2008. [2] B. C. Arimah, The face of urban poverty: Explaining the prevalence of slums in developing countries: Working paper. World Institute for Development Economics Research, 2010.
3
[3] M. R. Montgomery, "The urban transformation of the developing world," Science, vol. 319, pp. 761-764, 2008. [4] J. R. Philip Paar, "Low-Cost Mapping and Publishing Methods for Landscape Architectural Analysis and Design in Slum-Upgrading Projects," Future Internet vol. 3, pp. 228-247, 2011. [5] J. Barros and F. Sobreira, "City of Slums: self-organisation across scales," in Unifying Themes in Complex Systems IV, Minai Ali A. and B.-Y. Yaneer, Eds., ed: Springer, 2008, pp. 265-273. [6] C. Ioannidis, C. Psaltis, and C. Potsiou, "Towards a strategy for control of suburban informal buildings through automatic change detection," Computers,Environment & Urban Systems, vol. 33, pp. 64-74, 1// 2009. [7] D. Kohli, R. Sliuzas, N. Kerle, and A. Stein, "An ontology of slums for image-based classification," Computers, Environment and Urban Systems, vol. 36, pp. 154-163, 2012. [8] P. Hofmann, "Detecting informal settlements from IKONOS image data using methods of object oriented image analysis-an example from Cape Town (South Africa)," Jürgens, C.(Ed.): Remote Sensing of Urban Areas/Fernerkundung in urbanen Räumen, pp. 41-42, 2001. [9] R. Sliuzas, G. Mboup, & A. de Sherbinin, "Report of the expert group meeting on slum identification and mapping," CIESIN, UN-Habitat, ITC2008. [10] M. Stasolla and P. Gamba, "Spatial Indexes for the Extraction of Formal and Informal Human Settlements From High-Resolution SAR Images," Selected Topics in Applied Earth Observations and Remote Sensing, IEEE Journal of, vol. 1, pp. 98-106, 2008. [11] H. Taubenböck and N. Kraff, "The physical face of slums: a structural comparison of slums in Mumbai, India, based on remotely sensed data," Journal of Housing and the Built Environment, vol. 29, pp. 15-38, 2014. [12] P. Hofmann, J. Strobl, T. Blaschke, and H. Kux, "Detecting informal settlements from Quickbird data in Rio de Janeiro using an object based approach" Object-based image analysis, ed: Springer, 2008, pp. 531-553. [13] H. Rüther, H. M. Martine, and E. G. Mtalo, "Application of snakes and dynamic programming optimisation technique in modeling of buildings in informal settlement areas," ISPRS Journal of Photogrammetry and Remote Sensing, vol. 56, pp. 269-282, 7// 2002. [14] S. Mayunga, D. Coleman, and Y. Zhang, "A semi‐automated approach for extracting buildings from QuickBird imagery applied to informal settlement mapping," International Journal of Remote Sensing, vol. 28, pp. 2343-2357, 2007. [15] P. Joshi, S. Sen, and J. Hobson, "Experiences with surveying and mapping Pune and Sangli slums on a geographical information system (GIS)," Environment and Urbanization, vol. 14, pp. 225-240, 2002. [16] K. K. Owen and D. W. Wong, "An approach to differentiate informal settlements using spectral, texture, geomorphology and road accessibility metrics," Applied Geography, vol. 38, pp. 107-118, 2013. [17] T. Kemper and M. Pesaresi, "Searching for robust and objective EO based indicators for slum characterization," CIESIN, UN-Habitat, ITC, Enshede, Netherlands2008. [18] P. Hofmann, "Defining Robustness Measures for OBIA Framework- A Case Study for Detecting Informal Settlements," in Global Urban Monitoring and Assessment through Earth Observation, Q. Weng, Ed., ed Boca Raton: CRC Press Taylor & Francis, 2014, pp. 303-324. [19] H. Taubenböck and K. N. J, "Das globale Gesicht urbaner Armut? Siedlungsstrukturen in Slums," in Globale Urbanisierung - Perspektive aus dem All., ed: Springer, 2015. [20] C. Weber and A. Puissant, "Urbanization pressure and modeling of urban growth: example of the Tunis Metropolitan Area," Remote Sensing of Environment, vol. 86, pp. 341-352, 8/15/ 2003. [21] P. Hurskainen and P. Pellikka, "Change detection of informal settlements using multi-temporal aerial photographs–the case of Voi, SE-Kenya," in Proceedings of the 5th African Association of Remote Sensing of the Environment conference, Nairobi, Kenya, unpaginated CD-ROM, 2004. [22] D. Maktav and F. S. Erbek, "Analysis of urban growth using multi‐temporal satellite data in Istanbul, Turkey," International Journal of Remote Sensing, vol. 26, pp. 797-810, 2005/02/01 2005. [23] A. M. Dewan and Y. Yamaguchi, "Using remote sensing and GIS to detect and monitor land use and land cover change in Dhaka Metropolitan of Bangladesh during 1960–2005," Environmental monitoring and assessment, vol. 150, pp. 237-249, 2009. [24] O. Kit and M. Lüdeke, "Automated detection of slum area change in Hyderabad, India using multitemporal satellite imagery," ISPRS Journal of Photogrammetry and Remote Sensing, vol. 83, pp. 130-137, 9// 2013.
[25] O. Gruebner, J. Sachs, A. Nockert, M. Frings, M. M. H. Khan, T. Lakes, et al., "Mapping the Slums of Dhaka from 2006 to 2010," Dataset Papers in Science, vol. 2014, p. 7, 2014. [26] D. Stockwell & D. Peters, "The GARP modelling system: problems and solutions to automated spatial prediction," International Journal of Geogr. Information Science, vol. 13, pp. 143-158, 1999/03/01 1999. [27] O. François, S. Ancelet, and G. Guillot, "Bayesian clustering using hidden Markov random fields in spatial population genetics," Genetics, vol. 174, pp. 805-816, 2006. [28] L. Anselin & S. Rey, "Properties of tests for spatial dependence in linear regression models," Geographical analysis, vol. 23, pp. 112-131, 1991. [29] M. J. Hill, R. Braaten, S. M. Veitch, B. G. Lees, and S. Sharma, "Multicriteria decision analysis in spatial decision support: the ASSESS analytic hierarchy process and the role of quantitative methods and spatially explicit analysis," Environmental modelling & software, vol. 20, pp. 955-976, 2005. [30] R. Wilby, R. Abrahart, and C. Dawson, "Detection of conceptual model rainfall—runoff processes inside an artificial neural network," Hydrological Sciences Journal, vol. 48, pp. 163-181, 2003. [31] K. C. Clarke and L. J. Gaydos, "Loose-coupling a cellular automaton model and GIS: long-term urban growth prediction for San Francisco and Washington/Baltimore," International Journal of Geographical Information Science, vol. 12, pp. 699-714, 1998. [32] A. T. Crooks, "Constructing and implementing an agent-based model of residential segregation through vector GIS," International Journal of Geographical Information Science, vol. 24, pp. 661-675, 2010. [33] S. Strebelle, "Conditional simulation of complex geological structures using multiple-point statistics," Math. Geology, vol. 34, pp.1-21, 2002. [34] F. K. Shuvo & P. Janssen, "Modeling informal settlements using a hybrid automata approach," in Conference on Computer-Aided Architectural Design Research in Asia (CAADRIA 2013), 2013, p. 600. [35] M. Wooldridge & N. R. Jennings, "Intelligent agents: Theory and practice," The knowledge engineering review, vol.10, pp. 115-152, 1995. [36] N. Minar, R. Burkhart, C. Langton, and M. Askenazi, "The swarm simulation system: A toolkit for building multi-agent simulations," 1996. [37] N. Moreno, A. Ménard, and D. J. Marceau, "VecGCA: a vector-based geographic cellular automata model allowing geometric transformations of objects," Environment and Planning B: Planning and Design, vol. 35, pp. 647-665, 2008. [38] F. Wang and D. J. Marceau, "A Patch-based Cellular Automaton for Simulating Land-use Changes at Fine Spatial Resolution," Transactions in GIS, vol. 17, pp. 828-846, 2013. [39] P. Hofmann, P. Lettmayer, T. Blaschke, M. Belgiu, S. Wegenkittl, R. Graf, et al., "ABIA – A Conceptual Framework for Agent Based Image Analysis," South-Eastern European Journal of Earth Observation and Geomatics, vol. 3, pp. 125-130, 2014. [40] D. Roy, M. H. Lees, B. Palavalli, K. Pfeffer, and M. A. P. Sloot, "The emergence of slums: A contemporary view on simulation models," Environmental Modelling & Software, vol. 59, pp. 76-90, 9// 2014. [41] R. Sietchiping, "A Geographic Information Systems and Cellular Automata-Based Model of Informal Settlement Growth," PhD, School of Anthropology, Geography & Environmental Studies The University of Melbourne, 2004. [42] J. Barros, "Exploring Urban Dynamics in Latin American Cities Using an Agent-Based Simulation Approach," in Agent-Based Models of Geographical Systems, A. J. Heppenstall, A. T. Crooks, L. M. See, and M. Batty, Eds., ed: Springer Netherlands, 2012, pp. 571-589. [43] O. O. Vincent, "Exploring spatial growth pattern of informal settlements through agent based simulation," Thesis submitted to Utrecht University, Delft University of Technology, Wageningen University and the International Institute for Geo-Information Science and Earth Observation, 2009. [44] J. F. Ellen-Wien Augustijn-Beckers, Bas Retsios, "Simulating informal settlement growth in Dar es Salaam, Tanzania: An agent-based housing model," Computers, Environment and Urban Systems, vol. 35, pp. 93103, 2011. [45] A. C. Amit Patel, Naoru Koizumi "Slumulation: An Agent-Based Modeling Approach to Slum Formations," Journal of Artificial Societies and Social Simulation vol. 15 (4), 2012. [46] M. Felsen and U. Wilensky, "NetLogo urban suite-economic disparity model," ed: Center for Connected Learning and Computer-Based Modelling, Northwestern University Evanston.
4
