MONIKA KUFFER
GLOBAL URBAN DATA GAPS: MACHINE LEARNING, EARTH
OBSERVATION AND DEPRIVED URBAN AREAS
Machine Learning for Sustainable and Resilient Built Infrastructure in Urban Areas
27/4/20 2
MAPPING URBAN AREAS FROM SPACE
Global datasets of urban areas oftenvery coarse
Costs of high resolution imagery No detailed urban classes
Large data gaps in Global South
URBAN DATA GAPS - SDG INDICATOR 11.7.1 OPEN SPACES
Source:https://www.mdpi.com/2072-4292/12/7/1144 Example of Kampala, Uganda
27/4/20 3
SPATIAL DATA ON SLUM POPULATION
Missing the spatial dimension of deprivation Population estimates are very uncertain Official slum data have large data gaps
URBAN DATA GAPS - 11.1.1 INDICATORS ON SLUMS
Source: UN-Habitat, 2014
Sea level rise
-effected population?
26/4/20 4
THE IMPORTANCE OF REALISTIC DATA ON DEPRIVED AREAS
26/4/20 4
BASE DATA IS MISSING FOR MANY DOMAINS
Accounting for all urban
areas
Urban health Planning and
service provision Monitoring and decision support Environmental analysis Disaster prevention and response Climate change Community advocacy
Example of Tacloban city in the Philippines
M. Sheykhmousa, N. Kerle, M. Kuffer and S. Ghaffarian (2019), https://www.mdpi.com/2072-4292/11/10/1174
THE GLOBAL DATA GAPS
A new initiative coordinated by
UN-Habitat: Building the Climate Resilience
of the Urban Poor
Solutions are required in cities to
increase adaptation putting the most
vulnerable first
https://www.itc.nl/news/2019/9/66846/building-the-climate-resilience-of-the-urban-poor
MAJOR REASONS FOR GLOBAL DATA GAPS
Local context and training data Scalability and transferability – VHR imagery Official slum data have large data gaps
Aggregation and validation - ethics
Wang, Kuffer and Pfeffer, 2019:
https://www.sciencedirect.com/science/article/pii/ S0198971518301947 Scalable methods Predicting depriv ati on
26/4/20 7
SLUMAP PROJECT
SLUMAP (Remote Sensing for Slum Mapping and Characterization in sub-Saharan African Cities): Open-source framework that allows for the processing of remote sensing images for (i)
providing information on the location and extension of slums (ii) characterizing the
physical environment within slums at limited cost.
Methods will be tested in sub-Saharan cities, e.g. Ouagadougou (Burkina Faso), Nairobi and Kisumu (Kenya)
TWO-YEAR RESEARCH PROJECT (2019-2021) FUNDED BY THE STEREO-III PROGRAMME, BELGIAN SCIENCE POLICY (BELSPO)
DEPARTING FROM BINARY MAPS OF DEPRIVATION
Ajami, Kuffer, Persello and Pfeffer, 2019
https://www.mdpi.com/2072-4292/11/11/1282
INTEGRATED DEPRIVATION AREA MAPPING SYSTEM (IDEAMAPS)
Growing Network of Slum Mapping Experts
Source: https://www.mdpi.com/2072-4292/12/6/982
PEOPLE, PIXELS AND