Abstract :
[en] Urban flood risk in rapidly growing cities emerges from interactions among flood
hazard, exposure, and vulnerability, which are shaped over time by urban development,
infrastructure, and socioeconomic and institutional conditions. This thesis examines
how flood exposure has evolved over time, how its spatial distribution relates to urban
deprivation, which processes contribute to the production and persistence of risk, and
what these processes imply for flood-risk reduction. Greater Antananarivo (GT),
Madagascar, provides the empirical case, with particular attention to deprived urban
areas.
The research combines spatiotemporal analysis of urban expansion, comparative flood
mapping and simulation, multidimensional deprivation assessment, a survey of 400
households across 20 flood-prone neighborhoods, systemic impact-chain analysis,
analysis of flood-risk-reduction proposals formulated in different participatory settings,
and simulation-based assessment of selected intervention configurations. The analyses
span multiple levels, from GT and the Urban Commune of Antananarivo (CUA) to
neighborhoods and households.
Flood exposure accumulated over time as urban expansion and densification increased
development in flood-prone areas. Built-up land within mapped flood-prone areas
increased from approximately 399 ha in 1953 to 3,675 ha in 2017, when about 23% of
GT’s built-up area was in these areas. Its current spatial distribution, however, does not
uniformly correspond to the distribution of urban deprivation: some neighborhoods
combine high exposure and high deprivation, but this pattern is not consistent across
the study area. The results also show that analytical choices matter: different flood-
mapping approaches represent different aspects of flooding and produce different
estimates of affected areas and buildings, while the geographic scale of assessment
influences the observed spatial pattern of deprivation.
The impact-chain analysis represents flood-risk production through interacting urban,
infrastructural, and socioeconomic processes, organized around three connected
pathways: urbanization in hydrologically sensitive areas, failures in drainage and urban
services, and socioeconomic constraints on adaptive capacity. These processes link
urban conditions to exposure, disruption, and flood-related losses. The analysis also
represents a plausible feedback mechanism through which flood impacts can weaken
household capacities and constrain opportunities to reduce vulnerability or exposure,
thereby contributing to the persistence of existing risk conditions.
Community resilience cannot be inferred from household vulnerability and adaptive-
capacity profiles alone. Collective indicators did not differ significantly across these
profiles, while perceived community preparedness and reported household protection
measures were associated with household preparedness, residential context, and
perceived authorities–community communication.
In the multicriteria assessment of risk factors, high importance and urgency did not
necessarily coincide with high perceived capacity to act. Several factors related to
governance, exposure production, and socioeconomic conditions received high
priority but low perceived capacity-to-act scores, indicating that factors considered
important and urgent may also be difficult to address under existing technical,
institutional, and financial constraints. Participatory proposals addressed risk-
production mechanisms unevenly, with drainage-related mechanisms more strongly
represented than those associated with exposure production and household
socioeconomic vulnerability.
For the selected intervention configurations, changes in flood extent or water depth
did not consistently correspond to equivalent changes in building exposure. Under the
100-year current rainfall scenario, residential buildings remained exposed in
neighborhoods belonging to the two highest deprivation classes in the no-intervention
case and under all modeled intervention options.
Overall, the thesis advances a cross-scale, mechanism-based understanding of flood
risk in deprived urban neighborhoods by linking locally observed conditions to the
processes that produce and sustain exposure and vulnerability. It extends this
explanatory framework to risk reduction by examining the extent to which proposed
actions target those same mechanisms. This establishes a critical distinction between
reducing a component of risk and targeting the processes that sustain exposure and
vulnerability over time. For policy, this points to the need to align intervention choices
with the mechanisms they are intended to modify and with the actors and decision-
making levels that have the capacity to act on them.
