International Journal of Disaster Risk Reduction 67 (2022) 102683

Contents lists available at ScienceDirect

International Journal of Disaster Risk Reduction

journal homepage: www.elsevier.com/locate/ijdrr

GIS-based multi-criteria modelling of flood risk susceptibility in

Port Louis, Mauritius: Towards resilient flood management
Jay R.S. Doorga a, b, *, Leonard Magerl a, Priyal Bunwaree a, Jiaxin Zhao a,
Sophia Watkins a, Caroline G. Staub c, Soonil D.D.V. Rughooputh d, Tyagaraja S.
M. Cunden b, Roddy Lollchund d, Ravindra Boojhawon d
a
School of Geography and the Environment, University of Oxford, Oxford, United Kingdom
b
Faculty of Sustainable Development and Engineering, Université des Mascareignes, Rose Hill, Mauritius
c
Center for Urban Research, Engagement, and Scholarship, University of the District of Columbia, Washington, DC, USA
d
Department of Physics and Mathematics, Faculty of Science, University of Mauritius, Reduit, Mauritius

A R T I C L E I N F O A B S T R A C T

Keywords: Economic growth and population expansion in Mauritius have driven unprecedented rates of
Flood risk development. Most notably, increased urbanization of flood-prone regions has exacerbated
Site suitability analysis challenges around resilience and vulnerability to recurrent flooding. We implement a multi-
Geospatial assessment criteria model consisting of a physical-oriented, a social-oriented and an economic-oriented
MCDM scenario to identify highly vulnerable sites to flooding in the capital city, Port Louis. Social,
AHP
technical, economic and legal perspectives are incorporated to propose comprehensive flood
Mauritius
management strategies. The aim is to minimize the physical impacts of floods while also
addressing the broader social and economic risks. Location-based flood management strategies
are thereafter proposed to increase the resilience of the city to flooding. Measures include, inter
alia, the construction of an underground stormwater storage tank at Champ de Mars, establish­
ment of flood refugee camps at Vallée Pitot and Cité La Cure, reviewing building regulations and
urban planning, replacement of the Caudan Underpass by an Overhead Bridge, green renaturing
of rivers in downstream areas, construction of a retention pond on the downhill slope of the
mountain at Bell Village, and the setting up of a diversion tunnel linking St Louis River to Grand
River North West. The approach adopted will allow an understanding of the geography of
flooding in Port Louis and support management decisions for limiting its impacts.

1. Introduction
Floods are recurrent phenomena occurring worldwide with significant socio-economic impacts resulting in loss of lives and damage
to property [1]. In 2020, flood accounted for the highest share in death and missing toll by disaster with a value of 47.2% and resulted
in the highest direct economic loss by disaster value amounting to a share of 72.1% [2]. The frequency and intensity of flood events are
being amplified by climate change, thereby hampering the ability of Mauritius and other Small Island Developing States (SIDS) to cope
and recover between extreme weather events [3].
The geographical location of Mauritius near the tail of the Indian Ocean cyclone belt results in frequent exposure to tropical

* Corresponding author. School of Geography and the Environment, University of Oxford, Oxford, United Kingdom.
E-mail address: [email protected] (J.R.S. Doorga).

https://doi.org/10.1016/j.ijdrr.2021.102683
Received 28 August 2021; Received in revised form 7 November 2021; Accepted 9 November 2021
Available online 19 November 2021
2212-4209/© 2021 Elsevier Ltd. All rights reserved.
J.R.S. Doorga et al. International Journal of Disaster Risk Reduction 67 (2022) 102683

cyclones, floods and torrential rains. Flooding is the second biggest threat to society after cyclones, contributing to a direct loss of about
$22 million annually, equivalent to 20% of the country’s economic losses associated with disasters [4]. The majority of the incurred
losses (approx. 60%) occurs within the residential sector, with the larger ramifications felt by low-income households which are the
most vulnerable and the least likely to recover from flooding events [5]. The deadliest floods the island ever witnessed occurred in the
capital city of Port Louis, where 7 major flooding events have been reported between 1865 (30 deaths) and 2013 (11 deaths), with
water levels on the streets reaching as high as 2 m [6]. This vulnerability is further compounded by a related amplification of
water-borne and vector-bone disease transmission, with reported outbreaks of typhoid fever in the 1980s and chikungunya in the
2005–2006 rainy season in Mauritius [7]. The natural resources on which the island is highly dependent on for economic revenue and
environmental protection are also affected by floods in multiple ways: Reduced aeration in water-saturated soil leads to lower soil
fertility and muddy waters flowing towards the lagoons cause turbid sediment plumes. These can smother coral reefs and limit their
access to sunlight and oxygen, resulting in environmental stress and degradation of natural reef habitats which prompt long-term
ecological repercussions.
Mauritius experiences development challenges typical of SIDS and is faced with a dilemma: Implementing growth-conducive
structural reforms to attract foreign direct investments whilst having to enforce its regulatory framework to protect its natural re­
sources. In that optic, the local government alongside with the private sector and the civil society are being urged to address the
impacts of natural disasters on human settlements and infrastructures through resilience building, as enshrined in the Sendai
Framework for Disaster Risk Reduction. The country invests about 230 million USD annually, representing 2% of its GDP, in disaster
risk reduction through investments in climate resilient infrastructures such as improved drainage systems and elevated roads [8].
Guided by the core principles of the Sendai Framework (2015–2030) and its predecessor, the Hyogo Framework for Action
(2005–2015), the government of the island established a National Disaster Risk Reduction and Management Centre and promulgated a
National Disaster Reduction and Management Act in 2016, clearly stipulating the roles and responsibilities of key actors in the event of
a major flood. A national multi-hazard emergency alert system has also been set up to enable prompt and informed decisions in the face
of potential floods [9]. Recently, the government amended the regulations of the Prime Minister’s Relief Fund to provide assistance
and offer compensation to flood victims [10]. The Climate Change Act of 2020 was also established to assess the vulnerability of
Mauritius to climate change and to set out appropriate adaptation measures. Despite the range of policy responses adopted to alleviate
flooding impacts, flooding in Mauritius frequently results in emergency evacuations and severe damage to infrastructure and property.
Aside from climate change, policy-makers should take into consider other key drivers of flood risk. In Mauritius, these include
unregulated land development and drainage systems that are no longer adequate given the rapid pace of urbanization. The exploitation
of wetlands for building and infrastructure construction reduces the natural hydrological buffering capacity to rainwater runoff. About
70% of wetland extent in the northern coastal region of Mauritius has been lost, resulting in increased incidences and intensity of
flooding in the region [11]. The construction of drains with inadequate carrying capacities and the allocation of building permits in
low-lying, flood-prone areas further exacerbate the flood problem on the island [12]. Based on the result of a recent survey, victims of
flooding events in Mauritius were of the opinion that poorly maintained drainage facilities, as witnessed through clogged culverts and
blocked waterways by debris and garbage, was the primary reason for the increase in flooding events recorded in their localities [11].
The first step in informing the development of effective flood mitigation policies consists of identifying flood-prone areas [13,14].
Geospatial approaches in flood analysis are commonly categorized as: (1) Statistical methods, which rely on a database of historical
data processed on a GIS platform; (2) Hydrological modelling based on shallow water equations initialized by rainfall; and (3)
Multi-Criteria Decision Analysis (MCDA) [15–17]. MCDA has emerged as a formal approach to determine the optimum alternative in a
set of multiple, conflicting and interactive criteria. Çetinkaya et al. (2016) [18] mentioned that the distinctive fields of Geographical
Information System (GIS) and MCDA can operate in symbiosis to exploit the intersections between spatial representation and multiple
influential factors. Whilst GIS permits the integration of spatially referenced flood factors on a geospatial platform, MCDA provides a
structure for designing, evaluating and prioritizing decisions. Since all factors do not contribute equally to the flood problem, an
Analytical Hierarchy Process (AHP) technique is used to determine the weights of individual factors. Integral to the GIS-based MCDA
process are the determination of factor weights using AHP and combination of weighted factors using Weighted Linear Combination
(WLC) to generate a flood risk map [15–17]. MCDA has been used for flood risk analysis in multiple countries including Greece [19],
China [20], and Canada [21]. Flash-flood prone areas have also been identified in the metropolitan urban area of Athens using AHP
and multi-criteria GIS-based analysis [22]. A study conducted by de Brito and Evers (2016) [23] on the use of state-of-the-art flood
models revealed an 82% rise in the use of MCDA for flooding analysis since 2009. In contrast to approaches which focus primarily on
the economic and physical risks, the MCDA technique enables the consideration of social, environmental and cultural factors in flood
risk assessment [24]. Factoring in the socio-economic variability helps draw attention to the socially vulnerable communities and
inform more equitable policy decisions.
This study uses MCDA to identify regions that are vulnerable to flooding in the capital city of Port Louis based on three scenarios:
Physical, social and economic impacts. The authors discuss policy-based flood management strategies to reduce the impacts of flooding
in high-risk regions. The proposed strategies are discussed in relation to legal, technical, environmental, social and economic factors
and suggested avenues for future research on flooding in Mauritius.

2. Study area
Located some 800 km east of Madagascar, the main island of Mauritius (20◦ 9′ 42′′ South; 57◦ 29′ 56′′ East) measures about 1865
km2 (Fig. 1a). Of volcanic origin, the island was formed by successive submarine eruptions ca. 12 million years ago near the edge of the
South West Indian Ocean (SWIO) ridge. The topography of Mauritius is complex, consisting of a discontinuous ring of mountain ranges

2
J.R.S. Doorga et al. International Journal of Disaster Risk Reduction 67 (2022) 102683

encircling a central tableland at an elevation of around 550 m, which gently slopes down to the coastal plain (Fig. 1b). A number of
rivers and rivulets, formed in the crevices between land created by new and old lava flows, act as natural drainage system and channel
rainwater swiftly down these slopes into the sea.
Sandwiched between the North West littoral side and the Moka Range, the capital city of Port Louis is the economic centre-point of
the island (Fig. 1c). The city hosts the only trade port of Mauritius and has been, for over two centuries, the focal administrative centre
from a judicial, political and business outlook [25]. With a population of about 155,226 in 2020, Port Louis contains several suburbs on
the fringes of the city that provide shelter to squatters and low-income households in flood-prone regions. Low-income citizens living
in the suburbs are the most vulnerable and face longer recovery periods in the aftermath of a flood. This has been especially true for
communities in Cité La Cure and Canal Dayot with the overflow of Lataniers and St Louis rivers, respectively [5,26]. The City Centre is
not spared from floods as the inadequate capacities of Ruisseau Creole and Le Pouce, both natural streams which drain the lower areas,
are further limited through obstruction by debris and sediment, thereby decreasing the net outflow of rainwater into the sea at the
Caudan Basin [6].
The bedrock lithology of Port Louis is ancient basalt whilst the overlying soil type is classified as Lithosol in the mountain range,
Dark Magnesium Clay near the central section of the city, and Low Humic Latosols in the North and South West of the city [27]. The
region also hosts the Mer Rouge site located west of the Port Area, which is land reclaimed from the sea (mostly sand). Colluvium is
present near the base of hillslopes due to sediment runoff whilst some alluvium deposits have been reported near St Louis river [28].
From a geomorphological perspective, the steep-sloped relief of the concave-shaped mountain range concentrate rainwater and
channel it into the capital city. Whilst there is sparse tree vegetation with patches of heavy tree canopy in the suburbs, rapid devel­
opment near the City Centre resulted in an impermeability factor ranging from 90% to 100% [6].

3. Materials and methods

3.1. Framework overview
The methodological framework adopted in the current study is illustrated in Fig. 2. Factors contributing to flooding in Port Louis
are identified and weighted based on their importance using the Analytical Hierarchy Process (AHP). The factors identified and
weights attributed are based on three scenarios: Physical (Flood risk level related to physical impacts), social (Flood risk level related to
social vulnerability) and economic (Flood risk level related to economic losses), aimed at providing a panoply of divergent perspectives
on the flood problem. The weighted factors are combined in order to obtain flood risk maps of the city, delineating regions of high,
intermediate and low risks. All geospatial analysis is conducted using ESRI ArcMap - version 10.3. Regions of high risks are further
explored and management responses are proposed in order to mitigate the flood impacts. The political, economic and legal dimensions
of the proposed management measures are discussed with the objective of assessing the feasibility, conformity and acceptability of
project implementation.

3.2. Datasets
The datasets (Fig. 3) used in the formulation of the models investigated, are presented in Table 1.

Fig. 1. (a) Geographical location of the island of Mauritius in the SWIO basin. (b) Topographic relief map of Mauritius alongside with the location of the study area,
Port Louis, on the North West coast. (c) Aerial view of the landscape of Port Louis.

3
J.R.S. Doorga et al. International Journal of Disaster Risk Reduction 67 (2022) 102683

Fig. 2. Methodological flowchart illustrating the scenarios explored and steps adopted to assess flood risks in Port Louis.

3.3. Identification of influential factors used in flood model analysis

The implementation of a flood risk model requires the integration of factors that influence flooding within a certain locality. A
synthesis of the main factors used in flood risk studies worldwide is presented in Table 2. Also presented, are the weights of the factors
used, indicating the degree of importance to the flood problem in the different localities.
The selection of criteria to implement a flood risk model is arbitrary in nature and varies from region to region, as observed in
Table 2. However, there is some overlap among factors employed in the different studies, whilst the order of importance tends to follow
a general trend. Proximity to discharge channels has been strongly recommended by Fernandez and Lutz (2010) [39] and Hossain and
Meng (2020) [37] whilst implementing their flood risk models. Hammami et al. (2019) [41] considered land use as the most important
factor in the construction of a flood risk model for Tunis City. Studies by Mahmoud and Gan (2018) [40] and Vignesh et al. (2021) [42],
report drainage density as a key factor of influence, followed by slope, soil type and curvature. This general order of importance is
repeatedly featured in the scientific literature, which led the authors to adopt this framework as a basis for the construction of a flood
risk model for Port Louis.

3.3.1. Selected influential factors in flood risk modelling for Port Louis
A description of the factors used in flood risk modelling for the region of Port Louis is presented below. Aside from the factors
discussed in section 3.3, the authors include proximity to blocked waterways as a pivotal factor in our flood risk model. Additional
factors allow for socio-economic considerations to be included in the analysis.
3.3.1.1. Elevation. The elevation factor is an important criterion dictating the control of the overflow direction movement and the
depth of the flood [35]. Owing to its influence in impacting the propagation of the flood, the elevation factor is taken into consideration
such that regions of high elevations would have inherently low flood risks, whilst regions of low elevations would experience higher
flooding risks. The Digital Elevation Model (DEM) (10 m resolution) was acquired from the Cartographic section of the Ministry of
Housing and Lands.
3.3.1.2. Blocked waterways. As mentioned by Jhingut and Proag (2014) [10], the principal cause of flooding in Port Louis is due to the
accumulation of debris blocking waterways downstream. Weeks et al. (2013) [36] support the argument that floating debris is swept
up and lodged on the upstream side of culverts as the flooding increases. This causes other debris to accumulate due to blockage in the
culvert structure, eventually cutting off the waterway. Potential clogging points are identified in the current study at road-stream
intersections.
3.3.1.3. Proximity to discharge channel. Areas situated near to discharge channel networks have a higher probability of being flooded
with the overflow of rivers during heavy rainfall [37]. Records of flooding events in Mauritius suggest that communities living near
riverbanks and channels are the most vulnerable to floods. River density data was digitized from Google Earth.
3.3.1.4. Drainage density. The drainage density, defined as the total length of all streams per unit area of a drainage basin, is an
important factor in flood risk analysis. In a drainage basin containing several flow accumulation paths feeding into a main river, the
probability of flooding may be higher where multiple rivers join as they add the precipitation runoff from multiple unit areas. As
mentioned by Islam and Sado (2000) [38], regions located in close proximity to drainage networks and flow accumulation paths tend

4
J.R.S. Doorga et al. International Journal of Disaster Risk Reduction 67 (2022) 102683

Fig. 3. Factors employed in the flood risk model for the region of Port Louis.

to have higher flood risk susceptibilities. The drainage density was derived from the DEM.
3.3.1.5. Land use. Land use/cover influences water infiltration rates such that impervious materials, including concrete buildings and
asphalt pavements, tend to increase runoff and reduce permeability as compared to vegetated areas [39]. In an urban setting such as
that of Port Louis where impermeable surfaces are dominant, factoring the land use factor is of paramount importance to identifying
flood-prone regions. The land use map was derived through reclassification of a 15 cm-5 m resolution aerial photograph of Port Louis.
3.3.1.6. Slope. Flood risk is also influenced by the slope of the terrain. Steep slopes lead to less infiltration and lower flood risk due to

5
J.R.S. Doorga et al. International Journal of Disaster Risk Reduction 67 (2022) 102683

Table 1
Specifications of the data employed in the current study.

Data type Temporal coverage Spatial resolution Data source

Elevation (DEM) 2008 10 m Ministry of Housing & Lands

Rivers 2021 15 cm–5 m Digitized from Google Earth
Roads 2021 15 cm–5 m Digitized from Google Earth
Blocked waterways 2021 15 cm–5 m Derived from rivers and roads
Proximity to rivers 2021 15 cm–5 m Derived from rivers
Drainage density 2008 10 m Derived from DEM
Land use 2021 15 cm–5 m Reclassified from Google Earth image
Slope 2008 10 m Derived from DEM
Soil type 1962 – ESDAC
Curvature 2008 10 m Derived from DEM
Poverty rate 2007 – Statistics Mauritius
Economic zones 2021 15 cm–5 m Digitized from Google Earth

the increased velocity of surface runoff [40]. Flat surfaces where pooling can occur will tend to retain more surface runoff, thereby
increasing the risks of flooding. The slope map was derived from the DEM.
3.3.1.7. Soil type. The type of the soil also affects the infiltration of rainwater to the subsurface and influences flood risk level in a
given region. Impermeable rocks increase surface runoff leading to flooding, whilst permeable soil layers enhance rainwater infil­
tration thereby dampening flood risk. In the region of Port Louis, besides the reclaimed land near the Port Area, the three main soil
groups are: Lithosol, Dark Magnesium Clay and Low Humic Latosols [27]. Lithosol has a high runoff potential (very poorly drained),
Dark Magnesium Clay has a relatively lower runoff potential (poorly drained), whilst Low Humic Latosols has the lowest runoff po­
tential (well drained) [41]. The soil type map was acquired from ESDAC.
3.3.1.8. Curvature. Profile curvature influences the surface flow such that concave and flat surfaces would tend to retain water for a
prolonged period as compared to convex surfaces [42]. Rainwater accumulation occurs in concave and flat areas where there is lower
surface runoff, in contrast to convex surfaces where the runoff potential is significantly higher. Concave depressions upstream (i.e.
retention ponds) are highly desirable to retain surge water and slow the flow. The curvature map was derived from the DEM.
3.3.1.9. Poverty rate. A strong relationship exists between flooding and poverty such that flooding exacerbates poverty, whilst poverty
increases flood vulnerability [43]. With the aim of providing a social orientation to the flood risk assessment in Port Louis, the poverty
rate factor is taken into consideration. Regions with the highest poverty rates would be most vulnerable to flooding events. The poverty
rate dataset was made available from Statistics Mauritius [44].
3.3.1.10. Economically significant regions. Inclusion of an economic factor which considers the economic impacts of flooding may be
useful for identifying economically sensitive regions in order to devise strategies to minimize losses. Flooding impacts at the industrial
level could propagate to the rest of the economy and disrupt the country’s economic stability, as the manufacturing sector contributes a
major share of the GDP of SIDS. The Thai floods of 2011, for instance, inundated 7 industrial parks in which 804 companies were
operating, causing a decline in real GDP of about 8.6% between October and December 2011 [45]. The region of Port Louis hosts some
regions of economic importance, including the City Centre, the Port Area, the Champ de Mars Racecourse, as well as hospitals and
government schools. Economically significant regions were identified using Google Earth.

3.4. GIS-based MCDA

3.4.1. Analytical Hierarchy Process (AHP)
Flood risk assessment is an intrinsically complex, multidimensional process involving highly uncertain quantitative and qualitative
factors [46]. Initially proposed by Saaty (1977) [47] and applied by Ouma and Tateishi (2014) [48], Gigović et al. (2017) [49], and
Lyu et al. (2018) [50], AHP seeks to reduce complexity through the implementation of a hierarchical representation of a
decision-making problem achieved by pairwise comparisons. After the flood problem is decomposed into component factors, the
weights of individual factors are acquired through subjective judgements based on a pairwise comparison scale (Table 3), leading to an
evaluation matrix from which the normalized eigenvector is derived.
The result of the pairwise comparisons of n criteria can be represented by the evaluation matrix A, where element
aij (i, j = 1, 2…, n) indicates the quotient of weight, as given in Eq. (1).

⎡ a11 a12 … a1n ⎤

⎢ a21 a22 … a2n ⎥ 1

A=⎣ ⎦, aii = 1, aij = , aji ∕
= 0. (1)
⋮ ⋮ ⋱ ⋮ aji
an1 an2 … ann
The right eigenvector, w, corresponding to the maximum eigenvalue, λmax is computed to determine the relative weights, Aw of the
matrix as follows:
Aw = λmax w (2)
Prior to the application of derived weights in the flood risk model, a consistency check is used to ensure the reliability of the
adopted weighting scheme. The level of consistency is calculated using Eq. (3) as follows:

6
J.R.S. Doorga et al. International Journal of Disaster Risk Reduction 67 (2022) 102683

Table 2
Synthesis of influential factors and weights employed in flood risk models worldwide.

Country Region Technique Factor and Weight Reference

USA Birmingham city, Alabama • AHP; • Elevation (30%) Hossain and Meng (2020) [29]
• WLC; • Distance to river (29%)
• Land use (17%)
• Flow accumulation (11%)
• Slope (9%)
• Soil type (4%)
Philippines Davao Oriental • AHP; • Rainfall (41.8%) Cabrera and Lee (2019) [30]
• WLC; • Slope (23.8%)
• Elevation (14.7%)
• Distance to river (9.5%)
• Drainage density (6.1%)
• Soil type (4.1%)
Argentina Tucumán Province • AHP; • Distance to the discharge channels (41.4%) Fernandez and Lutz (2010) [31]
• WLC; • Elevation (20.6%)
• Slope (20.6%)
• Depth to groundwater table (10.9%)
• Cover Type (6.5%)
Egypt Beheira and Alexandria • AHP; • Flow accumulation (17.8%) Mahmoud and Gan (2018) [32]
• WLC; • Distance drainage network (8.9%)
• Elevation (9%)
• Land use (7.3%)
• Rainfall intensity (6.6%)
• Slope (5.2%)
• Geology (2.5%)
• Runoff (18.9%)
• Soil type (13.8%)
• Drainage density (7.9%)
• Curvature (2.2%)
Tunisia Tunis City • AHP; • Land use (23.8%) Hammami et al. (2019) [33]
• WLC; • Elevation (18.5%)
• Lithology (18.5%)
• Rainfall (15.9%)
• Drainage density (10.6%)
• Slope (8.4%)
• Soil (5.3%)
• Ground water level (2.7%)
India Kanyakumari district, Tamil Nadu • AHP; • Rainfall (22%) Vignesh et al. (2021) [34]
• WLC; • Slope (12.4%)
• Drainage density (15.4%)
• Land use (15.3%)
• Elevation (10.9%)
• Soil (6.8%)
• Geology (5.2%)
• Geomorphology (4.8%)
• Runoff (4.2%)
• Topographic wetness index (3%)

Table 3
Fundamental scale for pairwise comparisons.

Intensity of Importance Definition Explanation

1 Equally significant Two elements contributing equally to the objective

3 Moderate significance Experience and judgement slightly favour one element over another
5 Strong significance Experience and judgement strongly favour one element over another
7 Very strong significance Activity is strongly favoured and its dominance is demonstrated in practice
9 Extreme significance The evidence favoring one parameter over another is of the highest possible order
2,4,6,8 Intermediate values Used to represent compromise between priorities listed above

CI
CR = , (3)
RI

7
J.R.S. Doorga et al. International Journal of Disaster Risk Reduction 67 (2022) 102683

where RI represents the consistency of a randomly generated pairwise comparison matrix and CI is the consistency index expressed as
follows:
λmax − n
CI = . (4)
n− 1
The subjective judgement is deemed acceptable when CR ≤ 10%. However, when CR > 10%, the subjective judgement is incon­
sistent and the pairwise comparisons are repeated until consistency is achieved.

3.4.2. Weighted Linear Combination (WLC)

The flood risk map of Port Louis, delineating regions of high, intermediate and low risks, is generated using the WLC technique.
Falling under the domain of bivariate statistical analysis, this technique combines factors by assigning respective weights acquired
from AHP [51] and is expressed as follows:
∑
n
Flood risk = wj rij , (5)
j=1

where wj is the weight of factor j, rij is the normalized value of area i under factor j, and n is the total number of factors.

3.4.3. Decision matrix and AHP employed

The decision matrix and AHP employed for the three models investigated are presented below. A pairwise comparison is first
performed for subsequent criteria pairing based on Saaty’s scale where a score of one is attributed to an equally good comparison
whilst a score of nine is given for an extremely significant comparison. The assigned intensity of importance for the factors compared
are based on a thorough investigation of wide-ranging flood risk models in literature, as detailed in section 3.3. A consistency ratio test
is performed to ensure consistency of the AHP and criteria weights are thereafter derived from the resulting decision matrix. Factors
are normalized and criteria weights are applied to respective factors prior to applying the WLC combinatorial process.

1) Physical-oriented flood risk model

The physical-oriented flood risk model aims at identifying low-lying regions near discharge channels, with extensive impervious
surfaces and in close proximity to debris-blockage points along waterways, all of which maximize the likelihood of flooding. The
physical-oriented model seeks to identify regions of highest impact to flooding.
The MCDA model begins with the standardization of factors in order to bring them on a common scale of measurement to enable
comparison and combination of factors. The weights of the factors indicating their importance in the combinatorial process are
thereafter derived using AHP. Tables 4a and 4b presents the AHP results for the factor weights and decision matrix, respectively.
The AHP method involved 28 pairwise comparisons, yielding a consistency ratio of 2.9%, which is deemed acceptable due to being
lower than the 10% threshold value. The derived weights presented in Table 4a are based on the principal eigenvalue of 8.28, with an

Table 4
AHP results showing the (a) factor weights and (b) decision matrix for the physical-oriented flood risk model.

(a) Factor weights

Cat Criteria Weight Rank

Z1 Proximity to potential debris-blockage point 33.1% 1

Z2 Proximity to discharge channel 23.1% 2
Z3 Land use 15.7% 3
Z4 Elevation 10.6% 4
Z5 Drainage density 7.1% 5
Z6 Slope 4.8% 6
Z7 Soil type 3.3% 7
Z8 Curvature 2.3% 8

(b) Decision matrix

Z1 Z2 Z3 Z4 Z5 Z6 Z7 Z8

Z1 1 2 3 4 5 6 7 8
Z2 1 1 2 3 4 5 6 7
2
Z3 1 1 1 2. 3 4 5 6
3 2
Z4 1 1 1 1 2 3 4 5
4 3 2
Z5 1 1 1 1 1 2 3 4
5 4 3 2
Z6 1 1 1 1 1 1 2 3
6 5 4 3 2
Z7 1 1 1 1 1 1 1 2
7 6 5 4 3 2
Z8 1 1 1 1 1 1 1 1
8 7 6 5 4 3 2

8
J.R.S. Doorga et al. International Journal of Disaster Risk Reduction 67 (2022) 102683

eigenvector solution comprising of 5 iterations and a Kronecker delta, δ, value of 3.3 × 10− 8.
2) Social-oriented flood risk model
The social-oriented flood risk model aims at identifying low-lying regions characterized by low infiltration rates, high poverty rates,
and which are located in close proximity to discharge channels and debris-blockage points along waterways. Poverty increases
vulnerability to flooding as poor communities, trapped in flood-prone regions, often lack the resources to cope and respond to flood
disasters. With repeated flood events and limited capacity to cope, the resilience and recovery capacities of these communities are
seriously impacted, resulting in a vicious cycle of vulnerability. Consequently, the implemented model aims at identifying these poor
vulnerable communities prior to formulating inclusive management strategies and policies in order to alleviate the effects of floods at a
social level.
The weights of the factors that influence the social-oriented flood risk model are derived through pairwise analysis, with the AHP
results presented in Tables 5a and 5b. The factors are then normalized and combined using the WLC process, resulting in a social-
oriented flood risk map.
For the social-oriented model, the AHP method involved 36 pairwise comparisons, yielding a consistency ratio of 3.5%, which is
also deemed acceptable due to being lower than the 10% threshold value. The derived weights presented in Table 5a are based on the
principal eigenvalue of 9.40, with an eigenvector solution comprising of 5 iterations and a Kronecker delta, δ, value of 8.1 × 10− 8.
3) Economic-oriented flood risk model
The economic-oriented flood risk model targets regions of low altitudes, located in close proximity to economically significant
regions which are covered predominantly by impervious surfaces, and within close distance to potential debris-blockage points and
discharge channels. It seeks to identify flood-prone and economically-significant regions with potentially economic-wide impact.
The MCDA process begins with the estimation of weights for the factors used in the combinatorial process, derived using AHP.
Tables 6a and 6b presents the weights derived using pairwise analysis. The factors are normalized prior to combination, resulting in an
economic-oriented flood risk map.
For the economic-oriented model, the AHP method also involved 36 pairwise comparisons, yielding a consistency ratio of 3.5%,
which is also deemed acceptable. The derived weights presented in Table 6a are based on the principal eigenvalue of 9.40, with an
eigenvector solution comprising of 5 iterations and a Kronecker delta, δ, value of 8.1 × 10− 8.

3.4.4. Standardization of factors

The standardization of factors’ scales is performed using the Z-score normalization method. This approach gives an indication of the
distance of a data value from the geometrical mean, and is given as follows:

Table 5
AHP results showing the (a) factor weights and (b) decision matrix for the socially-oriented flood risk model.

(a) Factor weights

Cat Criteria Weight Rank

Z1 Proximity to potential debris-blockage point 31.2% 1

Z2 Proximity to discharge channel 22.2% 2
Z3 Land use 15.5% 3
Z4 Poverty rate 10.8% 4
Z5 Elevation 7.4% 5
Z6 Drainage density 5.1% 6
Z7 Slope 3.5% 7
Z8 Soil type 2.5% 8
Z9 Curvature 1.8% 9

(b) Decision matrix

Z1 Z2 Z3 Z4 Z5 Z6 Z7 Z8 Z9

Z1 1 2 3 4 5 6 7 8 9
Z2 1 1 2 3 4 5 6 7 8
2
Z3 1 1 1 2 3 4 5 6 7
3 2
Z4 1 1 1 1 2 3 4 5 6
4 3 2
Z5 1 1 1 1 1 2 3 4 5
5 4 3 2
Z6 1 1 1 1 1 1 2 3 4
6 5 4 3 2
Z7 1 1 1 1 1 1 1 2 3
7 6 5 4 3 2
Z8 1 1 1 1 1 1 1 1 2
8 7 6 5 4 3 2
Z9 1 1 1 1 1 1 1 1 1
9 8 7 6 5 4 3 2

9
J.R.S. Doorga et al. International Journal of Disaster Risk Reduction 67 (2022) 102683

Table 6
AHP results showing the (a) factor weights and (b) decision matrix for the economic-oriented flood risk model.

(a) Factor weights

Cat Criteria Weight Rank

Z1 Proximity to potential debris-blockage point 31.2% 1

Z2 Proximity to discharge channel 22.2% 2
Z3 Land use 15.5% 3
Z4 Proximity to economically significant regions 10.8% 4
Z5 Elevation 7.4% 5
Z6 Drainage density 5.1% 6
Z7 Slope 3.5% 7
Z8 Soil type 2.5% 8
Z9 Curvature 1.8% 9

(b) Decision matrix

Z1 Z2 Z3 Z4 Z5 Z6 Z7 Z8 Z9

Z1 1 2 3 4 5 6 7 8 9
Z2 1 1 2 3 4 5 6 7 8
2
Z3 1 1 1 2 3 4 5 6 7
3 2
Z4 1 1 1 1 2 3 4 5 6
4 3 2
Z5 1 1 1 1 1 2 3 4 5
5 4 3 2
Z6 1 1 1 1 1 1 2 3 4
6 5 4 3 2
Z7 1 1 1 1 1 1 1 2 3
7 6 5 4 3 2
Z8 1 1 1 1 1 1 1 1 2
8 7 6 5 4 3 2
Z9 1 1 1 1 1 1 1 1 1
9 8 7 6 5 4 3 2

Fig. 4. Normalized factors employed in the models implemented for Port Louis.

10
J.R.S. Doorga et al. International Journal of Disaster Risk Reduction 67 (2022) 102683

xi − μ
Z − score =
σ

where xi denotes the data value, μ denotes the geometrical mean, and σ is the standardization of the distribution.
The standardized factors, graded from 1 (low influence on flooding) to 9 (high influence on flooding), are presented in Fig. 4.

4. Results
The factors used in the construction of the flood risk models for Port Louis, as described in section 3.3.1, are shown in Fig. 3. Three
models are explored in order to offer different perspectives to the problem of flooding in the locality. The flood models implemented
are: Physical-oriented, social-oriented and economic-oriented. A grading comprised of five classes is used for the three models being
implemented. The grades are as follows: Very high (0.8–1); High (0.6–0.8); Moderate (0.4–0.6); Low (0.2–0.4); Very low (0–0.2).

4.1. Physical-oriented flood risk model

Following factor reclassification and allocation of weights derived using AHP, aggregation of reclassified and weighted factors was
performed using the WLC decision rule algorithm. The physical-oriented flood risk map for Port Louis, generated using the GIS-based
MCDA technique, is shown in Fig. 5. The regions are graded from very high to very low flood risk levels. The most vulnerable sites to
flooding correspond to regions that lie close to debris blockage points and discharge channels, whilst having the characteristics of
being relatively flat and impervious lands of low altitudes and high drainage densities. The highly vulnerable sites identified by the
physical-oriented model were also classified as vulnerable by the Water Resources Unit of Mauritius and GIBB consulting in a study
investigating land drainage issues of Mauritius [52]. The results presented in this study are in line with flood vulnerability assessments
conducted by the Water Resources Unit of Mauritius and GIBB consulting as highlighted in Fig. 5. Highly vulnerable sites are identified
in Fig. 5 at Canal Dayot, Bell Village, in the vicinity of Line Barracks Police Station, along Le Pouce stream from Tranquebar to Les
Jardins de la Compagnie near the City Centre, at Vallée Pitot, Plaine Verte, and along Lataniers River from Vallée des Prêtres to La
Cure. The delineated highly vulnerable regions are located near potential debris-blockage points and have low infiltration rates, as
characterized by their low slope angles and surface properties (predominantly build-up urban areas). However, the factors having the
highest influence on the result obtained are proximity to potential debris-blockage points near discharge channels and the impervious
nature of land in the vicinity of the vulnerable sites.

4.2. Social-oriented flood risk model

From a social perspective, identified flood-prone regions with relatively high poverty rates are concentrated to the East of Port
Louis near Tranquebar, Vallée Pitot, Camp Yoloff, Roche Bois, La Cure and Vallée des Prêtres, as shown in Fig. 6. The results of the
social oriented model concur with the findings obtained by Chacowry et al. (2018) for La Cure [5], Rushdie (2006) for Roche Bois [53],
CADRI (2020) for Tranquebar, Vallée Pitot and Camp Yoloff [54] and Le Mauricien (2021) for Vallée des Prêtres [55]. These regions
host low-cost housings in land-affordable, flood-prone areas, and squatter settlements with poor housing state which are highly
vulnerable to floods [56]. The squatter communities, driven by poverty and housing problems, occupy state lands without authori­
zation, refuse to invest in stronger houses and take longer time to recover from flood impacts [5]. The factors having the most sig­
nificant influence on the result generated are poverty rate, proximity to debris-blockage points near discharge channels, and regions
with low infiltration rates. The high poverty rates in these regions increases vulnerability of disadvantaged groups to flooding and

Fig. 5. Physical-oriented flood risk map for the city of Port Louis.

11
J.R.S. Doorga et al. International Journal of Disaster Risk Reduction 67 (2022) 102683

Fig. 6. Social-oriented flood risk map for the city of Port Louis.

decreases their ability to cope and recover from successive flood impacts. The identification of highly vulnerable groups will help
devise appropriate flood mitigation strategies to dampen social vulnerability.

4.3. Economic-oriented flood risk model

The result from the economic-oriented scenario which seeks to reveal flood-prone regions that may have economically significant
repercussions is shown in Fig. 7. As shown in the figure, highly vulnerable regions are located near the city centre in the vicinity of
Jeetoo Hospital, Line Barracks Police Station, City Centre, Champ de Mars, Port Area and public schools. Economically vulnerable
regions that have been affected by flooding have previously been reported in literature and include the Port Area [57], Bell-Village
[58], Champ de Mars [59], and Line Barracks Police Station [12]. The city centre is mainly a commercial and business area which
contributes significantly to the national GDP [60]. Consequently, identification of regions highly vulnerable to flooding near the city
centre that have the potential to cause long-term economic repercussions, is of national interest. The horse racing industry is another
economic pillar of the country, and generates worth $0.15 billion in GDP, creating more than 2000 jobs and contributing to $16.7
million of Government revenue [61]. Additionally, reparation of damaged public infrastructures, including public schools, hospitals
and road network cause significant economic strain. Besides the proximity to potential debris-blockage points near discharge channels,
and the nature of the terrains which are relatively impervious given the urban context, the proximity to economically significant

Fig. 7. Economic-oriented flood risk map for the city of Port Louis.

12
J.R.S. Doorga et al. International Journal of Disaster Risk Reduction 67 (2022) 102683

regions have a significant influence on the result obtained. The implemented model is tailored to identify economically vulnerable sites
in order to devise mitigation strategies to reduce the economic threats posed by flooding.

4.4. Model similarities and differences

Analysis of the flood risk models investigated revealed similarities and differences that exist among the three scenarios: physical-
oriented, social-oriented, and economic-oriented. The differences in highly vulnerable regions identified between the social and
economic-oriented scenarios are blatant. The reason for this is that highly vulnerable flood-prone regions identified in regions with
high poverty rates (social-oriented scenario) are not located anywhere near economically significant regions which are predominantly
revealed in the economic-oriented scenario. Nonetheless, most of the highly vulnerable sites identified in the social and economic-
oriented models are captured in the physical-oriented model, representing a subset of this impact-oriented physical model.

5. Discussion
Flood risk reduction and adaptation responses to climate change have evolved spatially from a one-size-fits all strategy to place-
based responses, in order to offer effective solutions to the problem of flooding. This section elaborates on some place-based flood
management strategies at highly vulnerable sites identified previously that would help improve the resilience of Port Louis city to
flooding. In view of addressing the flood problem in Mauritius, the Government has recently allocated a sum of USD 0.28 billion for the
establishment of a National Management Flood Programme aimed at implementing some 1500 drainage projects around the island
[62]. Whilst improving the efficiency of the drainage system is of absolute necessity, its effectiveness is limited by the spatial re­
strictions imposed by the pre-existing urban fabric. We provide a diverse and inclusive policy package proposal to complement the
drainage improvement project of the Government. The package incorporates social, technical, legal and economic perspectives, aiming
at dampening the physical impacts of flood, curtailing economic losses and addressing the social risks associated with flood events.
The flood risk models implemented in the current study identified highly vulnerable sites to flooding upstream and downstream of
Le Pouce and La Paix streams, posing a socio-economic threat to the City Centre (Figs. 5 and 7). In order to minimize flood risks in that
region, an underground stormwater storage facility could be constructed, as shown in Fig. 8a. Underground stormwater storage
systems have been highly effective at reducing the risk of flooding in the Wan Chai district of Hong Kong and in the Saitama Prefecture
of Japan [63,64]. The construction of a 60,000 m3 underground stormwater storage tank beneath the Hong Kong Happy Valley
racecourse dampened the flood threats posed to over 150,000 people in low-lying areas, in addition to providing stored water for
landscape irrigation throughout the year [65]. In the context of Port Louis, the Champ de Mars racecourse is strategically located
mid-stream, in-between Le Pouce and La Paix streams and can accommodate a 50,000 m3 underground stormwater storage tank of
similar specifications to that at Happy Valley. Stormwater can be temporarily stored in the tank through deviation using a culvert
system in order to attenuate the peak flow and used in landscape irrigation throughout the year.
Vulnerable, low-income communities living in flood-prone areas often seek shelter in refugee camps constructed at strategic lo­
cations around the island. The locations of socially vulnerable sites identified using the social-oriented flood risk model (Fig. 6) and the
placements of refugee camps, made available by the National Disaster Risk Reduction and Management Centre [66], are illustrated in
Fig. 8b. It can be observed that all vulnerable sites, at the exception of those located at Vallée Pitot and Cité La Cure, are located in
proximity to refugee camps. Vulnerable communities residing in these two locations have to travel long distances of about 1 km to find
shelter in the closest refugee stations during heavy rainfall. We propose the construction of refugee camps in proximity to low-income,
vulnerable communities at Vallée Pitot and Cité La Cure in order to compensate for their low mobility capacities in the event of a
torrential rainfall. However, whilst the construction of refugee camps can contribute in minimizing the loss of lives, it does not address
the root cause of vulnerability. Whilst relocating marginalized groups to safer regions may be regarded as a long-term solution,
communities living in those regions have developed a strong sense of belonging and attachment to the place and are unwilling to leave
the comfort of their homes [5]. Consequently, we propose that the construction of refugee camps at strategic locations be accompanied
by a plan to combat poverty and social exclusion. Financial stability and restoration of social justice may be beneficial to marginalized
groups. Activation of the Marshall Plan to eliminate poverty and reduce inequality [67] through the mobilization of multi-stakeholders
and facilitating the dialogue between marginalized communities and Government authorities, is a progressive step in the right
direction.
Weak law enforcement and inadequate planning have resulted in the construction of buildings in flood-prone regions, with little
consideration given to green infrastructures that would allow stormwater to soak into the ground. In several localities of Port Louis, as
characterized by the region of Vallée des Prêtres, vegetation has been replaced by impervious surfaces whilst buildings have been
constructed within a distance of 33 m from rivers, streams and canals, non-compliant with the Rivers and Canals Act of 1863 [68]. In
the recent budget speech 2021–2022, the Government recently proposed the amendment of the Rivers and Canal Act of 1863, the Land
Drainage Authority Act of 2017, and the Local Government Act of 2011 to remove legal roadblocks that would slow down the
implementation of the drainage improvement project of the Government [69]. There are also proposals for the amendment of the
Forests and Reserves Act to ensure natural drains are preserved. Whilst these are progressive measures, we advocate the enforcement of
the Building Control Act of 2012 and propose amendment of the Professional Architects’ Council Act of 2011 to prohibit ‘signature de
complaisance’, thereby allowing architects to be an integral part in building planning to minimize flood risks, rather than merely
apposing their signature on projects they have not developed. We also propose the enforcement of the Town & Country Planning Act of
1954 to respect minimum setback distances from water bodies, and prohibit construction in flood-prone regions.
Six of the eleven victims of the 2013 Port Louis flood event lost their lives in the Caudan Pedestrian Underpass after the latter was
completely flooded. A judicial enquiry into the matter revealed that security officers posted there failed to close the gates to prevent

13
J.R.S. Doorga et al. International Journal of Disaster Risk Reduction 67 (2022) 102683

Fig. 8. Proposed flood risk management strategies in Port Louis: (a) Construction of an underground stormwater storage tank at Champ de Mars; (b) Establishment of
refugee camps in proximity to vulnerable sites; (c) Review of building regulations and urban planning; (d) Construction of an overhead pedestrian bridge to replace the
Caudan underpass; (e) Green renaturing of rivers in downstream areas; (f) Construction of a retention basin to dampen flood risks at Bell Village; (g) Implementation of
a diversion tunnel to deviate excess river water from St Louis River to Grand River North West so as to reduce flood risk at Canal Dayot. (For interpretation of the
references to colour in this figure legend, the reader is referred to the Web version of this article.)

pedestrian access whilst the water pumps were ineffective, causing it to be flooded in split-seconds [70]. Whilst measures have been
taken to enforce and temporarily close the underpass as soon as water level exceeds critical level, we propose the permanent closure of
the underpass, relocation of underpass shop owners, and the construction of an Overhead Pedestrian Bridge leading to the Caudan

14
J.R.S. Doorga et al. International Journal of Disaster Risk Reduction 67 (2022) 102683

Waterfront, instead (Fig. 8d). An additional remedial measure to minimize flooding within the vicinity of the waterfront includes
investing in green renaturing of rivers in downstream areas, building several small bends and planting vegetation to slow the flow and
combat bank erosion. (Fig. 8e).
To address the problem of flooding further southwards in the vicinity of Bell Village, we propose the construction of a retention
pond, the capacity of which will require a separate analysis beyond the scope of the current paper. Model simulation results have
indicated that the flooding occurring in the vicinity of Bell Village is predominantly attributed to the convergence of multiple flow
accumulation pathways associated with the relatively steep slopes and complex mountain terrain. Due to the pre-existing urban fabric
at the foot of the mountain and the increasing pace of regional development, widening of existing drainage system is proving to be a
challenge. Instead, a parcel of barren land has been identified on the downhill slope of the mountain (Fig. 8f) to temporarily store
surface runoff, reduce flood peak flow, and minimize flooding condition in low-lying Bell Village region. Finally, to dampen the flood
risk in the vicinity of Canal Dayot, we propose the construction of a diversion tunnel to deviate excess stormwater from the limited
drainage capacity of St Louis River to the higher drainage capacity of Grand River North West (Fig. 8g).

5.1. Limitations and future research

The spatial resolution of the geospatial information used in this study can be improved and higher resolution geospatial data would
likely produce more accurate results. The aggregation method used assumes a linear combination of criteria but the combination may
be non-linear in nature; this assumption must be put to the test. Importantly, the uncertainty associated with the weights applied to the
selected flood drivers has yet to be quantified. We hope to address these limitations in the future by using a mixed-methods approach
including surveys and interviews.

6. Conclusions
A GIS-based multi-criteria model was used to integrate elevation, slope, land use, soil type, drainage density, curvature, proximity
to rivers and blocked waterways into a geospatial representation of regions that are highly vulnerable to flooding in Port Louis,
Mauritius. Hazard-related factors and associated weights are used to estimate flood risk levels in terms of physical impacts, social
vulnerability and economic losses. Location-specific historical data were used to validate model results.
A more comprehensive assessment of flood risk and flood risk management in Port Louis would benefit from a cross-validation of
the results using different methods, a quantification of the uncertainty associated with factor weights and a cost-benefit analysis of
proposed strategies for each vulnerable locality, in line with the available financial budget for project implementation. While further
research is recommended, initial recommendations for reducing impacts in flood-prone areas likely include the re-engineering of the
urban landscape of Port Louis to manage rainwater runoff more effectively, an increase in underground stormwater storage, investing
in the renaturing of rivers, enforcing building and planning regulations and including green infrastructure to improve rainwater
infiltration rates in the capital.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to
influence the work reported in this paper.

Acknowledgements
We acknowledge the support provided by the University of Oxford in the provision of facilities for conducting the research pre­
sented in this paper. We are also grateful to the Université des Mascareignes and the University of Mauritius for the provision of data
and facilities. We wish to express our gratitude to Professor Jim Hall and Dr Linus Mattauch for their guidance in the group project
leading to this manuscript. Our thanks are extended to Mr Rajen Suntoo for his insights pertaining to the social risks associated with
flooding in Port Louis. Mr Ryan Tannoo also provided relevant insights in urban planning and architectural practices in Mauritius. We
thank the three anonymous reviewers who provided highly insightful comments to help improve the work presented in this paper. We
acknowledge the contribution of the Cartographic Section of the Ministry of Housing and Lands, Statistics Mauritius, and the European
Soil Data Centre in the provision of data used in the current study. Google Earth images (© 2021 Maxar Technologies) were sourced
and analyzed from Google Earth 3D.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.ijdrr.2021.102683.

References
[1] P. Adhikari, Y. Hong, K.R. Douglas, D.B. Kirschbaum, J. Gourley, R. Adler, G.R. Brakenridge, A digitized global flood inventory (1998–2008): compilation and
preliminary results, Nat. Hazards 55 (2) (2010) 405–422.
[2] Reliefweb, Global natural disaster assessment report 2020, Available online at:, 23rd October 2021, https://reliefweb.int/sites/reliefweb.int/files/resources/
2020%20Global%20Natural%20Disaster%20Assessment%20Report-EN%282%29.pdf, 2020.

15
J.R.S. Doorga et al. International Journal of Disaster Risk Reduction 67 (2022) 102683

[3] A. Cashman, M.R. Nagdee, Impacts of climate change on settlements and infrastructure in the coastal and marine environments of Caribbean small island
developing states (SIDS), Sci. Rev. (2017) 155–173, 2017.
[4] World Bank Group, Disaster risk profile of the Republic of Mauritius, Available online at: https://www.preventionweb.net/files/52172_mauritius.pdf, 2016.
(Accessed May 2021).
[5] A. Chacowry, L.J. McEwen, K. Lynch, Recovery and resilience of communities in flood risk zones in a small island developing state: a case study from a suburban
settlement of Port Louis, Mauritius, Int. J. Dis. Risk Reduct. 28 (2018) 826–838.
[6] N. Jhingut, V. Proag, Port Louis Is Flooding, University of Mauritius Discussion Paper, 2014.
[7] A. Chacowry, Public perceptions of living with flood risk from media coverage in the small island developing state of Mauritius, Int. J. Dis. Risk Reduct. 19
(2016) 303–310.
[8] D. McClean, Mauritius invests 2% GDP in resilience, Available online at:, 13th May 2021, https://reliefweb.int/report/mauritius/mauritius-invests-2-gdp-
resilience, 2016.
[9] GoM, National multi hazard emergency alert system to go live in april 2020, Available online at:, 14th May 2021, http://www.govmu.org/English/News/Pages/
National-Multi-Hazard-Emergency-Alert-System-to-go-live-in-April-2020.aspx, 2019.
[10] GoM, Sixth national assembly: parliamentary debates, Available online at:, 14th May 2021, https://mauritiusassembly.govmu.org/Documents/Hansard/2019/
hansard0419.pdf, 2019.
[11] C.M. Mamoun, R. Nigel, S.D. Rughooputh, Wetlands’ inventory, mapping and land cover index assessment on Mauritius, Wetlands 33 (4) (2013) 585–595.
[12] V. Proag, Drain Design for Dry Feet, Journal of The Institution of Engineers Mauritius, 2013, pp. 7–18.
[13] K. Chapi, V.P. Singh, A. Shirzadi, H. Shahabi, D.T. Bui, B.T. Pham, K. Khosravi, A novel hybrid artificial intelligence approach for flood susceptibility assessment,
Environ. Model. Software 95 (2017) 229–245.
[14] Z. Tang, H. Zhang, S. Yi, Y. Xiao, Assessment of flood susceptible areas using spatially explicit, probabilistic multi-criteria decision analysis, J. Hydrol. 558
(2018) 144–158.
[15] A. Arabameri, K. Rezaei, A. Cerdà, C. Conoscenti, Z. Kalantari, A comparison of statistical methods and multi-criteria decision making to map flood hazard
susceptibility in Northern Iran, Sci. Total Environ. 660 (2019) 443–458.
[16] M. Rahman, C. Ningsheng, M.M. Islam, A. Dewan, J. Iqbal, R.M.A. Washakh, T. Shufeng, Flood susceptibility assessment in Bangladesh using machine learning
and multi-criteria decision analysis, Earth Syst. Environ. 3 (3) (2019) 585–601.
[17] A. Azareh, E. Rafiei Sardooi, B. Choubin, S. Barkhori, A. Shahdadi, J. Adamowski, S. Shamshirband, Incorporating multi-criteria decision-making and fuzzy-
value functions for flood susceptibility assessment, Geocarto Int. (2019) 1–21.
[18] C. Cetinkaya, E. Özceylan, M. Erbaş, M. Kabak, GIS-based fuzzy MCDA approach for siting refugee camp: a case study for southeastern Turkey, Int. J. Dis. Risk
Reduct. 18 (2016) 218–231.
[19] G. Papaioannou, L. Vasiliades, A. Loukas, Multi-criteria analysis framework for potential flood prone areas mapping, Water Resour. Manag. 29 (2) (2015)
399–418.
[20] K. Khosravi, H. Shahabi, B.T. Pham, J. Adamowski, A. Shirzadi, B. Pradhan, J. Dou, H.B. Ly, G. Gróf, H.L. Ho, H. Hong, A comparative assessment of flood
susceptibility modeling using multi-criteria decision-making analysis and machine learning methods, J. Hydrol. 573 (2019) 311–323.
[21] D. Rincón, U.T. Khan, C. Armenakis, Flood risk mapping using GIS and multi-criteria analysis: a greater Toronto area case study, Geosciences 8 (8) (2018) 275.
[22] G.D. Bathrellos, E. Karymbalis, H.D. Skilodimou, K. Gaki-Papanastassiou, E.A. Baltas, Urban flood hazard assessment in the basin of Athens Metropolitan city,
Greece, Environ. Earth Sci. 75 (4) (2016) 319.
[23] M.M. de Brito, M. Evers, Multi-criteria decision-making for flood risk management: a survey of the current state of the art, Nat. Hazards Earth Syst. Sci. 16 (4)
(2016) 1019–1033.
[24] V. Meyer, D. Haase, S. Scheuer, GIS-based Multicriteria Analysis as Decision Support in Flood Risk Management (No. 6/2007), UFZ Discussion Paper, 2007.
[25] Z. Allam, P. Newman, Economically incentivising smart urban regeneration. Case study of Port Louis, Mauritius, Smart Cities 1 (1) (2018) 53–74.
[26] K.K. Khedo, Real-time flood monitoring using wireless sensor networks, J. Inst. Eng. Mauritius-IEM J. 1 (2013) 59–69.
[27] ESDAC, Soil map of Mauritius (provisional classification), Available online at: https://esdac.jrc.ec.europa.eu/images/Eudasm/Africa/images/maps/download/
PDF/afr_musoil.pdf, 2021. (Accessed May 2021).
[28] GoM, Drainage report including road layout and drainage layout, Available online at:, 19th May 2021, https://environment.govmu.org/Documents/eia/
eiareports/2018/1204-smartcitypailles/ann9.pdf, 2018.
[29] M.K. Hossain, Q. Meng, A fine-scale spatial analytics of the assessment and mapping of buildings and population at different risk levels of urban flood, Land Use
Pol. 99 (2020) 104829.
[30] J.S. Cabrera, H.S. Lee, Flood-prone area assessment using GIS-based multi-criteria analysis: a case study in Davao Oriental, Philippines, Water 11 (11) (2019)
2203.
[31] D.S. Fernandez, M.A. Lutz, Urban flood hazard zoning in Tucumán Province, Argentina, using GIS and multicriteria decision analysis, Eng. Geol. 111 (1–4)
(2010) 90–98.
[32] S.H. Mahmoud, T.Y. Gan, Urbanization and climate change implications in flood risk management: developing an efficient decision support system for flood
susceptibility mapping, Sci. Total Environ. 636 (2018) 152–167.
[33] S. Hammami, L. Zouhri, D. Souissi, A. Souei, A. Zghibi, A. Marzougui, M. Dlala, Application of the GIS based multi-criteria decision analysis and analytical
hierarchy process (AHP) in the flood susceptibility mapping (Tunisia), Arabian J. Geosci. 12 (21) (2019) 1–16.
[34] K.S. Vignesh, I. Anandakumar, R. Ranjan, D. Borah, Flood vulnerability assessment using an integrated approach of multi-criteria decision-making model and
geospatial techniques, Model. Earth Syst. Environ. 7 (2) (2021) 767–781.
[35] M. Stieglitz, D. Rind, J. Famiglietti, C. Rosenzweig, An efficient approach to modeling the topographic control of surface hydrology for regional and global
climate modeling, J. Clim. 10 (1) (1997) 118–137.
[36] W. Weeks, G. Witheridge, E. Rigby, A. Barthelmess, O‘Loughlin, G., Project 11: Blockage of Hydraulic Structures, Engineers Australia, Water Engineering, 2013,
p. 11.
[37] D. Butler, A. Kokkalidou, C.K. Makropoulos, Supporting the siting of new urban developments for integrated urban water resource management, in: Integrated
Urban Water Resources Management, Springer, Dordrecht, 2006, pp. 19–34.
[38] M.M. Islam, K. Sado, Flood hazard assessment in Bangladesh using NOAA AVHRR data with geographical information system, Hydrol. Process. 14 (3) (2000)
605–620.
[39] O. Rahmati, H. Zeinivand, M. Besharat, Flood hazard zoning in Yasooj region, Iran, using GIS and multi-criteria decision analysis, Geomatics, Nat. Hazards Risk
7 (3) (2016) 1000–1017.
[40] K. Khosravi, E. Nohani, E. Maroufinia, H.R. Pourghasemi, A GIS-based flood susceptibility assessment and its mapping in Iran: a comparison between frequency
ratio and weights-of-evidence bivariate statistical models with multi-criteria decision-making technique, Nat. Hazards 83 (2) (2016) 947–987.
[41] FAO, Land resources and agricultural suitability map of Mauritius, Available online at:, 1st June 2021, https://edepot.wur.nl/481794, 1973.
[42] S. Lee, B. Pradhan, Probabilistic landslide hazards and risk mapping on Penang Island, Malaysia, J. Earth Syst. Sci. 115 (6) (2006) 661–672.
[43] O. Mtapuri, E. Dube, J. Matunhu, Flooding and poverty: two interrelated social problems impacting rural development in Tsholotsho district of Matabeleland
North province in Zimbabwe, Jàmbá: J. Dis. Risk Stud. 10 (1) (2018) 1–7.
[44] SM, Poverty map 2006/07Analytical paper on poverty maps 2001/02 & 2006/07, Available online at:, 1st June 2021, https://statsmauritius.govmu.org/Pages/
Statistics/By_Subject/Poverty/Poverty_Map.aspx, 2007.
[45] X. Hu, R. Pant, J.W. Hall, S. Surminski, J. Huang, Multi-scale assessment of the economic impacts of flooding: evidence from firm to macro-level analysis in the
Chinese manufacturing sector, Sustainability 11 (7) (2019) 1933.
[46] X.L. Yang, J.H. Ding, H. Hou, Application of a triangular fuzzy AHP approach for flood risk evaluation and response measures analysis, Nat. Hazards 68 (2)
(2013) 657–674.

16
J.R.S. Doorga et al. International Journal of Disaster Risk Reduction 67 (2022) 102683

[47] T.L. Saaty, A scaling method for priorities in hierarchical structures, J. Math. Psychol. 15 (3) (1977) 234–281.
[48] Y.O. Ouma, R. Tateishi, Urban flood vulnerability and risk mapping using integrated multi-parametric AHP and GIS: methodological overview and case study
assessment, Water 6 (6) (2014) 1515–1545.
[49] L. Gigović, D. Pamučar, Z. Bajić, S. Drobnjak, Application of GIS-interval rough AHP methodology for flood hazard mapping in urban areas, Water 9 (6) (2017)
360.
[50] H.M. Lyu, W.J. Sun, S.L. Shen, A. Arulrajah, Flood risk assessment in metro systems of mega-cities using a GIS-based modeling approach, Sci. Total Environ. 626
(2018) 1012–1025.
[51] R. Umagandhi, A.S. Kumar, Search query recommendations in web information retrieval using query logs, in: Web Usage Mining Techniques and Applications
across Industries, IGI Global, 2017, pp. 199–222.
[52] WRU, GIBB, Watershed Management; 4 Report for the Land Drainage Study of Mauritius Island, GIBB (Mauritius) Ltd for Water Resources Unit (WRU),
Mauritius, 2002.
[53] S. Rushdie, the River CAMP, Le Malaise Créole: Ethnic Identity in Mauritius 26 (2006) 135.
[54] CADRI, Disaster risk management : a capacity diagnosis, Available online at:, 25th October 2021, https://www.cadri.net/system/files/2021-06/Mauritius%
20Capacity%20Diagnosis%20Report%20-%202020%20-%20Final.pdf, 2020.
[55] Le Mauricien, District No. 4: twenty-two areas prone to flooding – Le mauricien, Available online at:, 25th October 2021, https://mauritiushindinews.com/
hindi-newspaper-in-mauritius/district-no-4-twenty-two-areas-prone-to-flooding-le-mauricien/, 2021.
[56] Lexpress, Port-Louis: la situation de 322 familles de squatters régularisée, Available online at:, 9th June 2021, https://www.lexpress.mu/article/279775/port-
louis-situation-322-familles-squatters-regularisee, 2016.
[57] E. Ozsen, Deadly flash flood event in Mauritius on 30th March 2013 – how SUDS can help reduce flooding risks, Available online at:, 27th October 2021,
https://emmaozsen.wordpress.com/tag/port-louis/, 2013.
[58] Defimedia, Inondations : la capitale à genoux après trois heures de pluies battantes, Available online at:, 27th October 2021, https://defimedia.info/
inondations-la-capitale-genoux-apres-trois-heures-de-pluies-battantes, 2019.
[59] Le Mauricien, DRAME—SAMEDI 30 MARS 2013: Le “flash flood” fait 11 victimes à Port-Louis, Available online at:, 27th October 2021, http://www.
lemauricien.com/actualites/societe/drame-samedi-30-mars-2013-flash-flood-fait-11-victimes-port-louis/38801/, 2015.
[60] E. Bertacchini, R. Sultan, Valuing urban cultural heritage in African countries: a contingent valuation study of historic buildings in Port Louis, Mauritius, J. Afr.
Econ. 29 (2) (2020) 192–213.
[61] EY, Economic impact assessment of the Mauritian horse racing industry, Available online at:, 10th June 2021, https://media.mauritiusturfclub.com/images/
stories/pdf/mtc_eia.pdf, 2019.
[62] Inside News, Budget 2021-2022: rs 65 milliards injectées des projets prioritaires sur les trois prochaines années, Available online at:, 14th June 2021, https://
inside.news/2021/06/12/budget-2021-2022-rs-65-milliards-injectees-des-projets-prioritaires-sur-les-trois-prochaines-annees-%e2%80%a8/, 2021.
[63] K.W. Mak, J. Poon, K. Lau, E. Lo, K. Chan, Computational hydraulics aspects of the drainage system planning for Happy Valley flood protection scheme, Hong
Kong, in: IWA World Water Congress and Exhibition, 2012.
[64] N. Carmon, U. Shamir, Water-sensitive planning: integrating water considerations into urban and regional planning, Water Environ. J. 24 (3) (2010) 181–191.
[65] W.H. Luk, E. Cheng, J. Johnson, June. Happy Valley underground stormwater storage scheme, Hong Kong, in: Proceedings of the Institution of Civil Engineers-
Civil Engineering, vol. 169, Thomas Telford Ltd, 2016, pp. 169–175. No. 4.
[66] NDRRMC, Emergency shelters – port Louis, Available online at:, 15th June 2021, https://ndrrmc.govmu.org/Pages/shelters.aspx, 2021.
[67] UNDP, Mauritius Marshall Plan aims to eradicate poverty and exclusion, Available online at:, 15th June 2021, https://www.un.org/africarenewal/news/
mauritius-marshall-plan-aims-eradicate-poverty-and-exclusion, 2017.
[68] NAO, Performance Audit Office: government response to mitigate the impact of flooding, Available online at:, 16th June 2021, https://nao.govmu.org/
Documents/Reports/P03_Imp_of_Flooding.pdf, 2019.
[69] EDB, Government of Mauritius budget measures, Available online at:, 16th June 2021, https://news.edbmauritius.org/budget-newsletter/pdf/2021_
22Annexbudgetspeech.pdf, 2021.
[70] DPP, Press release from the director of public prosecutions, Available online at:, 16th June 2021, https://weluvmu.com/blog/2015/01/08/press-release-from-
the-director-of-public-prosecutions/, 2015.

17