The current study is supported by the secondary data (table. 1). Secondary data is gathered through the internet, published reports, and the relevant public departments. Fig. 2 graphically depicts the steps taken to determine which locations are most susceptible to flooding.
Table 1 Descriptive details of the data collected
Data Source
|
Specifications
|
Output (Flood Influencing Factor)
|
IMD, Pune
|
Hourly Data
|
Rainfall
|
SRTM (Shuttle Radar Topography Mission) DEM (Digital Elevation Model)
|
Spatial Resolution: 30 m
|
Slope
|
Satellite Image
Landsat 5 and 8
|
TM & OLI/TIRS Landsat TM and OLI, Spatial Resolution: 30mt.
|
Land Use Land Cover (LULC)
|
BRIMSTOWAD-II (Brihanmumbai Storm Water Disposal System) Draft Master Plan Maps of Storm water
drainage & Sewers
|
Scale - 1: 50000,
Year - 2014
|
Vicinity to sewers and storm water drainage
|
Survey of India OSM Sheet, SRTM DEM
|
Number - E43A/16(47A/16)
Scale - 1:50,000,
Spatial Resolution: 30 m
|
Vicinity to Natural Drainage
|
Satellite Image
Landsat 5 and 8
|
NDVI; Spatial Resolution: 30mt.
|
Vegetation
|
National Bureau of Soil Survey
|
Maharashtra
Scale – 1:50000, Year - 1996
|
Soil
|
Municipal Corporation of Greater Mumbai (MCGM)
|
Ward-wise flood locations
|
Map of waterlogging spots
|
3.1 Producing Maps of Flood Influencing Factors
For each component, GIS maps are initially constructed using a standard geo-referencing approach. Rainfall and soil maps were prepared from the data obtained from IMD, Pune and National Bureau of Soil Survey, Maharashtra. By using surface-slope tool in ArcGIS, slope in percentage for the study area was retrieved from the computed SRTM DEM and SOI OSM sheet, which was later modified, using high resolution satellite images, and again used to digitize the map of natural drainage. The Euclidean distance measure in ARCGIS was used to calculate the vicinity for a distance equivalent to 1000 metres for natural drainage. The vegetation map is produced using TM Landsat 8 (30 mt.) image for the year 2020 and Normalized Difference Vegetation Index (NDVI) is performed using Erdas Imagine 2010 software as NDVI= (NIR-VIS)/(NIR+VIS). Using Landsat 8 (OLI/TIRS 30 mt.) multi-spectral data, the map of LULC for the study area was retrieved in ARCGIS software by employing supervised classification. Finally, the maps of sewers and storm water drainage from the BRIMSTOWAD-II Draft Plan were used to retrieve the map of artificial drainage. It is a component of the storm water drains project being undertaken in 2014 by the office of storm water drains under department of disaster management, MCGM. Using the Euclidean distance measure, the vicinity for a distance equivalent to 500 mt. was calculated. All maps are transformed to grid-based integer raster format with the same pixel size of 30 mt. for each parameter. At last, all the maps were overlaid to produce a combined map of flood vulnerable zones. Also, a ward-wise map of 234 flood locations is prepared from the data gathered from MCGM and then manually assesses the relative importance of each parameter against the flood locations. MCA is used to create and combine spatial data for characterizing the causative aspects in order to determine the vulnerability of flooding. In GIS context, the Weighted Linear Combination (WLC) Approach was employed to implement the AHP Pairwise Comparison Method.
3.2 Description and Order of Influencing Parameters
Each aspect that is taken into account is ranked according to the preference of decision makers. Each component is rated according to the expected significance influence on floods in order to establish criterion scores for each sub-class category. These factors received an inverted ranking. Each sub-class is ranked 1–5 in decreasing order of impact based on reviewed literature and knowledge, where 5 represents high vulnerability to floods and 1 represents low vulnerability. Table 2 displays the grading scheme. The current study makes the assumption that the areas that are most susceptible to flooding depend on a variety of variables, including slope, amount of rainfall, vegetation, soil types, LULC practices, natural and artificial drainage network. As a result, depending on these variables, flood can vary considerably over time and location. The following seven parameters were used in the current study, and each of them is shown and saved in a distinct map with order of their sub-category is shown in fig. 3 and table 2.
Table 2 Ranking of Flood Parameters and their Sub Categories
Vulnerability Parameters
|
Sub-category of Parameters
|
Ranking
|
1. Rainfall
|
>2300.01
|
5
|
|
2200.01-2300
|
4
|
|
2100.01-2200
|
3
|
|
2000.01-2100
|
2
|
|
<2000
|
1
|
2. Slope
|
Very Gentle [below 5]
|
5
|
|
Gentle [5.01-10.00]
|
4
|
|
Moderate [10.01-15.00]
|
3
|
|
Steep [15.01-20.00]
|
2
|
|
Very Steep [20.01 & above]
|
1
|
3. LULC
|
Built-up Area
|
5
|
|
Open Land
|
4
|
|
Cultivated Land
|
3
|
|
Water body
|
2
|
|
Vegetation Cover
|
1
|
4. Vicinity to Sewers & Storm Water Drainage
|
0-125 mt.
|
5
|
|
125-250 mt.
|
4
|
|
250-375 mt.
|
3
|
|
375-500 mt.
|
2
|
|
>500mt.
|
1
|
5. Vicinity to Natural Drainage
|
0-250 mt.
|
5
|
|
250-500 mt.
|
4
|
|
500-750 mt.
|
3
|
|
750-1000 mt.
|
2
|
|
>1000 mt.
|
1
|
6. Vegetation
|
Lowest dense vegetation cover
|
5
|
|
Lower dense Vegetation Cover
|
4
|
|
Dense Vegetation Cover
|
3
|
|
Higher Dense Vegetation Cover
|
2
|
|
Highest dense Vegetation Cover
|
1
|
7. Soil
|
Settlement Coastal Alluvium
|
5
|
|
Mud Marsh
|
4
|
|
Vertic Halaquepts
|
3
|
|
Vertic Ustrepepts
|
2
|
|
Typic Ustorthents
|
1
|
3.2.1 Rainfall
Rainfall is the primary hydrological component that is most frequently employed in studies of floods. Rainfall is the term used to describe the dispersion of liquid droplets over space and time, which regulates the surface runoff (Goswami et al. 2006). Since areas with higher rainfall than the annual average are more likely to experience flooding, high rainfall amounts are a marker of substantial flood susceptibility. As a result, a category weight of 5 is allocated to heavy rainfall zone and 1 is allocated to relatively low rainfall zone (table. 2).
3.2.2 Slope
The slope is the most important aspect in hydrology since it directly affects the surface runoff and floods. Since sites of low-elevation often have a gentle or level slope, they are more susceptible to flooding and water logging because steep slopes generates huge velocity of runoff than flat or gentle slopes and dispose of storm runoff more quickly (Altaf et al. 2013). Runoff from a level or gently sloping land is accumulated and released gradually over time (Tehrany and Kumar 2018). In contrast to high gradient slopes, low gradient slopes even more susceptible to flooding. Historically, the study area was an archipelago of seven islands that has been reclaimed and established as a land of concrete slabs over a span of five centuries. Additionally, the steeper slope was exploited to provide flat homes for a large number of migrants. As a result, places with very gentle slopes were assigned a class rating of 5, whereas locations with high relief were given lower ranking i.e., 1.
3.3.3 LULC
Recognizing the activities taking on in a location and the various categories of LULC being impacted by recurrent floods is crucial for vulnerability mapping. Due to their significant use of impermeable surfaces, urban areas are impacted by storm water runoff (Fernandez and Lutz 2010). Built-up area dominates the LULC category in the study area. Slums or nucleated communities are the main components of dense built-up areas, which are primarily found in the city's central and Southern parts. Evidently, areas with dense built-up space are at a larger risk of flooding than areas with less built-up land cover. Therefore, rankings are allocated as shown in table 2 based on the kind of land use and its susceptibility to floods.
3.3.4 Vicinity to sewers and storm water drainage
With growing pollution and a lack of concern for it, the sewerage system has emerged as a crucial component and responsibility of the city administration. Sewers are man-made drains that are used to move sewage from homes to disposal sites. The Storm Water Drains (SWD) is specialized man-made drains that assist in moving and draining extra storm runoff from the city to the countryside and thus minimizes floods. The water-logging in most parts of a megacity like Mumbai is caused by the overflow from these drains, which are primarily ineffective at holding or draining water, therefore it is crucial to take this into account while analysing the vulnerability of flooding.
3.3.5 Vicinity to natural drainage
The intensity of flooding was thought to be impacted by vicinity to the natural drainage, implying that the area around some stream or river is quite vulnerable to floods. Here, all streams and rivers are considered to be a part of natural drainage.
3.3.6 Vegetation Cover
Vegetation cover is the parameter that has the significant impact on flood vulnerability. Flood susceptibility may grow as vegetation cover declines. This connection is the basis for the widespread inclusion of tree cover and land use in studies of flood susceptibility (Young et al. 2009). Insufficient tree cover under barren or open land makes the area highly vulnerable to flooding; a maximum class weight of 5 is allocated to it, with subsequent classes receiving less weight in the sequence of increasing the area under tree cover.
3.3.7 Soil Type
Since soil properties, particularly those that are more likely to erode, are more vulnerable to floods because the rate of permeability are depends on soil characteristic of a region. Therefore, soil type was chosen as an influencing parameter (Chung et al. 2011). The study area has two main soil types’ viz vertic halaquepts and coastal alluvium (Soil group-Inceptisols). These soils are very fine, slightly deep, poorly drained and moderately salinized. They are located on very gently sloping areas in residual hills. These soils generally equate to black and laterite in the context of Indian soil classification (Pal 2013).
4.1 Application of AHP to Evaluate the Parameters Weight
Determining the weights for flood influential parameters is a complex problem that involves different criterion functions. If such a scenario is not handled with a reasonable and well-processed methodology, often results in miscalculation of the facts. The MCE technique has the potential to logically resolve this matter regarding multiple criteria. The AHP technique which Saaty devised was applied in the current investigation (Saaty 1977). With the use of a preference matrix, in which all recognized relevant criteria are contrasted against one another with replicable preference parameters, AHP is a very well-known and widely used statistical method to determine the necessary weights of each parameter. A pair-wise evaluation matrix, a measure to represent the proportional priority among the components, examines all parameters that are thought to be essential for a decision against one another. As a result, each parameter needs to be given a quantitative score conveying a decision of the importance of one variable compared to another. A scale for comparison with scores ranging from 1 to 9 which represent the intensity of importance was proposed by Saaty and Vargas (1991). Since it has been recognized through psychological studies, a person cannot compare more than 7 ± 2 variables at once. One represents "equal importance," whereas nine represents variable that are "extremely important" in comparison to other criteria (table. 3).
4.2 Construction of a Pair-wise Comparison Matrix
A pair-wise comparison matrix of order 7 is shown in table 4 and compares seven parameters (C1, C2, C3, C4, C5, C6, and C7). When parameter C1 and parameter C2 are directly compared, parameter C1 is seen as equal to moderate relevance and the remaining parameter is allocated the same relative weight. The reciprocal of 1/4, or 0.25, is immediately applied to the transposed position.
4.3 Standardized Pair-Wise Comparison Matrix
Intensity of Importance
|
Explanation
|
1
|
Equal Importance
|
2
|
Equal to moderate Importance
|
3
|
Moderate Importance
|
4
|
Moderate to Strong Importance
|
5
|
Strong Importance
|
6
|
Strong to very Strong Importance
|
7
|
Very Strong Importance
|
8
|
Very to Extremely strong Importance
|
9
|
Extreme Importance
|
Corresponding
|
Values for Inverse Comparison
|
The stated preference scores are combined in the following phase to arrive at a quantitative score that represents the weights of the parameters. As a result, Eigen scores and vectors of square preference matrix, that disclose key information about patterns in the data matrix, are computed. Seven eigen scores are provided by the square matrix of order seven mentioned below, which can be used to calculate seven eigen vectors, each of which has seven vector components. Since this Eigen vector provides enough information to show by its eigen vector components - the relative emphases of the parameters being investigated, it is viewed as adequate to compute only its eigen vector deriving from the biggest eigen score (Saaty and Vargas 1991). The pair-wise matrix is standardized, and the parameter weights are represented by the standardized matrix's eigen scores, which are produced as shown in table
Table 3 A sampled comparison scale
Intensity of Importance
|
Explanation
|
1
|
Equal Importance
|
2
|
Equal to moderate Importance
|
3
|
Moderate Importance
|
4
|
Moderate to Strong Importance
|
5
|
Strong Importance
|
6
|
Strong to very Strong Importance
|
7
|
Very Strong Importance
|
8
|
Very to Extremely strong Importance
|
9
|
Extreme Importance
|
Corresponding
|
Values for Inverse Comparison
|
Table 4 Construction of a Pair-Wise Comparison Matrix
Table 5 Standardized Pair-Wise Matrix
Parameters
|
|
Rainfall
|
Slope
|
LULC
|
Vicinity to Sewers & Storm Water Drainage
|
Vicinity to Natural Drainage
|
Vegetation
|
Soil
|
Total
|
Priority Vector
|
Weight (%)
|
|
|
C1
|
C2
|
C3
|
C4
|
C5
|
C6
|
C7
|
|
|
|
Rainfall
|
C1
|
0.33
|
0.41
|
0.31
|
0.34
|
0.27
|
0.20
|
0.19
|
2.06
|
0.29
|
29.42
|
Slope
|
C2
|
0.17
|
0.21
|
0.31
|
0.23
|
0.21
|
0.20
|
0.15
|
1.47
|
0.21
|
20.96
|
LULC
|
C3
|
0.17
|
0.10
|
0.16
|
0.23
|
0.14
|
0.25
|
0.19
|
1.23
|
0.18
|
17.52
|
Vicinity to Sewers & Storm Water Drainage
|
C4
|
0.11
|
0.10
|
0.08
|
0.11
|
0.27
|
0.15
|
0.15
|
0.98
|
0.14
|
13.99
|
Vicinity to Natural Drainage
|
C5
|
0.08
|
0.07
|
0.08
|
0.03
|
0.07
|
0.15
|
0.15
|
0.63
|
0.09
|
8.97
|
Vegetation
|
C6
|
0.08
|
0.05
|
0.03
|
0.04
|
0.02
|
0.05
|
0.12
|
0.39
|
0.06
|
5.58
|
Soil
|
C7
|
0.07
|
0.05
|
0.03
|
0.03
|
0.02
|
0.02
|
0.04
|
0.25
|
0.04
|
3.56
|
Total
|
|
1.00
|
1.00
|
1.00
|
1.00
|
1.00
|
1.00
|
1.00
|
7.00
|
1.00
|
100.00
|
4.4 Computing Consistency Ratio
To evaluate how reliable the assessments have been in comparison to sizable samples of merely random assessments, the consistency ratio (CR) is determined at this point. The AHP always permits a certain degree of consistency, but it shouldn't go beyond a certain point. The consistency ratio (CR), that evaluates the level of consistency, is determined using the Random inconsistency index (RI) (table 6) created by Saaty (1980). If the CR is substantially greater than 0.1, assessments are unreliable because they are too close to randomness. If the CR value is less than or equivalent to 0.1, the inconsistency is acceptable, or else the pair-wise comparison may be altered (Saaty, 1980). The weights are therefore acceptable.
Table 6 Different size matrices with random indices
n
|
2
|
3
|
4
|
5
|
6
|
7
|
8
|
9
|
10
|
RI
|
0.00
|
0.52
|
0.90
|
1.12
|
1.24
|
1.32
|
1.41
|
1.45
|
1.49
|
Source: Saaty and Vargas 1991.
Table 7 Computing the Consistency Ratio (CR)
Parameters
|
|
C1
|
C2
|
C3
|
C4
|
C5
|
C6
|
C7
|
Criteria Weight
|
Criteria weight
|
Consistency Vector
|
Rainfall
|
C1
|
1
|
2
|
2
|
3
|
4
|
4
|
5
|
0.29
|
2.24
|
7.63
|
Slope
|
C2
|
0.50
|
1
|
2
|
2
|
3
|
4
|
4
|
0.21
|
1.62
|
7.74
|
LULC
|
C3
|
0.50
|
0.5
|
1
|
2
|
2
|
5
|
5
|
0.18
|
1.34
|
7.67
|
Vicinity to Sewers & Storm Water Drainage
|
C4
|
0.33
|
0.50
|
0.50
|
1
|
4
|
3
|
4
|
0.14
|
1.10
|
7.85
|
Vicinity to Natural Drainage
|
C5
|
0.25
|
0.33
|
0.50
|
0.25
|
1
|
3
|
4
|
0.09
|
0.61
|
6.75
|
Vegetation
|
C6
|
0.25
|
0.25
|
0.20
|
0.33
|
0.33
|
1
|
3
|
0.06
|
0.39
|
7.04
|
Soil
|
C7
|
0.20
|
0.25
|
0.20
|
0.25
|
0.25
|
0.33
|
1
|
0.04
|
0.26
|
7.24
|
Total
|
|
1.00
|
1.00
|
1.00
|
1.00
|
1.00
|
1.00
|
1.00
|
1.00
|
|
51.92
|