GIS-Based Risk Assessment of Building Vulnerability in Flood Zones of Naic, Cavite, Philippines Using AHP and TOPSIS

Abstract

Floods pose significant challenges globally, particularly in coastal regions like the Philippines, which are vulnerable to typhoons and subsequent inundations. This study focuses on Naic city in Cavite, Philippines, using Geographic Information Systems (GIS) to develop flood risk maps employing two Multi-Criteria Decision-Making (MCDM) methods including Analytical Hierarchy Process (AHP) and Technique for Order Preference by Similarity to Ideal Solution (TOPSIS). These maps integrate hazard, vulnerability, and exposure assessments to identify structures most vulnerable to flooding. Weight assignments in the study are derived from a literature review and expert opinions, reflecting the Philippines’ flood-prone geography and historical data. Structural attributes, categorized on a low to very high scale, were assessed based on field survey data from 555 buildings. AHP categorized 91.3% of buildings as moderate to very high risk, whereas TOPSIS placed 68% in this category, underscoring methodological disparities in data handling and assumptions. This research enhances understanding of flood threats and offers a decision-making framework for resilient flood risk management strategies. Identifying vulnerable buildings aims to support informed urban planning and disaster preparedness in flood-prone areas, thereby mitigating potential property, infrastructure, and livelihood damage.

1. Introduction

Natural disasters are becoming more common due to different factors, including changes in climate conditions, specifically global warming. Heavy rain, rapid snow melt, or a storm tide from a tropical cyclone or tsunami in coastal areas often cause floods, making them one of the most destructive natural disasters on the planet. Floods are natural phenomena that occur under various circumstances and are affected by multiple meteorological, hydrological, and human factors. Floods have destructive power and can destroy structures that cannot withstand the force of the water, such as bridges, houses, trees, cars, and even human life. As a result, flood studies and prevention plans have grown in recent years. The study growth also primarily focuses on metropolitan cities and their urban planning. The villages need to be more developed and need more attention in case of floods. However, damage to the land and structures in rural communities is overlooked and puts them in an unsafe position. Also, despite the importance of assessing the risks associated with building attributes, most risk assessment studies have focused on environmental factors, with little focus on building attributes. Also, a rapid visual evaluation technique is required to reduce the burden on the time-consuming, detailed, and expensive quantitative and analytical methods. Floods kill nearly 100 people in the United States each year and cause damage worth $7.5 billion. In the last century, the Yellow River valley in China has experienced some of the world’s worst floods. The 1931 Yellow River flood was one of the most devastating natural disasters ever recorded, killing nearly a million people and displacing millions more [1]. Owing to its geographical location, the Philippines finds itself susceptible to devastating floods.

Typhoons create floods mainly through intense rainfall and typhoon waves. The heavy and rapid precipitation overwhelms rivers and drainage systems, leading to overall flooding, while the powerful winds drive seawater onto coastal areas, elevating the flooding impact. On average, 20 typhoons affect the Philippines yearly [2], ranging in severity from annually recurring moderate up to extreme events such as the super-typhoon Haiyan. Typhoon Haiyan (local name Yolanda) in November 2013 and Typhoon Goni (local name Rolly) [3] in late October 2020 were the two strongest typhoons to make landfall in the country, with 1-min sustained winds of 315 km/h (195 mph). As of this date, Typhoon Haiyan [4] is also the deadliest Philippine typhoon during this period, killing 6300 people. Other notable storms in the Philippines include Typhoon Ketsana (Ondoy) [5] in September 2009, which became the most destructive tropical cyclone to hit Manila, and Typhoon Bopha (Pablo) [6] in December 2012, which became the strongest typhoon to hit Mindanao on record. Between 2008 and 2013, the Philippines experienced five of its costliest typhoons, resulting in a cumulative damage of $138 million. This indicates a lack of tools to aid in developing appropriate mitigation measures and programs in the country to reduce damage caused by typhoons and flooding [7].

Table 1. Most destructive and Costliest Philippine typhoons in years.

Typhoon Name (Local Name)	Year	Damage Cost (USD $)	Reference
Pepeng (Parma)	2009	581 million	[24]
Pedring (Nesat)	2011	356 million	[6]
Pablo (Bopha)	2012	1.06 billion	[6]
Yolanda (Haiyan)	2013	2.2 billion	[4]
Glenda (Rammasun)	2014	771 million	[25]
Lando (Koppu)	2015	313 million	[3]
Ompong (Mangkhut)	2018	627 million	[26]
Ulysses (Vamco)	2020	418 million	[3]
Rolly (Goni)	2020	369 million	[3]
Odette (Rai)	2021	1.02 billion	[3]

shows the costliest typhoon events in the country and the damage caused by those typhoons in the country in billions. Building risk assessment, particularly in natural disasters, can benefit from decision-making tools such as MCDM, Fuzzy Logic, and Machine Learning (ML) techniques. Many research works and studies [8,9,10,11,12,13,14,15,16,17] discovered the role of decision making and ML technologies in advancing seismic risk evaluation, damage assessment, and emergency response measures after exploring the landscape of structural damage assessments post-natural hazards. ML models are compelling but can not be feasible and certainly not usable under constrained financial situations and multifaceted decision-making problems such as floods. Therefore, this brings out the need to consider MCDM techniques more useful. Moreover, MCDM techniques use a region’s geographical, socio-economic, and building attributes data to predict the building risks. Several studies for flood risk assessment (FRA) employed different MCDM approaches. The most common and popular MCDM approach is AHP utilizing GIS [18,19,20]. Multiple studies on TOPSIS [21,22,23] prove it to be a valuable tool for decision support in FRA. All these preceding studies help determine the flood risk in the areas and not the structural resilience of the buildings. Only a few studies include the building FRA, such as [7,19], by considering environmental, building attributes, and socio-economic factors. Such studies can help the government utilize the results and plan to educate the people on improving the structural integrity of their houses.

Geographic Information System (GIS) is a technology used to create, manage, analyze, and map all types of location-based data; it links data to a map and integrates location data with descriptive information. Hence, it is helpful in FRA to create and visualize all spatial information of field data. Using GIS, planning decisions can be optimized, emergency and disaster response in a better way, environmental management can be improved, and business operations and marketing can be bettered [27]. ArcMap is one of two primary desktop GIS application components of the ArcGIS suite developed by Esri. Using ArcMap as a supporting framework in this research provides spatial analysis and visualization that is vital in determining the risk to structures within the flood-prone area using MCDM techniques. The flexibility and functionality of ArcMap provided quality and quantity of risk assessment depth as a precise value for dealing with complex geographical problems.

The flood risk in this study is based on three closely linked factors that determine the effect of flood events. Equation (1) shows that the hazard, exposure, and vulnerability underline the multi-dimensional character of the flood risk. The equation, then, explains the potential flood risk through the number of physical properties of floods (the hazard) present in the exposed area along with the material and human values within that area of flooding (the exposure) and the prospect that said damage will be incurred (the vulnerability) [28].

$$\mathrm{Flood}\mathrm{Risk}=\mathrm{Hazard}\times \mathrm{Exposure}\times \mathrm{Vulnerability}$$

(1)

The main objective of this study is to develop an in-depth qualitative flood risk index map for the town of Naic using MCDM techniques, including AHP and TOPSIS, using GIS. The analysis will identify the pros and cons of the MCDM technique, allowing us to judge suitable techniques based on the requirements. The integration of MCDM techniques with GIS technology will be the basis of the research to develop a comprehensive assessment of structures in flood-prone areas for risk reduction from floods. This study fills the gap in disaster management and further enriches the approach to building FRA, which integrates precise GIS information. The model would also guide policy-making and strategic planning by instituting a risk assessment in sensitive regions. The study developed a strategic framework for implementing implementation that protects lives and property by advancing risk identification and the most effective mitigation strategies. This study determines which MCDM technique is suitable for the type of dataset available and the specific risk assessment needs. This novel approach fits into the global guidance on disaster preparedness and community resilience enhancement.

2. Materials and Methods

2.1. Study Area, Geomorphological and Geological Considerations

Naic is a town in the province of Cavite, Philippines (see Figure A1), renowned for its significant cultural and historical importance. Naic municipality has a unique edge in FRA because of its geographical location, climate conditions, topographical features, and historical flood events. According to the Corona climate categorization system, Naic has a climate type I typically dry from November to April and wet the rest of the year [29]. The annual precipitation for Naic is 1780 mm, where July has the highest average rainfall while February has the lowest [30]. This study analyzes the development of building flood risks in Naic.

Figure 1 shows the covered study area of Naic with the included surveyed building. The municipality of Naic is vulnerable to natural disasters, especially during the rainy season and following typhoons, as seen by past flood catastrophes [31]. Topographical and geographic features greatly influence the spread and intensity of floods. Comprehending these dynamics facilitates flood management, strengthens mitigation attempts, and fosters community resilience. This historical study lessens the effect of floods on the people and infrastructure of Naic by forecasting future events.

2.1.1. Geological Framework

In this study, the focus was on developing flood risk maps, with hydraulic models for hazard assessment based on high-resolution LiDAR data to represent the terrain. Figure 2 shows the LiDAR data depicting the terrain of the project site, located downstream near the coastal area. The terrain data derived from LiDAR provided a detailed digital elevation model, which was essential for simulating water flow and flood inundation across the watershed. Data of this level of accuracy helps create near-realistic flood hazard scenarios [32]. The site has an average elevation ranging from 3 to 10 m above mean sea level (MSL), making it vulnerable to flooding risks.

2.1.2. Use of Geomorphological Parameters

The Figure 3 shows the delineated watershed for the Labac River. This area catches rainfall, which eventually reaches the downstream region where Naic is located. The properties of the area greatly affect the accumulation of water downstream [33]. The hydrologic model utilized various spatial datasets, including land cover and land use information, soil maps, elevation data, slope gradients, and river characteristics such as length and size, to accurately define the geomorphological features of the watershed. These datasets played a crucial role in determining key parameters that influence water flow within the watershed, including the roughness coefficient, infiltration rate, and time of concentration for different sub-basins. By incorporating these factors, the model was able to simulate water movement across the landscape, providing a comprehensive understanding of the watershed’s hydrological behavior and its response to rainfall and other environmental conditions. This detailed analysis is essential for assessing flood risks, water availability, and overall watershed management.

2.1.3. Drainage Basins and Their Role in Flood Events

The Labac River Basin, the focus of this study, was delineated into several sub-basins for hydrological modeling. These sub-basins were identified based on terrain data, following the flow of water from higher to lower elevations. The hydrological assessment for this study consists of two components. The first component involves a rainfall-runoff analysis, which addresses the basic water balance by considering key parameters that influence runoff and modeling rainfall-runoff events. The second component simulates runoff hydraulics to determine the extent of the inundated region.

The primary objective of the hydrological assessment is to quantify the flow and volume of water in the streams to evaluate their capacity and assess the potential impact of floodwaters downstream. This is achieved through a comprehensive flood modeling process that integrates both hydrologic and hydraulic models. The hydrologic model transforms rainfall data into a flood hydrograph, while the hydraulic model simulates how floodwaters move across the terrain. The outputs from the hydraulic modeling are then exported to GIS software for flood mapping, particularly in areas identified as vulnerable.

The Figure 4 illustrates the methodology used for flood hazard mapping, a critical component of the overall flood risk assessment conducted in this study. This approach provides a detailed understanding of flood dynamics, enabling more informed decision-making for mitigation and planning.

2.2. Methods for Assessing Flood Risk on Buildings

The study framework is depicted in Figure 5, which outlines the workflow used in this research, with a primary focus on constructing exposure models to assess flood risk. Identified parameters and their weight were derived through a literature review as shown in

Table 2. Parameters with Flood Risk Class and Percentage Weight.

Parameters	Percentage Weights (%)
Building Flood Hazard Parameters
Annual Average Rainfall	14.6
Slope	13.35
Elevation	30.07
Flood Depth	41.98
Building Flood Vulnerability Parameters
Average Income	4.55
Gender Ratio	3.04
Land Cover	11.55
Roofing Material	6.23
Flooring Material	5.91
Interior/Exterior Walling Material	4.79
Number of Floors	11.92
Types of Fencing Material	4.34
Age of Building	7.70
Total Height of Building	12.28
Distance to River	27.74
Building Flood Exposure Parameters
Building Density	43.74
Number of Buildings per Area	24.97
Use of Building	31.29

. All individual parameters have been utilized and overlayed to generate flood hazard, vulnerability, and exposure maps using two methods, AHP and TOPSIS, which were then overlayed to create a building flood risk map using raster multiplication.

In this study, the parameters and their weights were assigned using the literature by Gacu et al. [7] because of their relevance. Accurate weight determination in a FRA is needed to establish the significance of the different risk factors. The weights of parameters in the research were determined based on expert judgment, where ten experts in disaster risk reduction management participated in assessing the relevance of one parameter over another presented in a matrix using AHP. Table 2 shows the final percentage weights used in the research.

The

Table 3. Parameters and their data sources.

Parameters	Data Type	Period/Year	Source
Hazard Parameters
Annual Average Rainfall	PERSIANN-Cloud Classification System (PERSIANN-CCS)	2011–2020	CHRS [34]
Slope	Generated from DEM using slope tool in GIS	2018–2019	FRAMER Project of Mapua University
Elevation	Generated from DEM data	2018–2019	FRAMER Project of Mapua University
Flood Depth	Extracted from synthetic rainfall scenario at 100-year return period	2018–2019	FRAMER Project of Mapua University
Vulnerability Parameters
Average Income	Average income per barangay	2018–2019	Field Data
Gender Ratio	Barangay men-to-women gender ratio	2018–2019	Field Data
Land Cover	Land cover map	2018	NAMRIA
Roofing Material	Roofing material per building	2018–2019	Field data
Flooring Material	Flooring material per building	2018–2019	Field data
Walling Material	Wall material per building	2018–2019	Field data
Number of Floors	Number of floors per building	2018–2019	Field data
Type of Fencing Material	Fencing material per building	2018–2019	Field data
Age of Building	Age of building structure	2018–2019	Field data
Total Height of Building	Earth to roof height per building	2018–2019	Field data
Distance to River	Shapefile clipped from water courses (river) map	2018–2019	FRAMER Project of Mapua University
Exposure Parameters
Building Density	Area of buildings in 50 m by 50 m grid	2022	Field data
Number of Buildings	Number of buildings in 50 m by 50 m land area	2022	Field data
Use of Building	List of building types according to use	2022	Field data

Note: All Field Data is taken from FRAMER Project of Mapua University.

displays every parameter, data type, and source. The study used four parameters, average rainfall, slope elevation, and flood depth, for hazard parameters, which assess the possibility and severity of floods in an area affecting buildings. Vulnerability parameters refer to the properties of exposed entities and the effectiveness of support and protection mechanisms. The following eleven vulnerability factors were taken into consideration when assessing a building’s vulnerability: average income, gender ratio, land cover/use, roofing, flooring, and walling materials, number of floors, types of fencing material, building age, building height, and distance to river bodies. The exposure factor quantifies how inhabitants, property, and infrastructural systems are found within flood-prone areas. It includes building density, number of buildings, and use of buildings.

2.3. Mapping of Individual Risk Parameters

Data collection of all parameters identified through the literature review were transformed into spatial maps. Data were collected from field surveys from earlier research on the FRAMER project of Mapua University and from different government agencies (CHRS for rainfall and NAMRIA for land cover) for parameters. All spatial maps have a 1:30,000 scale for improved visualization, and cell sizes (x,y) of 0.99982856 and 0.99982856 are comparable to those created using DEM data. To establish the risk level of each building or structure against floods, 555 buildings or structures were evaluated, and their geographical analyses were done. Figure 6, Figure 7 and Figure 8 show the maps of respective individual parameters generated with the help of GIS. All these maps were reclassified to risk levels varying on the scale of 1–5 as shown in

Table A1. Reclassified Values of Flood Hazard Parameters.

Parameter	Categories	Reclassified Value
Flood Depth	0.00100279–0.5	1
	0.5–1	2
	1.000000001–2	3
	2.000000001–3	4
	>3	5
Annual Average Rainfall	2163.01–2201.53	1
	2201.54–2241.68	2
	2241.69–2282.66	3
	2282.67–2322	4
	2322.01–2371.99	5
Slope	0–3	5
	3.001–8	4
	8.001–18	3
	18.01–30	2
	>30.01	1
Elevation	−2.2268–6.248	5
	6.2481–10.485	4
	10.486–14.252	3
	14.253–18.019	2
	>18.02	1

,

Table A2. Reclassified Values of Flood Exposure Parameters.

Parameter	Categories	Reclassified Value
Type of Building	Residential	4
	Commercial	3
	Institutional	2
	Recreational	1
Building Density	0.001–16.201	5
	16.202–26.56	4
	26.561–35.855	3
	35.856–45.682	2
	45.683–67.727	1
Number of Buildings	0.001–9	1
	9.001–16	2
	16.001–23	3
	23.001–31	4
	31.001–43	5

and

Table A4. Reclassified Values of Flood Vulnerability Parameters.

Parameter	Categories	Reclassified Values
Average Income	Less than 10 k	4
	10 k to 25 k	3
	Above 25 k to 40 k	2
	40 k and above	1
Gender Ratio	0.25	4
	0.5	3
	0.75	2
	1	1
Land Cover	Cultivated area	1
	Built-up	2
Roofing Material	Tarp	5
	Metal	1
	Concrete	1
	Liha	4
	Thatch	5
	Mix Materials	2
	Wood	3
Flooring Material	Soil	4
	Concrete	1
	Mix Materials	3
	Wood	2
	Dirt	5
Walling Material	Wood	3
	Concrete	1
	Mix Materials	4
	Metal	2
Number of Floors	1	3
	2	2
	3	1
Types of Fencing Material	Wood	3
	Concrete	1
	Mix Materials	2
	Metal	1
	Liha	4
	Thatch	5
	Tarp	5
	Others	3
Age of Building	0–11.57647059	1
	11.5764706–28.94117647	2
	28.94117648–44.85882353	3
	44.85882354–67.04705882	4
	67.04705883–123	5
Total Height of Building	Less than 5	3
	5–20 m	2
	More than 20	1
Distance to River	0–104.99	5
	104.9967968–221.6599042	4
	221.6599043–342.2117819	3
	342.211782–505.5401324	2
	505.5401325–991.6364136	1

.

2.4. Utilizing AHP to Develop a Flood Risk Map

AHP is a technique for organizing and analyzing complicated decisions using mathematical and psychological principles. The process starts by breaking down the decision-making goal into a hierarchical structure of criteria and alternatives. The criteria are then assigned weights, and the alternatives are scored with each other based on pairwise comparisons made by decision-makers. This weighting and scoring procedure generates a total score for each alternative, allowing them to be ranked [35]. AHP is widely used for decision-making in numerous fields, such as business, government, engineering, health care, and education. AHP is reliable and effective in mapping flood-prone areas and assessing the effects of flood risk [36,37]. It allows for the integration of numerous factors under different criteria, enabling an analysis of flood vulnerability and risk [37].

Procedure for Implementing the AHP

• Literature Review for Weight Assignment:
  • AHP (Analytical Hierarchy Process) was utilized to assign weights to various factors influencing floods.
  • The weights were determined based on a comprehensive literature review, ensuring that the prioritization reflects empirical evidence and expert opinions.
• Generation and Reclassification of Individual Maps: Individual flood-related maps were generated and reclassified using ArcMap which involved mapping various flood-related parameters such as hazard, vulnerability, and exposure.
• Overlaying Individual Maps Using Weighted Sum Tool: The Weighted Sum tool in ArcMap was used to overlay the individual weighted parameter maps. This step produced the flood hazard map, flood vulnerability and exposure maps by overlaying individual parameters using the Weighted Sum tool.
• The resulting maps (Figure 9a–c) represent building flood hazard, vulnerability, and exposure indices. Each map was divided into five index levels, ranging from very low to very high risk.

Each map was divided into five index levels ranging from very low to very high. The building flood hazard map Figure 9a using AHP shows that 50% of the area has moderate to very high risk. Meanwhile, areas without buildings are categorized as very low risk in the flood vulnerability map Figure 9b and the flood exposure map Figure 9c.

2.5. Utilizing TOPSIS to Develop a Flood Risk Map

TOPSIS is a method that identifies solutions from a finite set of alternatives based upon simultaneous minimization of distance from an ideal point and maximization of distance from a nadir point [38].

Researchers widely apply the TOPSIS method in FRA in MCDM. TOPSIS integrates various criteria, such as maximum flood depth, slope, elevation, and Average rainfall, to assess flood risk and create flood hazard maps. It allows for identifying vulnerable areas and estimating flood risk coverage, making it a valuable tool for decision support in flood management [21,23]. TOPSIS differs from traditional methods by providing a robust approach to FRA. It enables the systematic consideration of multiple criteria and the systematic integration of various factors, which allows for a more sensitive and accurate flood risk evaluation, mainly when dealing with uncertain or imprecise information. Additionally, TOPSIS has yielded more precise flood risk coverage estimates than other methods, making it a valuable approach for assessing and managing flood risks [21,22]. In our study, to determine the flood risk for buildings regarding hazard, vulnerability, and exposure, the study region has been divided into 41 distinct zones based on the catchment area. The catchment zone in TOPSIS was selected only for the area with survey data for accurate building risk assessment in the Area of Interest (AOI). This division is an analysis that enables the classification of these zones into five categories based on their risk levels to assess the risk of the buildings falling in that region.

The first step is Data Integration for each map (flood hazard, vulnerability, exposure). The value for each parameter was calculated within the defined catchment area zones. Afterwards, the computed statistics for the zone were added to the catchment area. This will provide us with the integrated data for hazard, vulnerability, and exposure maps to create the decision matrix and further calculate the TOPSIS method.

Procedure for Implementing the TOPSIS

After integrating the data, the TOPSIS method analyses the compiled data. Then, the final $Pi$ value with the defined rank and categories was integrated into the ArcMap again. Following the specific steps stated below:

• Establishment of the Decision Matrix: The exported Excel was included for all alternatives and criteria for respective maps.
• Normalization of the Decision Matrix $\left(R\right)$: Normalize the decision matrix to allow for comparison across different measurement scales using the Equation (2) in Excel:

  $$r_{ij}={\displaystyle \frac{x_{ij}}{\sqrt{{\sum }_{i=1}^n{x}_{ij}^2}}}$$

  (2)

  where $x_{ij}$ is the value of alternative i under criterion j in decision matrix,
  $r_{ij}$ is the normalized value of alternative i under criterion j, as calculated from the decision matrix.
• Weight Assignment and Weighted Normalized Decision Matrix: Assign weights to the criteria based on findings from the literature review using AHP, reflecting their relative importance in FRA and creating a weighted normalized decision matrix as described in the Equation (3).

  $$V_{ij}=w_j·r_{ij}$$

  (3)
  $V_{ij}$ represents the weighted normalized value of alternative i under criterion j,
  $w_j$ is the weight of criterion j, assigned based on a literature review,
  $r_{ij}$ is normalized value of alternative i under criterion j.
• Determination of Ideal and Nadir Solutions: The Ideal solution ($A_j^{+}$) is composed of the best outcome, and the Nadir solution ($A_j^{-}$) of the worst possible outcomes for each parameter from the weighted decision Matrix as shown in

  Table A3. Ideal best and ideal worst values of 18 criteria according to category.

  Parameters	Criteria Type	Ideal Best	Ideal Worst
  Building Flood Hazard Parameters
  Annual Average Rainfall	Non-Beneficial	2.17	2.35
  Slope	Beneficial	5.291009	0.448891727
  Elevation	Beneficial	9.012045996	1.315811282
  Flood Depth	Non-Beneficial	1.44	12.70
  Building Flood Vulnerability Parameters
  Average Income	Beneficial	0.858053051	0.429026526
  Gender Ratio	Beneficial	0.738466461	0.307694359
  Land Cover	Non-Beneficial	0.993674515	1.98734903
  Roofing Material	Non-Beneficial	0.96	0.96
  Flooring Material	Non-Beneficial	0.88	1.76
  Interior/Exterior Walling Material	Non-Beneficial	0.44	1.31
  Number of Floors	Beneficial	1.958888661	1.31
  Types of Fencing Material	Non-Beneficial	0.51	1.02
  Age of Building	Non-Beneficial	0.00	2.56
  Total Height of Building	Beneficial	2.467459204	1.233729602
  Distance to River	Beneficial	8.790167321	0.390308829
  Building Flood Exposure Parameters
  Building Density	Beneficial	12.60431978	1.101386616
  Number of Buildings per Area	Non-Beneficial	0.097842463	15.06773935
  Use of Building	Beneficial	1.48	5.91

  using the Equations (4) and (5). The beneficial criteria are the ones that give us low risk with a high value, and non-beneficial criteria are the ones that provide low risk with a lower value.

  $$A_j^{+}=\left\{\left(\underset{j}{max}{V}_{ij}|j\in J_1\right),\left(\underset{j}{min}{V}_{ij}|j\in J_2\right)\right\}$$

  (4)

  $$A_j^{-}=\left\{\left(\underset{j}{min}{V}_{ij}|j\in J_1\right),\left(\underset{j}{max}{V}_{ij}|j\in J_2\right)\right\}$$

  (5)

  where, $J_1$ represents the beneficial criteria, and $J_2$ represents the non-beneficial criteria,
  $V_{ij}$ represents the weighted normalized value of alternative i under criterion j.
  The equation above (4) and (5) means for beneficial criteria, the ideal best value is based on the maximum weighted normalized value among all alternatives, and the ideal worst is the minimum weighted normalized value among all alternatives. For non-beneficial criteria, the minimum weighted normalized among 41 alternatives is the ideal best, and the maximum weighted normalized value is the ideal worst.
• Calculation of Separation Measures (Euclidean Distance): Determine the distance of each alternative from the ideal and nadir solutions [(6) and (7)].

  $$S_i^{+}=\sqrt{\sum _{j=1}^m{(V_{ij}-A_j^{+})}^2}$$

  (6)

  $$S_i^{-}=\sqrt{\sum _{j=1}^m{(V_{ij}-A_j^{-})}^2}$$

  (7)

  where, $S_i^{+}$ is the distance of alternative i from the ideal solution,
  $S_i^{-}$ is the distance of alternative i from the nadir solution,
  $V_{ij}$ is the value of weighted normalized matrix for alternative i under criterion j,
  $A_j^{+}$ is the ideal best score for criterion j,
  $A_j^{-}$ is the ideal worst score for criterion j.
• Relative Closeness to the Ideal Solution: Afterwards, the area’s ranking was determined based on their proximity to the ideal solution, indicating their flood risk levels. The relative closeness of an alternative $Pi$ to the ideal solution is given by the Equation (8) as follows:

  $$Pi={\displaystyle \frac{S_i^{-}}{S_i^{+}+S_i^{-}}}$$

  (8)

  where a higher $Pi$ value indicates a performance score for alternative i. This measure helps rank the alternatives by determining their closeness to the ideal condition compared to the worst solution. The value of $Pi$ was utilized to rank each catchment zone. $Pi$ values closer to zeros are not desirable with high rank, and closer to 1 are desirable with low rank [39].
• Data Analysis and Interpretation: The relative closeness scores were used to categorize and rank each zone. Excel was utilized to support the analysis, handling complex calculations and visualizing the distribution of vulnerability across zones. The zones were classified into five categories based on their performance index:
  • Category 1 (Very Low): Zones with performance index values greater than or equal to the 80th percentile.
  • Category 2 (Low): Zones scoring above the 60th percentile but less than or equal to the 80th percentile.
  • Category 3 (Moderate): Zones with values greater than the 40th percentile but less than or equal to the 60th percentile.
  • Category 4 (High): Zones scoring above the 20th percentile but less than or equal to the 40th percentile.
  • Category 5 (Very High): Zones with values less than the 20th or equal to percentile.
• Integration of Excel Data with ArcMap: After finalizing the rankings in Excel, the dataset was exported and incorporated into ArcMap to facilitate geospatial visualization by zones.
• Mapping and Spatial Analysis: The integrated analysis, ranking, and categorization for each alternative (catchment area) were integrated from Excel to ArcMap. These were used to create the raster maps for flood hazard, vulnerability, and exposure maps. In ArcMap, each zone was mapped based on its calculated analysis, using colour gradients for different levels of risk as shown in Figure 10a–c. This visual method effectively identified areas requiring immediate focus and comprehended the risk factors’ spatial distribution.

3. Results and Analysis

All three components of the flood risk Equation (1) have been analyzed to develop the final building flood risk map. Flood risk is a triad of flood hazards, vulnerabilities, and exposure maps. Hence, they are combined and overlayed to get the final building flood risk map in GIS, considering their equal importance. By integrating the hazard, vulnerability, and exposure map utilizing AHP and TOPSIS methodologies, final building flood risk maps were generated for AHP and TOPSIS as shown in Figure 11 and Figure 12. The impact of floods on buildings is represented in maps by categorizing them from very-low to very high-risk levels. This can help determine the risk level for each building in the AOI.

3.1. Operational Differences between AHP and TOPSIS

The maps of the flood impact on the building using AHP and TOPSIS can differ due to their methodological differences. For instance, on a flood hazard map produced by the AHP, a building may be assigned a very low risk, yet on a map created by TOPSIS, the same building may be assigned a moderate risk category. This comparison highlights the variations in risk assessment using different multifaceted approaches to deal with flood risks. For comparison, each map was associated with 555 surveyed buildings for Naic and was classified by risk levels using the ‘spatial join’ tool to count the number of buildings in each category.

Both methods, AHP and TOPSIS, handle individual parameters differently. AHP relies on pairwise comparison to determine the weights of the parameters based on the hierarchical structure. The final maps are created using a weighted sum of individual parameters. Therefore, AHP methodology results are sensitive to the accuracy of the pairwise comparison and subjective to expert judgments. Due to the crisp nature of data and the hierarchical structure of this methodology, the building count in risk categories can show more variability based on the weights. Small changes in expert judgments can significantly lead to changes in the weights of parameters, which impact the results of AHP. On the other hand, TOPSIS provides a novel methodology compared to AHP, providing a ranking based on relative closeness to the ideal best solution. In TOPSIS, data is explicitly normalized to ensure compatibility across individual parameters in varying scales, after which AHP weights are applied. This methodology is sensitive to accurately determining the ideal best and ideal worst for each parameter and the weights assigned to the parameters. Further discussion of analysis of building shows how the operation difference in methods impacts the number of building counts in each risk level (very low–very high) is done in the Section 3.2, Section 3.3 and Section 3.4 for hazard, vulnerability, and exposure maps respectively and also for the integrated building flood risk map in Section 3.5.

3.2. Analysis of Building Count under Flood Vulnerability Map

In the AHP results, only 2 buildings fall under the very low category. AHP methodology combines individual maps using a weighted sum approach. Maps with cumulative weights exceeding 50% indicate a bias towards the moderate to high-risk area in building data. The results reflect this bias, with 229 buildings classified as very high risk, 178 buildings as high risk, and 138 buildings as moderate risk (see

Table 4. Building Count Analysis on Flood Vulnerability Map.

Category	AHP	TOPSIS
Very Low	2	80
Low	8	78
Moderate	138	117
High	178	210
Very High	229	70

). This distribution means that almost more than 90% of the buildings are moderate to very high-risk buildings according to the AHP method.

In TOPSIS, the catchment area was selected based on the distribution of survey data. As a result, zones that did not include survey data were excluded from the AOI, leaving only 41 zones for analysis. The results show that, out of 41 zones, 16 are very low to low-risk, with 158 structures falling into that risk category, suggesting that the buildings’ attributes determine each zone’s vulnerability. 8 out of 41 zones have 117 buildings classified in the moderate risk category, and 17 out of 41 are classified as high to very high risk, putting 280 of the examined buildings in that category. The result in the Table 4 is neither biased towards very low nor very high values by the TOPSIS method. It is a straightforward ranking system based on the distance from the ideal solution, making it useful for precise decision-making scenarios.

3.3. Analysis of Building Count under Flood Exposure Map

A comparable impact trend is evident across the individual parameters in the flood exposure and vulnerability maps. The formulation of the flood exposure map relies on attributes derived from surveyed building data.

Table 5. Building Count Analysis on Flood Exposure Map.

Category	AHP	TOPSIS
Very Low	4	57
Low	0	52
Moderate	101	78
High	193	144
Very High	257	224

shows that only 4 buildings were classified as very low risk using AHP. However, in the TOPSIS method, 75 buildings are classified as low risk. In the AHP method, the maximum number of data points in building density and use of the building are in the high-risk category, and the weightage of these two categories is approximately 74%. This results in 257 buildings and 193 in very high and high-risk categories, respectively. In the TOPSIS methodology, as seen in Table 5, 446 buildings under moderate to very high-risk zones can be observed, and 109 buildings are under very low to low risk. The TOPSIS methodology results do not represent the explicit data categories of individual maps as observed in AHP. However, the similarity in the results can be seen due to the influence of individual parameters. The exposure map shows that most of the buildings are under the moderate to very high-risk category in both methods because of the skewed data on the use of buildings and the number of buildings.

3.4. Analysis of Building Count under Flood Hazard Map

The buildings under the flood hazard map are analysed by including environmental factors such as annual average rainfall, flood height, slope, and elevation. The flood hazard map does not include a direct analysis of the building attributes as done in the vulnerability and exposure map. It analyses the parameters that affect the condition and vulnerability of buildings during and after flood events.

The flood hazard map using the AHP method shows that 101 buildings fall under the very low-risk category, 104 buildings are at low risk, 218 are at moderate risk, 101 are at high risk, and 31 are at very high risk by considering the geographical and topographical conditions around the buildings. More than 63% of the surveyed buildings are under moderate to very high-risk regions of the flood hazard map as presented in

Table 6. Building Count Analysis on Flood Hazard Map.

Category	AHP	TOPSIS
Very Low	101	60
Low	104	81
Moderate	218	60
High	101	188
Very High	31	166

. Only around 37% of the buildings are under the very low and low-risk category, with 101 buildings at very low risk and 104 buildings at low risk. By TOPSIS, around 74% of the buildings are classified as moderate to the very high-risk category, and the rest of the buildings with only 26% are in the very low to low-risk category. Since flood hazards focus on environmental factors, the risk categories are estimated using historical and geographical data.

3.5. Analysis of Building Count under Flood Risk Map

Finally, after integrating all three maps (hazard, vulnerability, exposure) in both methods, a flood risk map was generated as described by Equation (1). The AHP method uses crisp data and the weightage of individual parameters to generate the final maps; the data available in individual maps having the highest weightage are more sampled towards moderate to very high-risk categories. This characteristic of data is directly reflected in the output of AHP, which makes 91.3% of the building under moderate to very high-risk categories, which can be alarming to the residents of the building and measure action towards it should be taken when a flood occurs. On the other hand, TOPSIS is better at handling such imbalanced data sets through its unique methodology, as seen in

Table 7. Flood Risk Map Building Count.

Category	AHP	TOPSIS
Very Low	4	51
Low	44	125
Moderate	201	78
High	212	61
Very High	94	240

where 68% buildings are moderate to high risk. In the AHP method, only 4 buildings are classified as very low-risk. This may be due to the scarcity of buildings in this category on vulnerability and exposure maps. While there are 101 buildings in the very low-risk category on the flood hazard map, their structural integrity may not be robust, making them susceptible to minimal flood damage.

4. Discussion

The result presented in the study shows a notable difference in the risk categorization between AHP and TOPSIS using GIS. Similar to our study, other researchers have employed AHP, TOPSIS, and other MCDA methods to asses the flood risk and have noted the outcomes due to inherent methodological differences. For instance, a comparative analysis study by Pathan et al. [21] examined both AHP and TOPSIS methodologies. This perspective leads to interpreting the outcomes of this study more effectively and shows that TOPSIS provides greater consistency with actual values. Numerous studies have evaluated flood risk using a single MCDM method. For instance, Huang et al. [40] used AHP and discovered it was susceptible to expert opinion, causing variations in risk categorization. A similar observation was made in a study by Sarmiento et al. [41] that AHP might magnify biases brought about by expert opinion, especially when the parameter weights are highly skewed. The results show that AHP often placed more buildings in the moderate to very high-risk categories, which is overall 91.3% buildings from survey data. It can be due to the subjective nature of pairwise comparisons and the hierarchical structure inherent in AHP.

On the other hand, TOPSIS normalizes data to ensure compatibility across different scales and ranks alternatives based on their distance from the ideal solution, which provides a more dispersed risk distribution. As observed in the study, 68% of buildings were categorized from moderate to high risk, which is lower than AHP but still indicative of significant risk. It means TOPSIS provides a more balanced distribution, affirming the results from the study by Ekmekcioğlu et al. [42] and Nyimbili et al. [43], where TOPSIS shows robustness in handling imbalanced datasets and is more suited for the implementation in a raster data structure as used in our study in a GIS environment. TOPSIS also offers a straightforward ranking based on relative closeness to the ideal solution.

5. Conclusions

Risk assessment of the buildings in the Naic region can assist in flood risk management and strategy planning, as buildings serve as a primary defence against flood disasters for a community. Also, the study aimed to identify the extent of buildings’ vulnerabilities and risk zones, which can help recommend practical measures to improve buildings’ strength on a priority basis before and after floods. Utilizing these methodologies can be the first step in understanding the impact of building attributes on flood risk and helping rural communities like Naic to be more resilient against flood hazards. Different methodologies presented in this study help understand the use case for qualitative risk assessment based on data and risk categorization. Both methods have pros and cons. The building count in both risk categories differs when using AHP and TOPSIS due to their fundamental methodological differences in handling the data to produce flood risk maps. As building FRA is determined using the MCDM technique with numerous variables such as assumptions, expert opinions, and conflicting parameters, no optimal solution exists as the output or the optimal solution based on the mentioned variables [44].

Both methods consider the relative importance of each parameter for hazard, vulnerability, and exposure maps. In AHP, the weighted sum overlay is done with a crisp value for each parameter. On the other hand, the TOPSIS method involves normalizing the decision matrix of the parameter values, which helps bring the dataset into a uniform scale. Therefore, the effect of individual parameters is based on their importance rather than their true value, which is then used to create the weighted decision matrix. Even though TOPSIS involves more complex calculations during analysis, making it computationally complex and time-consuming compared to AHP, it is more suitable for datasets with different units. Calculating the relative closeness to the ideal best solution ensures that the parameters are analyzed and compared to the benchmark of optimal performance, promoting more reliable and effective decision-making. TOPSIS approach has methodological differences compared to AHP because it calculates the Euclidean distance of each parameter from the ideal best to the worst. TOPSIS can rank AOI in terms of zones or exact building locations by determining the relative closeness to the ideal best solution. Finally, the choice of the method depends on the particular need for risk assessment and the nature of input data. Moreover, both MCDM techniques analyzed in this study are reasonably basic, with a few computational processes that can be created and used in GIS. The results of both methods can be analyzed in a GIS environment. AHP and TOPSIS differed in GIS-based processing and analysis of a flood risk map after each parameter was selected and evaluated.

Despite the advantages of MCDM techniques, some inherent limitations are associated with them. Less data can severely affect the performance of outputs using the MCDM technique. The survey data of this study is only available for 555 buildings in the Naic region (AOI), but there are more than 10 thousand buildings in AOI that have yet to be surveyed. The results could be more accurate if a complete dataset for all parameters within the AOI were available rather than assuming the absence of buildings in certain areas.

In all 18 parameters, 4 were for hazard, 11 were for vulnerability, and 3 were for exposure, considering the environmental and socio-economic conditions with buildings’ attributes and how they affect the people of the Naic region. In future studies, more parameters, such as drainage density, soil type, and population density, can be considered. A sound drainage system helps reduce flood water accumulation in a region, so drainage density is essential to consider during flood risk analysis. The same goes for soil type in the context of water absorption and population density to check the number of people exposed in certain high-risk zones during floods. Future research should also explore geomorphological parameters that influence flood behavior and assess the role of drainage basins contributing to flood events, considering their size, shape, and hydrological characteristics.

Further research and comparison of MCDM techniques can be made on similar datasets to find suitable methods for building risk assessment such as Fuzzy Logic [45], ELECTRE (ELimination and Choice Expressing Reality) [46], PROMETHEE (Preference Ranking Organization Method for Enrichment Evaluations) [47] and SAW (Simple Additive Weighting Method) [48]. Fuzzy Logic has a wide variety to handle different datasets by providing various fuzzification methods. It includes a rule-based inference engine where custom rules can be designed based on the expert opinion and available dataset.

Overall, the MCDM technique is beneficial and efficient in buildings FRA as it provides decision-making with objectivity for multiple conflicting criteria and the participation of the researcher and experts. MCDM addresses the need for a more robust structure by avoiding inappropriate and fragile materials.