Geospatial Mapping of Inland Flood Susceptibility Based on Multi-Criteria Analysis – A Case Study in the Final Flow of Busu River Basin, Papua New Guinea

Kumne, W. 1 and Samanta, S.2*

1Post-Graduate Scholar, Department of Surveying and Land Studies, The Papua New Guinea University of Technology, Papua New Guinea

2Department of Surveying and Land Studies, The Papua New Guinea University of Technology Papua New Guinea

*Corresponding Author

DOI:https://doi.org/10.52939/ijg.v19i6.2693

Abstract

Floods are one of the most extensively dispersed risks all over the world. The aim of this research is to map out potential flood susceptible zones and analyse the risk using integrated Remote Sensing (RS) and Geographic Information System (GIS) Technology. RS techniques present inexpensive and quicker decisions for retrieving spatial data about the flood occurrence in the actually unreachable areas. On the other hand, GIS techniques enable hydrological demonstrations in data gathering, examination, querying, and demonstration of information in a further easy layout. This research focused on the flooding problem in the Busu River basin and demonstrated a detailed geospatial mapping of inland flood susceptibility based on Multi-Criteria Analysis (MCA) approaches like Pairwise comparison, ranking and weight, and Boolean logic. Different flood factors namely slope, elevation, flow accumulation, soil drainage, soil texture, surface runoff, distance to the active channel, land use land cover, and lithological characteristics were considered for the flood hazard assessment in the lower catchment of the Busu River. Storm rainfall data were used to calculate surface runoff, while the population layer was used for flood vulnerability analysis by overlaying them on the flood hazard map. These flood susceptibility maps especially from Pairwise comparison and ranking approaches can be used for flood mitigation and flood hazard preparedness, as they are more accurate than the Boolean logic approach.

Keywords: Flood, Geographic Information System, Multi-Criteria Decision Analysis, Remote Sensing, Risk assessment

1. Introduction

Floods cause damage to natural resources and the quality of the environment and indirectly contribute to increasing poverty, which consequently increases the vulnerability of both natural and human systems [1]. The effect and rates of natural disasters are on the rise each year. More than 18% of all-natural disasters occur in developing countries and 50% to 60% of these countries are extremely vulnerable [2]. Flood remains highly a customary natural disaster and the size and the economical approach to flood disasters are at their peak in the affected area around the world. The terms “floods”, “flood hazard”, and “flood risk” cover a broad range of phenomena. The terms such as “flood risk” and “flood losses” are essentially our interpretation of the negative economic losses and social consequences of natural events. Flood risk may increase due to human activity and may decrease through appropriate flood management and planning [3].

Flood is a general temporary condition of partial or complete inundation of normally dry areas from overflow of inland or tidal waters or from unusual and rapid accumulation or runoff [4]. Floods are a natural phenomenon that results in temporary submerging with water and destructs agricultural lands, built environment, and infrastructure and may result in casualty, loss of lives, and economic loss [5] [6] [7] [8] [9] [10] and [11]. Flood does not occur under normal conditions and cannot be prevented causing displacement of people [12]. Floods pose a threat to environmental and socio-economic effects [13] [14] [15] and [16]. A number of factors trigger inland flooding.

Heavy precipitation is a major contributing factor to flood disasters when the normal waterway cannot transport the extra water from the rain due to the unavailability of a proper drainage network [17]. Nonetheless, heavy rainfall is not the only factor to flood disasters; floods can also be triggered by dam failures, tsunamis, storm surges, etc. that may have resulted from geological activities. Floods may occur due to anthropogenic activities and human interventions in the natural processes such as the increase in settlement areas, population growth, and economic assets over low-lying plains prone to flooding leading to alterations in the natural drainage and river basin patterns, deforestation, and climate change [18].

Lae, the second largest city of Papua New Guinea had experienced two major flood and mudslide disasters. In both cases, hundreds of people lost their homes. The 1983 floods remain the worst since the establishment of the town in the late 1920s. These floods left hundreds of people homeless particularly those living along the banks of the Bumbu River [18]. Many houses were damaged and hundreds of people at the Five Mile settlement along the Highlands Highway were affected by mudslides. The Busu River in the Naweb District of Morobe Province is one of the most flood-problematic rivers. The basin is classified as one of the most devastating flooding rivers in Papua New Guinea.Reviewing the past records and oral histories of the Busu River, the area near the basin encountered a lot of flood occurrences which led to the loss of lives and properties [18]. Today Busu River is still a threat to the people living around and near the river. This is endorsed by the climate changes and other major flood factors that made adjustments in sedimentation weight and water capacity. The river flows wide due to the flatness of the landforms in the floodplain zone. These are the consequences of the flooding of the extremely huge region and there are many additional channels established which do not connect to the major channel. The rivers geomorphology changes due to the additional streams, which presents a clear understanding about the river and causes of flooding. The geomorphic, hydrological, and topographic features along the river with other persuasive factors aid in the flood hazard examination for Busu River that revealed information for governmental departments, planners, settlers, and other benefactors to make better decisions that will minimize the flood damages in terms of properties, farms, and even lives. It is important to create flood hazard maps that are user-friendly that will help focus on the mitigation validities.

Many methods were used to study flood issues by different workers around the world with their own sets of parameters according to individual river situations. This flood assessment and flood risk mapping approach include geomorphology, hydrology, topography, demography, and meteorology as subsidiary parameters [19]. RS and GIS technology has become suitable platform that has the capability to deliver large bulks of information on a well-timed basis and cost-effectively. Till now, a lot of recognized achievements have been encountered with the use of satellite data and geomorphology in flood risk assessment and mapping. Compound analyses, like flood susceptibility evaluation and flood loss evaluation, are still active issues to work on.

The GIS-based models with varying numbers and types of flood-influencing factors have been successfully utilized in the mapping of inland flood hazard areas by several studies [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] and [29]. In the recent past, different geospatial methods were used in flood analyses, which have proven themselves much more effective and realistic. Hydrological models are the first category which includes SWAT [29], WetSpa and HYDROTEL [30], and hydrodynamic approaches based on the shallow water equations initialized by rainfall [31] [32] and [33]. Statistical and knowledge-based methods were also proven to be much more effective in flood analysis, which includes weights of evidence (WoE) [29] [34] and [35], logistic regression (LR) [35] [36] and [37], analytic hierarchy process (AHP) [38] [39] [40] and [41], frequency ratios (FR) [28] [42] and [43]. Machine learning models can handle complex nonlinear problems such as neuro-fuzzy logic [44], artificial neural networks (ANNs) [45] and [46], decision trees (DTs) [47], support vector machines (SVMs) [48] and [49], adaptive neuro-fuzzy inference system (ANFIS) [50], biogeography based optimization and BAT algorithms [51], reduced error pruning trees [50], multivariate adaptive regression splines [52]. Multi-Criteria Decision Analysis (MCDA) approach can employ the combination of socioeconomic, environmental, and technical objectives to achieve the most favorable decision [19] and [53] and has proven to be the best model in flood analysis [54] and [55].

The study aims to focus on flood susceptibility mapping using various spatial models like Pairwise comparison, ranking, and Boolean logic. The core assignment of this research is to the identification of flood hazard zones in the lower catchment area of the Busu River. Three objectives were considered to achieve the aim of this study.

First objective was to identification and preparation database on different input parameters for flood susceptibility mapping in the lower catchment area of Busu catchment. Secondly, to develop potential flood risk zones through multi-criteria decision approaches and establishes the best method after the validation process. Finally, to perform flood vulnerability assessment after overlaying flood risk zone database and population information and other infrastructure.

2. Study Area

Busu River is considered one of the fast-flowing rivers in Morobe Province and is located just outside of Lae City, the capital of Morobe Province. Busu River is made up of all the tributaries in the Sankwep catchment (Figure 1). Sankwep is one of the many catchments in Morobe Province. Sankwep catchment mostly takes up Nawae and Lae Districts with the lower portion of the Kabwum district. This study only concentrates on the Busu flood plain zone. This area is situated in the Ahi Local Level Government in Lae District, situated between the longitudes of 147º03′00′′ E and 147º01′12′′ E and the latitudes of 06º39′52′′ S to 06º43′30′′ S. Morobe is one of the provinces with well-known rivers with Markham the largest river with Busu the fast-flowing river. Busu originated from the Gain, Boana, Lambaip, and Kwapsanek mountain ranges and it is named based on the famous Busu Mountain, which is the main tributary. Along with its tributaries, the river drains a large amount of area in Morobe Province, particularly in the Naweb and Lae Districts. Mountains and ridges in the source zone bound Busu River to flow faster and it is the fastest flow in the country. The river is merged by main tributaries in Kwapsanek range north of the Revival Centers of Papua New Guinea Headquarters, the other major tributaries joined from the Gain, Boana, and Kwalan areas, which are east of the Nazab Airport.

Figure 1: Location map of the study area: (a) Papua New Guinea, (b) Morobe province and (c) Longitudinal structure of the Sankwep catchment

The river has built a mega-fan of some 25,000 square kilometers in extent breaking into two districts, mostly the Naweb District, all with changing passages due to the flooding. The Sankwep, Busip, and Gobadik are the major tributaries of the Busu River with other minor tributaries. The watercourse often changes the path and affects the local settlement [56]. This shift of the channel happens due to the flood, which is repetitive because of rainy weather all year round. The flood related damages caused by the Busu River have extreme effects on the people and the environment in the flood-prone areas. The longitudinal structure of the Sankwep catchment showed the source, transitional, and flood plan zone, where the flood plain zone is taken into account for flood susceptibility analysis using different GIS models (Figure 1). The sediments carried during flood return to the old channel (paleo) which causes the river stream to divert in a different direction each year. The main settlements and villages that are assembled near the flood-challenging river are Bumeng, Malanhang, Bumayong, Situm, and other smaller settlements. Settlements that are often affected by the floods were Second Seven, Pepsi, Nabak, and Wain-Busu.

Several attempts have been made to avoid forthcoming river penetration and flooding with local flood protection systems like the establishment of ridges, planting of the bamboo plant, and digging up to redirect the breaching flows to its main canal. Their previous protection schemes failed several times which steered them into great fear, according to the locals living along the river catchment. They feared that the course of the river near its main current will run to the flooding of the settlement if the height of the water increased. Although several works are done previously that considered mitigation measures, it is still uncertain, natural disasters are inevitable. Therefore, it is essential to reflect on the vulnerability and do risk assessments by analyzing the parameters that contributed to the inundations and do better validations upon the results produced using GIS and RS techniques. The Morobe provincial government and the Naweb and Lae Districts local government have raised this concern of risk assessments to do flood mitigations but not to the extent where they can fully mitigate the flood.

3. Methodology

The Busu River flood risk assessment and flood risk zonation database were developed after integrating all thematic layers (controlling factors) which are mandatory in flood susceptibility mapping. Data analysis was done in ArcGIS 10.5 with the use of the DEM to generate the slope map, elevation map and flow accumulation map, and other hydrological analyses. All the data obtained from several resource centers and organizations were converted to the grid or GIS-compatible format for the Busu flood assessment. All the data sets were combined through a multi-criteria analysis approach for the planning of inundation exposure and flood susceptibility map.

3.1 Data used and their Description

Satellite images, topographic maps, rainfall data, population data, digital elevation model (DEM), and historical flood inventory database were used to derive all mandatory input factors for flood susceptibility mapping. Some of these decisive factors include geomorphic features, elevation, slope, land use/land cover, drainage density, flow accumulation, flow direction, rainfall, soil texture, and soil drainage. The usable data with their descriptions for this study are listed in Table 1. Elevation data is essential in any flood risk assessment within a designated drainage basin and Busu is not an exception. Elevation layers explain elevation values from corner to corner of the extent of maps depicting the real world.

Table 1: Data sets and input parameters for flood susceptibility modeling

Type of Data

Resolution

Source

Data

Output Layer

Lidar Image (Orthophoto)

Land Use layer

20 cm

PNG University of Technology, Surveying and Land Studies Department

Digital Elevation Model

Elevation, Slope, Flow Accumulation

30 m

Rainfall

Strom Rainfall, Total rainfall

Stationed data

250 m

PNG Resource Information System [57]

Soil

Texture, Drainage

250 m

The PNG Geobook [58]

Geology

Lithological description

500 m

Flood

Flood points

Point Location

Field Survey using hand held GPS and other historical flood inventory

Population

Population Density

5 people per house

National Population and Housing Census of Papua New Guinea, 2011.

Many other researchers discovered that good elevation data must be included in any analysis relating to elevation to generate improved outcomes with minimal inaccuracy, which have a great ability to implement. The advanced space-borne thermal emission and reflection radiometer (ASTER) DEM is one of the topographic data sets (spatial resolution of 30m) used as the input raster. For regional or local assessments, the 30m ASTER DEM is available for the required topographic information to project potential impacts on the flooding area. The elevation, slope, and flow accumulation layers were generated from the 30m ASTER DEM. Elevations layers are also helpful in 3D visualization in terms of creating watershed delineation. Slopes have greater impacts on hydrology studies. Slope decides whether water is concentrated or isolated [59]. The slope tools under surface analysis of spatial analyst tool bar helps to develop the slope in degree through the simple logic ‘the inverse-tangent of the rise divided by the run’ from DEM data. The water runs faster when the river source zone has a steep and lengthy slope. In this case, the catchment has a bigger or higher chance of erosion, which leads to flooding in the floodplain zone. The length and the steepness of the slope also affected the flooding of the Busu River [60]. The lowland area with low slope approach floods first as contrasted to the high slope area throughout flood inundation. The regions of the sheer slope indicate excessive topmost flux contrasting to the low-lying region and trigger storage reduction throughout the upper catchments. Flow direction is the direction in which the stream water flows from one pixel to its neighboring pixels and gradually accumulated into each down slope cell called Flow accumulation. Both parameters are very essential for flood modeling [61]. Flow accumulation layer was generated through the navigation of ‘Spatial Analyst Tools’ followed by ‘Hydrology’ and ‘Flow Accumulation tool’. Before flow accumulation analysis, flow direction layer was prepared using DEM by the fill method. Rainfall is a major factor in flood and flood-related disasters [62]. Local water accumulation and river overflowing are the core sources of flood-associated catastrophes. Lae is known as ‘Rainy-Lae’ because of unpredictable and heavy rainfall in this area all year round. The continuous heavy downpour has a higher chance of flood. Rainfall data is necessary for considering the water depth in a flood risk assessment. Rainfall has a major impact on the depth of water. Soil type also stimulates rainfall when it reaches the ground. Impermeable soils and rocks such as clay restrict water to infiltrate and increase the chances of long-term water logging and flooding. Therefore, soil data is another important factor to consider in flood assessment [63]. Two separate layers were extracted from the soil data. These layers are soil texture and soil drainage layers. Soil texture describes the amount of sand, silt, and clay-sized units that compose the soil mineral portion. They are classified as loam, sandy loam, or clay. Soil drainage is a regular practice in which water shifts past, within, and away from the soil which results from the gravitational force. Surface runoff is one parameter to understand the flood in any area that is generally derived using the storm rainfall, soil, and land use/land cover based on the Soil Conservation Service (SCS) model. The population is another significant factor in flood risk assessment [43]. Population data enable the authors to calculate the magnitude of the populace exposed to the flood hazard. Better risk assessment is done when accurate and updated data is available. The population data were collected by field survey and interview. The population data was used for vulnerability assessment only.

3.2 Multi Criteria Analysis

Multi-criteria analysis was practiced with the contributing factors of a flood occurrence. The susceptibility areas were demarcated by numerically overlaying all the flood factor layers. Based on the expert's opinions and the availability of data the criteria were selected, compared, and ranked. Three methods were used in this study to identify the inland flood susceptibility zone. These three methods are (i) the Pairwise comparison method, (ii) the Ranking method, and (iii) the Boolean logic method. Furthermore, the resulted outputs were compared to identify the best method after the verification and validation of each using historical flood occurrences (Figure 2).

The Pairwise comparison method was introduced in 1980 [64] to determine the weight of every criterion. This method is the popular method being used by researchers successively [38] [39] [40] and [41]. This method can convert subjective assessments of relative importance into a linear set of weights. This method involves the comparison of the criteria and allows the comparison of two criteria at a time. The values obtained from the weighting of the criteria are entered into the pair-wise comparison matrix. The values in the Pairwise comparison entirely depend on the author’s knowledge based on the understanding of the flood factors, the flood history, and the flood characteristics in the study area. Table 2 is filled by comparing one flood factor with another until all factors are compared to each other.

Figure 2: The overall methodological flowchart for inland flood susceptibility mapping

Table 2: Pairwise comparison matrix of the flood factors

Flood Factors

SR

DAC

EL

SL

LULC

FA

ST

SD

LIT

SR

1

2

3

3

5

6

7

9

9

DAC

0.5

1

2

2

3

5

7

8

9

EL

0.33

0.5

1

2

3

5

4

5

6

SL

0.33

0.5

0.5

1

4

6

5

5

7

LULC

0.2

0.33

0.33

0.25

1

7

5

4

6

FA

0.17

0.2

0.2

0.17

0.14

1

3

2

5

ST

0.14

0.14

0.25

0.2

0.2

0.33

1

2

4

SD

0.11

0.125

0.2

0.2

0.25

0.5

0.5

1

3

LIT

0.11

0.11

0.17

0.14

0.17

0.2

0.25

0.33

1

SUM

2.89

4.905

7.65

8.96

16.76

31.03

32.75

36.33

50.00

These factors are shown in the table with their short form, namely Surface runoff (SR), elevation in meters (EL), distance to active channel (DAC), slope in degree (SL), land use/ land cover (LULC), flow accumulation (FA), soil texture (ST), soil drainage (SD), and lithology (LIT).

After the Pairwise comparison matrix is filled, the criteria are normalized to obtain the normalized matrix. The normalized values were obtained by dividing the value in each column by the sum in each column (Table 3). The priority vector is calculated by dividing the row total of the flood factor by the number of flood factors (Table 4). For example, the priority vector of Surface Runoff is 2.59/9 = 0.29. In the next step, lambda (λ) values of each criterion were calculated by multiplying the Priority Vector (PV) with the criteria sum. The λ max (Lamdamax) was calculated as 10.299. The Consistency Index (CI) was calculated as 0.162 with the help of lamdamax. The consistency analysis was done through consistency ratio (CR) calculations to check whether the weights are accepted or not and it was recommended that CR should be definitely below 0.2 [65].

Table 3: Normalizing the criteria column to obtain the normalized matrix

Flood Factors

SR

DAC

EL

SL

LULC

FA

ST

SD

LIT

Row Total

SR

0.34

0.4

0.39

0.33

0.3

0.19

0.21

0.25

0.18

2.59

DAC

0.17

0.2

0.26

0.22

0.18

0.16

0.21

0.22

0.18

1.8

EL

0.11

0.1

0.13

0.22

0.18

0.16

0.12

0.14

0.12

1.28

SL

0.11

0.1

0.07

0.11

0.24

0.19

0.15

0.14

0.14

1.25

LULC

0.06

0.07

0.04

0.03

0.06

0.23

0.15

0.11

0.12

0.867

FA

0.05

0.04

0.03

0.02

0.008

0.03

0.09

0.06

0.1

0.428

ST

0.05

0.03

0.03

0.02

0.01

0.01

0.03

0.06

0.08

0.32

SD

0.04

0.03

0.03

0.02

0.01

0.02

0.02

0.03

0.06

0.26

LIT

0.04

0.02

0.02

0.02

0.01

0.006

0.008

0.009

0.02

0.153

SUM

1.0

1.0

1.0

1.0

1.0

1.0

1.0

1.0

1.0

Table 4: Priority vector and weighted percentage

Flood Factors

Row Total

Priority Vector (PV)

SR

2.59

0.288

DAC

1.8

0.200

EL

1.28

0.142

SL

1.25

0.139

LULC

0.867

0.096

FA

0.428

0.048

ST

0.32

0.036

SD

0.26

0.029

LIT

0.153

0.017

In case the number of parameters exceeds 7; the consistency can be expected to be poor [64]. In this research, the CR value was calculated as 0.11 with the help of ratio index (RI) value of 1.46 for nine parameters. Although the CR value is marginally higher, logically it was considered as the acceptable consistency with the use higher numbers of (nine) parameters.

The ranking method is one of the most comfortable operative assessment techniques [66]. In the Ranking Method, multiple conflicting factors are ranked with various importance. The factor ranking of the flood conditioning factors was selected based on the author’s preferences and interviewing with the respective local experts. The ranking method is used in a matter where every criterion under consideration was ranked in the order of the decision maker’s preference. Each factor is rated according to the estimated significance of causing flooding and ranked using the inverse ranking method [19]. For example, the least influential criterion is ranked 1, then 2 for the next least. Weights in numbers were generated in order to do the criteria ranking [62] and [63]. The criterion values generated for each factor, every factor was weighted according to the predictable implication for flood triggering (Table 5).

A Boolean logic method is an overlay method, where logical operators combine all the criteria [68]. Boolean method is a helpful and commonly practiced analysis operator in GIS [68]. Boolean logic centered on fundamentals of binary logical operations, and its mathematical structures only centered on the values 1 for true and 0 for false. GIS Boolean operators for criteria selections are AND, OR, XOR, and NOT. Most GIS codes Boolean operations precisely are compatible with existing functions and have the same name. Intersect (AND), Union (OR), and Erase (NOT) are familiar operators in GIS. Boolean logic operates in the sense that one searches for areas that come across specific conditions. When the condition is specified or met, it is given the value 1 and when it is not, given zero (0). In the Boolean method, all the nine layers (flood factors) were reclassified in order to convert to integer values of 0, 3, 4 and 5. The non-zero values represent the areas of moderate risk (3), high risk (4) and very high risk (5) respectively and on the other hand, the very low risk and low risk were converted to zero (0). The Boolean AND operations were used to analyse four paired flood factors except surface runoff. The pair factors were (i) Elevation & Slope, (ii) LULC & Distance to active channel, (iii) Soil texture & Soil drainage, and finally (iv) Flow accumulation & Lithology. In the final step, these four pair of intermediate resulted factors and surface runoff factor (five layers) were summed in the raster calculator under spatial analyst tool in ArcGIS10.5 to generate the flood risk map (Figure 3).

Table 5: Rank and weighted of flood factors for the ranking method

Flood factors

Class

Rank

Risk vulnerability

weighted

Surface runoff (in mm)

0-25

25-106

106-157

157-189

187-229

1

2

3

4

5

Very Low risk

Low Risk

Moderate Risk

High Risk

Very High Risk

9

Distance to active channel (in meter)

Less than 200

200-400

400-600

600-800

More then 800

5

4

3

2

1

Very High Risk

High Risk

Moderate Risk

Low risk

Very Low Risk

8

Elevation in meters (in meter)

Less than 15.6

15.7 – 31.8

31.9 – 47.5

47.6 - 65.5

More than 65.6

5

4

3

2

1

Very High Risk

High Risk

Moderate Risk

Low risk

Very Low Risk

7

Slope (in Degree)

Less than 3

3-7

7-15

15-30

More than 30

5

4

3

2

1

Very High Risk

High Risk

Moderate Risk

Low risk

Very Low Risk

6

LULC

Active Channel

Flood Plain

Mix Vegetation

Agricultural Land

Barren Land

Settlement

Road

Sea

5

5

4

4

3

2

1

0

Very High Risk

Very High Risk

High Risk

High Risk

Moderate Risk

Moderate Risk

Low Risk

5

Flow accumulation

(Value)

Less than 51

51-102

102-153

153-204

More than 204

5

4

3

2

1

Very High Risk

High Risk

Moderate Risk

Low risk

Very Low Risk

4

Soil texture

Silty Clay Loam

Sandy Loam

Sandy

5

2

1

Very High Risk

Low Risk

Very Low Risk

3

Soil drainage

Well Drained

Moderately Well Drained

Poorly Drained

5

3

1

Very High Risk

Moderate Risk

Very Low Risk

2

Lithology

Alluvial Deposits

3

Moderate Risk

1

4. Results and Discussions

Database on nine different flood contributing factors, namely surface runoff, distance to the active channel, elevation, slope, land use/land cover, flow accumulation, soil texture, soil drainage, and lithology were selected based on the nature of the flood in the study area. These data sets were derived with the help of a digital elevation model, satellite image, rainfall, geology, Geobook, and PNGRIS GIS database of Papua New Guinea. ArcMap 10.5 was used to create these maps with the help of geographic data. Finally, a vulnerability assessment was performed with the help of local infrastructure and population distribution data sets, which were derived from the high-resolution LiDAR data sets of the area. The altitude database was prepared to utilize the 30m ASTER DEM data of the area. The maximum elevation is 93 meters above sea level observed in the northern part of the study area and the minimum altitude of 0 meters (close to the mean sea level) which is very prone to flood, found in the southern part of the basin where the Busu River empty into the Huon Golf (Figure 4a).

Figure 3: Methodological flowchart of Boolean logic for flood modeling

A flat slope, 0 to 3 degrees is identified in the middle part and the final section of the Busu catchment, which is demarked as a floodplain area. The maximum slope, more than 30 degrees is mapped in the northern part of the study area (Figure 4b). Rainfall is one of the major factors triggering surface runoff, which steers flood disasters in Papua New Guinea. As soon as rain strikes steeped ground it starts to stream overland downward. The surface runoff supplements the maximum amount of water in the river. A watershed is defined as a land area that generates runoff draining to a joint point. Based on the SCS model and curve number (CN) can estimate the surface runoff as a function of storm rainfall, hydrological soil group, and land use/land cover [69]. A three days (21/10/2012 to 23/10/2012) storm rainfall of 229.6 mm was considered as the input for surface runoff calculation. The SCS model derived a mean surface runoff of 191.59mm in the study area (Figure 4c). Supervised Classification was done with ERDAS Imagine 8.5 with the use of the high-resolution LiDAR satellite image to extract land use land cover characteristics of the study area. Through this classification, eight (8) different land use/land cover classes were extracted in the study area. These classes are active water channels, sea areas, mix-vegetation, agricultural land, floodplain, road, barren land, and settlement areas (Figure 4d). The lithological characteristics are an essential criterion for flood vulnerability assessment because it intensifies or makes less the degree of flood outcomes. Rocks with higher permeability coefficient devise minute resistance to water which permits excessive infiltration velocity and the rocks with lower permeability devise a greater resistance to water, which permits a smaller infiltration velocity. Only alluvial deposit is found in the study area (Figure 4e). The distance to the active channel was computed using the Euclidean distance method by 200 meters equally on each bank of the current channel with the help of spatial analyst tools in ArcMap10.5. After the distance to active channel analysis with Euclidean distance, the values were distributed into five classes concerning risk calculation (Figure 4f).

The rate and amount of water traveled through the soil are taken into consideration since they control the water movement. Soil textures, for example, loam, sandy loam, and clay made up the ratio of sand, silt, and clay number of particles that contribute to the mineral portion of the soil. Sand does not grasp much water while clay soils grasp much water. Textures do changes corresponding to the depth and there are three categories of this soil textures; uniform (same texture all through), texture contrast (sudden change in texture), and gradational (textures increase down the soil profile). In the determination of the soil textures, GIS and RS technologies come to aid with the spatial analysis, and the soil texture layer was used for the flood risk assessment. There are three soil textures identified in the study area. They are silty clay loam, sandy loam, and sand (Figure 4g). The three-soil texture falls under two major hydrological soil group (HSG). The sandy loam and sand fall under Group-A, and the silty clay loam under Group D. The soils within Group-A have a lower capacity for runoff, water passes freely by the soil, and the infiltration rate is greater than 0.3 inches/hr when wet. Group-A has less soil and more sand. Group D has a high capacity for runoff; the movements in water through the soils are much more constrained. The Group-D soils have more clay with sand associated and have clayey textures with a lower infiltration rate of 0 to 0.05 inches/hr.

Figure 4: Spatial distribution of flood conditioning factors; (a)Elevation in meter, (b) Slope in degree, (c) Surface runoff through SCS model, (d) Land use/land cover characteristics, (e) Lithological characteristics, (f) Distance from active channels, (g) Soil texture, (h) Soil drainage and (i) Flow accumulation

Soil drainage naturally talks about the occurrence and period of wet epochs under circumstances like those under which the soil developed. Fabricated adjustments to the water system through drainage or irrigation do not have much impact on the soil except a considerable change is made to the soil. Four soil drainage classes are identified according to the water tables, their positioning in the landscape, their wetness, and their form and structures in the study area. These groups are well drained, moderately well drained, poorly drained, and waterlogged (sea) (Figure 4h). Flow accumulation is one of the flood factors considered in this flood risk assessment.

The water from a cell flows towards the nearest down slope cell. Flow accumulation tools easily find out the accumulated flow with the known flow direction. All these methods concerning hydrology were performed using the hydrology tool extension in ArcGIS 10.5. The calculated flow accumulation values range from 0 to 255 (Figure 4i).

After the preparation of the input parameters different GIS models, like the Pairwise comparison, the ranking method, and the Boolean logic were used to develop flood susceptibility maps respectively. The resulting output through all the methods was reclassified into five (5) susceptibility zones, namely very high, high, moderate, low, and very low. Approximately 33.63% of the study areas are designated as very high susceptibility, 24.11% as high, 25.76% as moderate, 11.67% as low, and 4.83% as very low respectively through the Pairwise method (Table 6). The resulting data sets through the ranking method show 28.96% of the study areas as very high susceptibility, 18.14% as high, 23.40% as moderate, 21.22% as low, and 8.27% as very low respectively through the Pairwise method (Table 6). Most of the high susceptibility zones were identified in the middle portion and southern part of the study area where the active water channels are carrying water through it and empty into the Huon golf as per the resulting maps through the Pairwise method (Figure 5a) and Ranking method (Figure 5b) respectively. The Boolean logic derived a very significant result, which is very different from the first two methods. Only three classes were generated through the method, namely moderate (37.77%), low (39.95%), and very low (22.78%) respectively (Table 6 and Figure 5c).

Table 6: Statistical distribution of resulted flood susceptibility zones through Pairwise,

Flood susceptibility zones

Pairwise method

Ranking method

Boolean logic

Area (Km2)

Percentage

Area (Km2)

Percentage

Area (Km2)

Percentage

Very high

5.69

33.63

4.90

28.96

0

0

High

4.08

24.11

3.07

18.14

0

0

Moderate

4.36

25.76

3.96

23.40

6.39

37.77

Low

1.98

11.67

3.59

21.22

6.76

39.95

Very low

0.82

4.83

1.40

8.27

3.77

22.28

Total

16.92

100.0

16.92

100.00

16.92

100.00

Figure 5: Flood susceptibility zone based on (a) Pairwise comparison, (b) Ranking and (c) Boolean logic

Figure 6: Historical flood points and building footprints on the flood susceptibility map through Pairwise comparison method

Table 7: Comparison of flood points against each flood susceptibility mapping technique

Flood susceptibility zones

Pairwise comparison

Ranking

Boolean logic

Very high

17

12

None

High

15

16

None

Moderate

9

11

12

Low

5

7

22

Very low

None

None

12

Total flood points

46

46

46

Table 8: Population at risk based on vulnerability assessment

Flood susceptibility zones

Pairwise comparison

Ranking

Boolean

Household

Population

Household

Population

Household

Method

Very high

11

55

17

85

None

None

High

226

1130

189

945

None

None

Moderate

1213

6065

1134

5670

340

1700

Low

483

2415

667

3335

936

4680

Very low

122

610

48

240

779

3895

Flood validation is simply to check the authenticity of flood susceptibility analysis by using the existing flood point or through field verification by visiting the flood scene in the study area. Model validation is a key topic in flood risk analysis, as flood risk assessments are characterized by significant levels of uncertainty [70]. Validation is to evaluate risk valuation precision with the use of vulnerable parameters to higher susceptibility zones. It is also important for decision-makers to make decisions. There are many validation approaches to perform using the concerned parameters. Forty six (46) historical flood points were used to validate the susceptibility mapping through different GIS models. An attempt was taken to show the historical flood points on the resulted output through Pairwise comparison method (Figure 5). The overlay analysis suggests that thirty-two (32) flood points falls within the very high and high susceptibility zones and nine (9) flood points on moderate and five (5) flood points on low flood susceptibility zone (Figure 6 and Table 7). The similar analysis was performed for other two methods (Table 7).

The population is the most important parameter when doing a flood risk assessment and flood validation, thus it is eminent to deliberate on the population in the flood hazard zone. The building footprints were captured from LiDAR 30cm spatial resolution data sets of the area. Population data were gathered from the country’s national census and the national statistics office of Papua New Guinea. Alternatively, a survey was done around the Busu Plain to estimate the average household population. The survey suggests that on average five people are living in every household. The populations at risk were identified as per the degree of risk (Table 8). Figure 6 represents the building footprints on the resulted flood susceptibility zones derived through Pairwise comparison method.

5. Conclusions and Recommendation

The primary drive behind this work was to model inland floods using multiple approaches or multi-criteria analysis. Such approaches are; the Pairwise Comparison, the Ranking, and the Boolean logic, to establish flood assessment maps that identified areas that are likely to be affected by flooding and the degree of harm that can be caused to the local population. By using all the available data with the use of GIS software and other supplementary software, a better flood risk assessment was done which is user-friendly. These flood risk maps especially from Pairwise and Ranking can be used for flood mitigation and flood hazard preparedness. The MCA techniques offer a practicable approach for analyzing and implementing the influences of different factors, including multiple levels of contingent, autonomous, analytical, and measurable data after combining high-resolution satellite data (LiDAR), demographic data, topographic datasets, national soil data, and weather data. The DEM was used to categorize the areas of threat in the study vicinity. The defined flood impact parameters are precipitation, elevation and slope, geology, land-use land cover, soil, and other geomorphic characteristics. Thematic maps were generated by using various image analysis methods including specific GIS analysis using several GIS software. Each one of the thematic layers was analyzed by using the weighted overlay, weighted sum, and Boolean and operations to generate the final flood susceptibility maps. The weights are assigned by taking into account the flood susceptibility and vulnerability of the region. The three flood assessment maps generated were done using three different methods, namely Pairwise comparisons, Ranking and Boolean. In the Pairwise Comparison method, the criteria weights were generated by filling the comparison matrix, normalizing the matrix, and then calculated the priority vector. In order for the criteria weights to be accepted, a consistency analysis is done. In the Ranking method approach, the flood factors were reclassified and ranked. The ranking values are ranged from 1 to 5 according to their relative importance to the flood, where 1 is the very low risk and 5 is the very high risk. In the final method, Boolean logic the flood factors were reclassified again after the first reclassification. The second reclassification was done since Boolean operations only interpret integers. Boolean and operation was used for paired flood factors except Surface runoff. The validation process suggests the level of acceptability of each model used for flood susceptibility mapping after overlaying historical flood points on the resulted flood susceptibility database. Very high, high and moderate susceptibility zones were considered as positive for flooding zone, while on the other hand low and very low susceptibility zones were ignored for flooding. The analysis (Table 7) represents 41 (17 + 15 + 9) historical flood points falls within the positive flooding zone out of 46 historical flood points based on Pairwise comparison technique, which leads 89.13% accuracy. The accuracy of 84.78% and 26.09% were calculated for Ranking and Boolean logic respectively. The Boolean logic method is found not to be an accurate method for flood risk assessment as per the result compared to Pairwise comparisons and Ranking method.

Flood assessment maps are very vital as they generate a roadmap for the needed flood mitigation and preparedness methods. In all three approaches, the value of each factor was measured by determining how much each factor contributed to the flood. The monitoring and treatment of flood risk is fundamentally part of social and environmental growth, which targets reducing the loss of life, and natural and manmade resources that are of benefit to individuals. Flood Mitigation. Busu flood is real, real damages were done. The people living around the Busu River are victims of flood and have been living in fear for many years. The government authority must now consider both structural and non-structural flood mitigation measures. It is identified that even with considerable use of structural flood mitigation methods, the flood damages were still devastating due to the flood factors involved. The government must now initiate multifaceted flood management in Morobe Province as well as other parts of Papua New Guinea, by stretching out the use of nonstructural means of flood risk management. This study suggests that MCA is a better decision-making method and must be considered for better solutions by the local authority.

References

[1] Tran, P., Marincioni, F. and Shaw, R., (2010). Catastrophic Flood and Cover change in the Huong RiverBasin, Central Viet Nam: A Gap between Common Perceptions and Facts. Journal of Environmental Management , Vol. 91, 2186-2200. https://doi.org/10.1016/j.jenvman.2010.05.020.

[2] Munchener, M., (1999). Topics 2000, National Catastrophes - the Current Position, Special Millennium Issue. https://www.imia.com/wp-content/uploads/2013/05/EP17-2003-Loss-Potential-of-Natural-Hazzards-sm.pdf.

[3] Simonovic, S., (2009). A New Method for Spatial and Temporal Analysis of Risk in Water Resources Management. Journal of Hydro informatics , Vol. 11(3,4), 320-329. https://doi.org/10.2166/hydro.2009.018.

[4] Jeb, D. N. and Aggarwal, S. P., (2008). Flood Inundation Hazard Modeling of the River Kaduna Using Remote Sensing and Geographic Information Systems. Journal of Applied Sciences Research, Vol. 4(12), 1822 -1833. https://www.aensiweb.com/old/jasr/jasr/2008/1822-1833.pdf.

[5] Huang, X., Tan, H., Zhou, J., Yang, T., Benjamin, A., Wen, S. W., Li, S., Liu, A., Li, X., Fen, S. and Li, X., (2008). Flood Hazard in Hunan Province of China: An Economic Loss Analysis. Natural Hazards, Vol. 47, 65-73.

[6] Veerbeek, W. and Zevenbergen, C., (2009). Deconstructing Urban Flood Damages: Increasing the Expressiveness of Flood Damage Models Combining a High Level of Detail with a Broad Attribute Set. Journal of Flood Risk Management , Vol. 2(1), 45-57. https://doi.org/10.1111/j.1753-318X.2009.01021.x.

[7] Merz, B., Kreibich, H., Schwarze, R. and Thieken, A., (2010). Assessment of Economic Flood Damage. Natural Hazards and Earth System Sciences , Vol. 10(8), 1697-1724. https://doi.org/10.5194/nhess-10-1697-2010.

[8] Dewan, A., (2013). Floods in a Megacity: Geospatial Techniques . Assessing Hazards, Risk and Vulnerability, Dordrecht: Springer, Vol. 6, 1-199. https://doi.org/10.1007/978-94-007-5875-9.

[9] Markantonis, V., Meyer, V. and Lienhoop, N., (2013). Evaluation of the Environmental Impacts of Extreme Floods in the Evros River Basin using Contingent Valuation Method. Natural Hazards, Vol. 69, 1535-1549. https://doi.org/10.1007/s11069-013-0762-3.

[10] Hudson, P. G. M. B., Botzen, W. J. W., Kreibich, H., Bubeck, P. and Aerts, J. C. J. H., (2014). Evaluating the Effectiveness of Flood Damage Mitigation Measures by the Application of Propensity Score Matching. Natural Hazards and Earth System Sciences, Vol. 14(7), 1731-1747.

[11] Perera, E. D. P. and Lahat, L., (2015). Fuzzy Logic Based Flood Forecasting Model for the Kelantan River Basin, Malaysia. Journal of Hydro-Environment Research , Vol. 9(4), 542-553. https://doi.org/10.1016/j.jher.2014.12.001.

[12] Adeoye, N. O., Ayanlade, A. and Babatimehin, O., (2009). Climate Change and Menace of Floods in Nigerian Cities: Socio-Economic Implications. Advances in Natural and Applied Sciences, Vol. 3(3), 369-378.

[13] Youssef, A. M., Pradhan, B. and Hassan, A. M., (2011). Flash Flood Risk Estimation along the St. Katherine Road, Southern Sinai, Egypt using GIS Based Morphometry and Satellite Imagery. Environmental Earth Sciences , Vol. 62(3), 611-623. https://doi.org/10.1007/s12665-010-0551-1.

[14] Taylor, J. man Lai, K., Davies, M., Clifton, D., Ridley, I. and Biddulph, P., (2011). Flood Management: Prediction of Microbial Contamination in Large-Scale Floods in Urban Environments. Environment International , Vol. 37(5), 1019-1029. https://doi.org/10.1007/s12665-010-0551-1.

[15] Dawod, G. M., Mirza, M. N. and Al-Ghamdi, K. A., (2012). GIS-Based Estimation of Flood Hazard Impacts on Road Network in Makkah City, Saudi Arabia. Environmental Earth Sciences, Vol. 67, 2205-2215. https://doi.org/10.1007/s12665-012-1660-9.

[16] Foudi, S., Osés-Eraso, N. and Tamayo, I., (2015). Integrated Spatial Flood Risk Assessment: The Case of Zaragoza. Land Use Policy , Vol. 42, 278-292. https://doi.org/10.1016/j.landusepol.2014.08.002.

[17] Kwak, Y. and Kondoh, A., (2008). A Study on the Extraction of Multi-Factor Influencing Floods from RS Image and GIS Data; A Case Study in Nackdong Basin, S. Korea . The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, ISPRS Congress Beijing, Vol. 37, 421-426. https://www.isprs.org/proceedings/xxxvii/congress/8_pdf/2_wg-viii-2/50.pdf.

[18] Kaittila, S. and Yambui, A., (1996). Disaster Management and Government Intervention in PNG: the Case of Lae. Disasters, Vol. 20(1), 61-67. https://doi.org/10.1111/j.1467-7717.1996.tb00515.x.

[19] Samanta, S., Koloa, C., Pal, D. K.and Palsamanta, B., (2016). Flood Risk Analysis in Lower Part of Markham River Based on Multi-Criteria Decision Approach (MCDA). Hydrology, Vol. 3(3). https://doi.org/10.3390/hydrology3030029.

[20] Luu, C. and Von Meding, J., (2018). A Flood Risk Assessment of Quang Nam, Vietnam Using Spatial Multicriteria Decision Analysis. Water, Vol. 10(4). https://doi.org/10.3390/w10040461.

[21] Papaioannou, G., Vasiliades, L. and Loukas, A., (2015). Multi-Criteria Analysis Framework for Potential Flood Prone Areas Mapping. Water Resources Management, Vol. 29, 399-418. https://doi.org/10.1007/s11269-014-0817-6.

[22] Yang, X. L., Ding, J. H. and Hou, H., (2013). Application of a Triangular Fuzzy AHP Approach for Flood Risk Evaluation and Response Measures Analysis. Natural Hazards, Vol. 68, 657-674. https://doi.org/10.1007/s11069-013-0642-x.

[23] Sanyal, J. and Lu, X. X., (2004). Application of Remote Sensing in Flood Management with Special Reference to Monsoon Asia: A Review. Natural Hazards, Vol. 33, 283-301. https://doi.org/10.1023/B:NHAZ.0000037035.65105.95.

[24] Haq, M., Akhtar, M., Muhammad, S., Paras, S. and Rahmatullah, J., (2012). Techniques of Remote Sensing and GIS for Flood Monitoring and Damage Assessment: A Case Study of Sindh Province, Pakistan. The Egyptian Journal of Remote Sensing and Space Science , Vol. 15(2), 135-141. https://doi.org/10.1016/j.ejrs.2012.07.002.

[25] Patel, D. P. and Srivastava, P. K., (2013). Flood Hazards Mitigation Analysis Using Remote Sensing and GIS: Correspondence with Town Planning Scheme. Water Resources Management, Vol. 27(7), 2353-2368. https://doi.org/10.1007/s11269-013-0291-6.

[26] Tehrany, M. S., Pradhan, B. and Jebur, M. N., (2014). Flood Susceptibility Mapping Using a Novel Ensemble Weights-Of-Evidence and Support Vector Machine Models in GIS. Journal of Hydrology, Vol. 512, 332-343. https://doi.org/10.1016/j.jhydrol.2014.03.008.

[27] Wanders, N., Karssenberg, D., De Roo, A., De Jong, S. M. and Bierkens, M. F. P., (2014). The Suitability of Remotely Sensed Soil Moisture for Improving Operational Flood Forecasting. Hydrology and Earth System Sciences , Vol. 18(6), 2343-2357. https://doi.org/10.5194/hess-18-2343-2014.

[28] Rahmati, O., Pourghasemi, H. R. and Zeinivand, H., (2016). Flood Susceptibility Mapping Using Frequency Ratio and Weights-of-Evidence Models in the Golastan Province, Iran. Geocarto International, Vol. 31(1), 42-70. https://doi.org/10.1080/10106049.2015.1041559.

[29] Oeurng, C., Sauvage, S. and Sánchez-Pérez, J. M., (2011). Assessment of Hydrology, Sediment and Particulate Organic Carbon Yield in a Large Agricultural Catchment Using the SWAT Model. Journal of Hydrology , Vol. 401(3-4), 145-153. https://doi.org/10.1016/j.jhydrol.2011.02.017.

[30] Bahremand, A., De Smedt, F., Corluy, J., Liu, Y. B., Poorova, J., Velcicka, L. and Kunikova, E., (2007). WetSpaModel Application for Assessing Reforestation Impacts on Floods in Margecany–Hornad Watershed, Slovakia. Water Resources Management, Vol. 21, 1373-1391. https://doi.org/10.1007/s11269-006-9089-0.

[31] Rojas, R., Feyen, L. and Watkiss, P., (2013). Climate Change and River Floods in the European Union: Socio-economic Consequences and the Costs and Benefits of Adaptation. Global Environmental Change, Vol. 23(6), 737-1751. https://doi.org/10.1016/j.gloenvcha.2013.08.006.

[32] Xia, X., Liang, Q., Ming, X. and Hou, J., (2017). An Efficient and Stable Hydrodynamic Model with Novel Source Term Discretization Schemes for Overland Flow and flood Simulations. Water Resources Research , Vol. 53(5), 3730-3759. https://doi.org/10.1002/2016WR020055.

[33] Bout, V. B. and Jetten, V. G., (2018). The Validity of Flow Approximations when Simulating Catchment-Integrated Flash Floods. Journal of Hydrology , Vol. 556, 674-688. https://doi.org/10.1016/j.jhydrol.2017.11.033.

[34] Dahal, R. K., Hasegawa, S., Nonomura, A., Yamanaka, M., Masuda, T. and Nishino, K., (2008). GIS-Based Weights-Of-Evidence Modelling of Rainfall-Induced Landslides in Small Catchments for Landslide Susceptibility Mapping. Environmental Geology, Vol. 54, 311-324. https://doi.org/10.1007/s00254-007-0818-3.

[35] Tehrany, M. S. and Jones, S., (2017). Evaluating the Variations in the Flood Susceptibility Maps Accuracies due to the Alterations in the Type and Extend of the Flood Inventory. Proceedings of the International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, 209-214. https://doi.org/10.5194/isprs-archives-XLII-4-W5-209-2017.

[36] Al-Juaidi, A. E., Nassar, A. M. and Al-Juaidi, O. E., (2018). Evaluation of Flood Susceptibility Mapping Using Logistic Regression and GIS Conditioning Factors. Arabian Journal of Geosciences, Vol. 11, 1-10. https://doi.org/10.1007/s12517-018-4095-0.

[37] Nandi, A., Mandal, A., Wilson, M.and Smith, D., (2016). Flood Hazard Mapping in Jamaica using Principal Component Analysis and Logistic Regression. Environmental Earth Sciences, Vol. 75, 1-16. https://doi.org/10.1007/s12665-016-5323-0.

[38] Saaty, T.L., (2008). Decision Making with the Analytic Hierarchy Process. International Journal of Services Sciences, Vol. 1(1), 83-98. https://doi.org/10.1504/IJSSCI.2008.017590.

[39] Chen, Y.R., Yeh, C.H. and Yu, B., (2011). Integrated Application of the Analytic Hierarchy Process and the Geographic Information System for Flood Risk Assessment and Flood Plain Management in Taiwan. Natural Hazards , Vol. 59, 1261-1276. https://doi.org/10.1007/s11069-011-9831-7.

[40] Ouma, Y.O., and Tateishi, R., (2014). Urban Flood Vulnerability and Risk Mapping Using Integrated Multi-Parametric AHP and GIS: Methodological Overview and Case Study Assessment. Water, Vol. 6, 1515–1545. https://doi.org/10.3390/w6061515.

[41] Kazakis, N., Kougias, I. and Patsialis, T., (2015). Assessment of Flood Hazard Areas at a Regional Scale Using an Index-Based Approach and Analytical Hierarchy Process: Application in Rhodope–Evros Region, Greece. Science of the Total Environment, Vol. 538, 555-563. https://doi.org/10.1016/j.scitotenv.2015.08.055.

[42] Tehrany, M. S., Pradhan, B. and Jebur, M. N., (2015). Flood Susceptibility Analysis and Its Verification Using a Novel Ensemble Support Vector Machine and Frequency Ratio Method. Stochastic Environmental Research and Risk Assessment , Vol. 29, 1149-1165.

[43] Samanta, S., Pal, D. K. and Palsamanta, B., (2018). Flood Susceptibility Analysis through Remote Sensing, GIS and Frequency Ratio Model. Applied Water Science, Vol. 8(2), 1-14. https://doi.org/10.1007/s13201-018-0710-1.

[44] Zou, Q., Zhou, J., Zhou, C., Song, L. and Guo, J., (2013). Comprehensive Flood Risk Assessment Based on Set Pair Analysis-Variable Fuzzy Sets Model and Fuzzy AHP. Stochastic Environmental Research and Risk Assessment , Vol. 27, 525-546. https://doi.org/10.1007/s00477-012-0598-5.

[45] Campolo, M., Soldati, A. and Andreussi, P., (2003). Artificial Neural Network Approach to Flood Forecasting in the River Arno. Hydrological Sciences Journal , Vol. 48(3), 381-398. https://doi.org/10.1623/hysj.48.3.381.45286.

[46] Kia, M. B., Pirasteh, S., Pradhan, B., Mahmud, A. R., Sulaiman, W. N. A. and Moradi, A., (2012). An Artificial Neural Network Model for Flood Simulation using GIS: Johor River Basin, Malaysia. Environmental Earth Sciences , Vol. 67, 251-264. https://doi.org/10.1007/s12665-011-1504-z.

[47] Tehrany, M. S., Pradhan, B. and Jebur, M. N., (2013). Spatial Prediction of Flood Susceptible Areas Using Rule Based Decision Tree (DT) and a Novel Ensemble Bivariate and Multivariate Statistical Models in GIS. Journal of Hydrology, Vol. 504, 69-79. https://doi.org/10.1016/j.jhydrol.2013.09.034.

[48] Sahana, M., Rehman, S., Sajjad, H. and Hong, H., (2020). Exploring Effectiveness of Frequency Ratio and Support Vector Machine Models in Storm Surge Flood Susceptibility Assessment: A Study of Sundarban Biosphere Reserve, India. Catena, Vol. 189. https://doi.org/10.1016/j.catena.2019.104450.

[49] Mojaddadi, H., Pradhan, B., Nampak, H., Ahmad, N. and Ghazali, A. H. B., (2017). Ensemble Machine-Learning-Based Geospatial Approach for Flood Risk Assessment Using Multi-Sensor Remote-Sensing Data and GIS. Geomatics, Natural Hazards and Risk, Vol. 8(2), 1080-1102. https://doi.org/10.1080/19475705.2017.1294113.

[50] Khosravi, K., Panahi, M. and Tien Bui, D., (2018). Spatial Prediction of Groundwater Spring Potential Mapping Based on an Adaptive Neuro-Fuzzy Inference System and Metaheuristic Optimization. Hydrology and Earth System Sciences , Vol. 22(9), 4771-4792. https://doi.org/10.5194/hess-22-4771-2018.

[51] Ahmadlou, M., Karimi, M., Alizadeh, S., Shirzadi, A., Parvinnejhad, D., Shahabi, H. and Panahi, M., (2019). Flood Susceptibility Assessment Using Integration of Adaptive Network-Based Fuzzy Inference System (ANFIS) and Biogeography-Based Optimization (BBO) and BAT Algorithms (BA). Geocarto International, Vol. 34(11), 1252-1272. https://doi.org/10.1080/10106049.2018.1474276.

[52] Shafizadeh-Moghadam, H., Valavi, R., Shahabi, H., Chapi, K. and Shirzadi, A., (2018). Novel Forecasting Approaches Using Combination of Machine Learning and Statistical Models for Flood Susceptibility Mapping. Journal of Environmental Management, Vol. 217, 1-11. https://doi.org/10.1016/j.jenvman.2018.03.089.

[53] Ghanbarpour, M. R., Salimi, S. and Hipel, K. W., (2013). A Comparative Evaluation of Flood Mitigation Alternatives Using GIS‐Based River Hydraulics Modelling and Multicriteria Decision Analysis. Journal of Flood Risk Management , Vol. 6(4), 319-331. https://doi.org/10.1111/jfr3.12017.

[54] Malczewski, J., (2006). GIS‐based Multicriteria Decision Analysis: A Survey of the Literature. International Journal of Geographical Information Science , Vol. 20(7), 703-726. https://doi.org/10.1080/13658810600661508.

[55] Hwang, C. L. and Lin, M. J., (2012). Group Decision Making Under Multiple Criteria: Methods and Applications. Lecture Notes in Economics and Mathematical Systems , Vol. 281.

[56] Sekac, T. and Jana, K. S., (2014). Change Detection of Busu River Course in Papua New Guinea –Impact on Local Settlements Using Remote Sensing and GIS Technology. International Journal of Scientific and Engineering Research. Vol. 5(8), 891-899.

[57] PNGRIS-Papua New Guinea Resource Information System, (2009). The Land-Use Section, Science and Technology Branch, Department of Agriculture and Livestock . 3rd ed.; University of Papua New Guinea: Boroko, Papua New Guinea.

[58] Geobook, (2009). Provincial GIS-based Planning Tools. The Remote Sensing Centre of University of Papua New Guinea: Port Moresby, Papua New Guinea.

[59] Beullens, J., Velde D. V. and Nyssen, J., (2014). Impact of Slope Aspect on Hydrological Rainfall and on the Magnitude of Rill Erosion in Belgium and Northern France. Catena, Vol. 114, 129-139. https://doi.org/10.1016/j.catena.2013.10.016.

[60] Samanta, S. and Pal, B., (2011). A Case Study of Soil Loss Estimation using RUSLE, Remote Sensing and GIS in Sankwep Catchment of Papua New Guinea. Niugini Agrisaiens,Vol. 3, 50-58.

[61] Mahmoud, S. H. and Gan, T. Y., (2018). Multi-Criteria Approach to Develop Flood Susceptibility Maps in Arid Regions of Middle East. Journal of Cleaner Production , Vol.196, 216-229.
https://doi.org/10.1016/j.jclepro.2018.06.047
.

[62] Morea, H. and Samanta, S., (2020). Multi-Criteria Decision Approach to Identify Flood Vulnerability Zones Using Geospatial Technology in the Kemp-Welch Catchment, Central Province, Papua New Guinea. Appl Geomat, Vol. 12, 427–440. https://doi.org/10.1007/s12518-020-00315-6.

[63] Harley, P. and Samanta, S., (2018). Modeling of Inland Flood Vulnerability Zones through Remote Sensing and GIS Techniques in the Highland Region of Papua New Guinea. Appl Geomat, Vol. 10, 159–171. https://doi.org/10.1007/s12518-018-0220-8.

[64] Saaty, T. L., (1980). The Analytical Hierarchy Process. McGraw Hill, New York.

[65] Saaty, T. L., (1977). A Scaling Method for Priorities in Hierarchical Structures. The Journal of Mathematical Psychology , Vol. 15(3), 234-281.

[66] Roopnarine, R., Opadeyi, J., Eudoxie, G., Thongs, G. and Edwards, E., (2018). GIS-based Flood Susceptibility and Risk Mapping Trinidad Using Weight Factor Modeling. Caribb. J. Earth Sci., Vol. 49, 1–9. http://caribjes.com/CJESpdf/CJES49-1-TrinidadFloodRisk.pdf.

[67] Cheng, C. and Thompson, R. G., (2016). Application of Boolean Logic and GIS for Determining Suitable Locations for Temporary Disaster Waste Management Sites. International Journal of Disaster Risk Reduction, Vol. 20, 78-92. https://doi.org/10.1016/j.ijdrr.2016.10.011.

[68] Riad, P. H. S., Billib, M., Hassan, A. A., Salam, M. A. and El Din, M. N., (2011). Application of the Overlay Weighted Model and Boolean Logic to Determine the Best Locations for Artificial Recharge of Groundwater. Journal of Urban and Environmental Engineering, Vol. 5(2), 57–66. https://www.jstor.org/stable/26203357.

[69] Pal, B. and Samanta, S., (2011). Surface Runoff Estimation and Mapping Using Remote Sensing and Geographic Information System. International Journal of Advances in Science and Technology , Vol. 3(2), 106-114.

[70] Handmer, J., (2003). The Chimera of Precision: Inherent Uncertainties in Disaster Loss Assessment. The Australian Journal of Emergency Management , Vol. 18(2), 88-97.