Application of geographical information system-based analytical hierarchy process modeling for flood susceptibility mapping of Krishna District in Andhra Pradesh

Flooding is one of the most catastrophic natural disasters in terms of provoking socio-economic losses. The current study is to foster a flood susceptibility map of Krishna District in Andhra Pradesh (AP) through integrating remote sensing data, geographical information system (GIS), and the analytical hierarchy process (AHP). Eleven factors, including elevation, slope, aspect, land use/land cover (LULC), drainage density, topographic wetness index, stream power index, lithology, soil, precipitation, and distance from the streams, are considered for identifying and evaluating the spatial distribution of critical flood-susceptible regions. Thematic maps of different factors were derived in GIS using remote sensing data obtained from Sentinel-2A (satellite sensor), shuttle radar topography mission digital elevation model (SRTM DEM v3), and other scientific data products. An analytical hierarchy process is a mathematical approach for decision support, primarily based on the weight and rank of different causative factors. AHP technique is implemented for flood hazard modeling and ascertaining the Flood Hazard Index (FHI) to produce a flood susceptibility map. Different thematic maps weighed with the AHP framework are combined using overlay analysis to produce the flood susceptibility map of the study region. The outcomes of the study demonstrate the potential of GIS and AHP in providing a premise to recognize the vulnerable areas that are susceptible to flood. According to the findings, the Flood Hazard Index is 42% and the study region is classified into very high, high, moderate, low, and very low susceptible, respectively. Following that, historical flood data was used to validate the accuracy of the generated flood susceptibility map. This shows that a maximum of 90% of the data points are within floodplain.


Introduction
Flooding often termed an inundation, which is a hydrometeorological disaster that usually submerges dry land, is one of the most catastrophic natural hazards globally causing significant losses every year (Khosravi et al. 2019;Lawal et al. 2014;Mandal et al. 2021).It tends to happen, typically within 1 h of precipitation in the form of heavy rainfall (Mohammadi et al. 2020;Zheng et al. 2012;Petropoulos and Islam 2017), and results in human death, infrastructural destruction, environmental degradation, and loss of livelihood (Khosravi et al. 2019;P. G. 2013;Das 2020).
The flood events continue to increase every year as a result of several issues, including climate change, and the rapid development of urbanization, civilization, and abnormal rainfalls are becoming much more prevalent today, posing tremendous flood risks and threatening infrastructure and livelihoods (Khosravi et al. 2019;Uddin et al. 2013;Yousefi et al. 2020;Shit et al. 2021).
Flooding-related infrastructure damage has led to a change in the water supply and inadequate service delivery in the fields of wastewater treatment, electricity, transportation, communication, education, and health care (Petropoulos and Islam 2017; Talha et al. 2019;Darwin et al. 2018;Hammami et al. 2019).Many flood protection acts have been launched and implemented but unfortunately, many developments have not been fully implemented (Patel and Yadav 2021;Natarajan et al. 2021).Therefore, it is necessary to predict the flood risk with the aid of imposing appropriate strategies within the study area to reduce the impact of the disaster.
Flood susceptibility of different regions across the world have been studied by different researchers (Ahmadlou et al. 2021;Khosravi et al. 2019;Patel and Yadav 2021;Petropoulos and Islam 2017;Talha et al. 2019), and several techniques and methods have been explored in order to simulate flooding (Asim et al. 2022;Daneshparvar et al. 2022;Seejata et al. 2018).Multi-Criteria Decision-Making analysis and machine learning techniques are adopted for flood susceptibility mapping to reduce losses arising from floods (Khosravi et al. 2019;Swain et al. 2020;Ramu et al. 2020).Some of the most powerful machine learning algorithms (Rahman et al. 2019) and autoencoder neural networks that focused on reducing flood hazards include hybrid autoencoder-MLP (multilayer perceptron) (Ahmadlou et al. 2021) and MLP neural networks (Ahmadlou et al. 2021).Fuzzy mathematical models have been developed in order to enhance the probability estimate and evaluation of flood risk with inadequate data sets (Li et al. 2012).With the help of GIS and some analysis techniques, however, strategies such as Multi-Criteria Decision-Making analysis (Ibanga and Idehen 2020;Souissi et al. 2020;Allafta and Opp 2021;Dash and Sar 2020), machine learning, hybrid autoencoder-MLP (multilayer perceptron) (Ahmadlou et al. 2021), MLP neural networks (Ahmadlou et al. 2021), frequency ratio method (Ullah and Zhang 2020), fuzzy mathematical models (Ramu et al. 2020;Talha et al. 2019), and artificial neural network techniques and processes produce the most precise findings and weights as close to reality as possible.
Because of the timing and location of occurrence, geophysical interactions, and geographical expanse of floods, it is not feasible to entirely manage floods (Patel and Yadav 2021;Demir and Kisi 2016).Complete safety and flood protection are rarely considered viable choices (Seejata et al. 2018;Pham et al. 2021;Hashim et al. 2017).Advances in technology and remote sensing are essentially protecting us from terrible disasters and loss of life (Talha et al. 2019;Liuzzo et al. 2019;Bourenane et al. 2019).
Remote sensing is the study of gathering data about an object or an event without making direct contact with it (Uddin et al. 2013;Asim et al. 2022;Vilasan and Kapse 2021).The GIS tool uses this data to do numerous analyses and expose the results to various needs via maps, reports, charts, and tables, which are provided in printed or digital format (Ibanga and Idehen 2020;Souissi et al. 2020;Ullah and Zhang 2020;Lawal et al. 2014;Aravinthasamy et al. 2021).
Flood estimating is a challenging and complex subject in hydrology (Elkhrachy 2015).It plays a vital role in reducing human tragedy and economic loss caused by floods (Nahiduzzaman et al. 2015).The dependability of flood forecasting has improved in recent years as a result of advances in data collection via satellite observations, integration of hydrological and meteorological modeling capabilities for uncertainty analysis, and communication (Razavi et al. 2018;Demir and Kisi 2016;Danumah et al. 2016).It can also serve as a crucial foundation for emergency services when it comes to implementing an innovative strategy to prevent and mitigate future floods.
Recently, the advancement of GIS tools and the availability of remote sensing data sources have substantially improved the formulation and implementation of prediction models for natural disaster-prone locations (Ravinder and Ramu 2020).Geospatial tools now enable a diverse set of data sources for high-quality flood simulation (Ravinder and Ramu 2020).Keeping this as an essential part the main objective of this study is to derive the extent of flood susceptibility zones in Krishna District, to evaluate which region needs the greatest involvement in the creation of risk reduction or mitigation techniques through the GIS-based AHP modeling.Hamsaladeevi is where the Krishna River joins the ocean.Due to the geographical conditions, as well as the frequent cyclonic depression events in Krishna District, cyclones occur frequently here.In addition to these cyclones, upstream rainfall also contributes to floods from Krishna.During the past two decades, 19 floods have occurred in Krishna District.It is one of the most vulnerable regions for cyclones (Rao et al. 2007).Many previous studies have used Multi-Criteria Decision-Making techniques, such as AHP, to estimate the potential flood zones using remote sensing and GIS.Previous studies over Krishna District have used factors such as distance from river, rainfall, and socio-economic factors (Sai et al. 2021) for developing flood vulnerability index over Krishna District.Other studies conducted over Krishna District in the aspect of flood hazard potential of the Krishna river basin have not implemented AHP and GIS for flood hazard assessment.However, most of the studies considered river basin-wise implementation, and have considered different factors based on the data availability.Here in this study we have made use of 11 factors for implementing AHP.Similar to those studies, 11 causative factors, elevation, slope, aspect, land use/land cover (LULC), drainage density, topographic wetness index, stream power index, lithology, soil, precipitation, and distance from the streams, have been selected for the mapping the flood hazard potential over Krishna region.In addition to mapping the flood hazard zones, the region is classified into different zones based on hazard potential index.The conclusions from this study may be utilized as a decision-making tool to improve flood hazard control plans and projects.

Study area
Krishna District is located in the coastal area of Andhra Pradesh (16.6100°N, 80.7214°E).According to the 2011 Indian census, it has an area of around 8727 km 2 (3370 mi 2 ), a total coastline of 88 km (55 mi), and a population of 4,529,009 people.It is surrounded on the east by West Godavari and on the south by the Bay of Bengal.The Krishna River, India's third longest, travels through the state of Andhra Pradesh before emptying into the Bay of Bengal at Hamsaladevi hamlet in the Krishna District.The district's climatic characteristics may be characterized as tropical, with extremely hot summers and fairly hot winters.The Southwest monsoons give roughly 1028 mm of annual rainfall to the region, and the area is covered with three types of soils: black cotton (57.6%), sand clay loams (22.3%), and red loams (19.4%).According to historical data from the study area, catastrophic flash floods hit various parts of the Krishna District in 1998, 2009, and 2019, causing substantial loss of life and property.As a result, flood susceptibility mapping is critical in this area for risk management and flood mitigation.The study area location map is shown in Fig. 1.

Data preparation
The methodology included in this study is depicted in the Fig. 2.However, only a subset of factors was overlaid based on their significance in flood area selection.For inundation zoning, 11 thematic maps were considered, including elevation, slope, aspect, land use/land cover (LULC), drainage density, topographic wetness index, stream power index, lithology, soil, precipitation, and distance from the streams, for identifying and evaluating the spatial distribution of critical flood-susceptible regions.
In this study, the Shuttle Radar Topography Mission (SRTM) global 1 arc-second (30 m) DEM (digital elevation model) data is retrieved from USGS Earth Explorer using Krishna District coordinates with a cloud cover of less than 10%.The raw DEM data is imported into Arc-GIS 10.3, and an elevation map of the research area is established.The slope % results were computed using the spatial analysis tool, and further, the DEM was then used to generate the aspect map and distance from the stream map for the area.Land use/land cover (LULC) was acquired using the map, which is created from ESA Sentinel-2 imagery at 10-m resolution on a scale of 1/5,700,000 and mosaiced using data management tools.The research area is then extracted using the mask command.The USGS provides the lithology shape file for South Asia, and the lithology map for the research region is derived using the spatial analysis operation in ArcGIS Preparing rainfall maps is essential for identifying flood-prone areas, but there are several steps to this process.The data was gathered from the APSDPS website, and then interpolated by selecting IDW.However, the resultant map of this research revealed that the biannual precipitation ranges between 751 and 1138 mm.These were categorized into four classes: 800 mm, 800-950 mm, 950-1100 mm, and > 1100 mm.
The topographic wetness index (TWI) is a measure of wetness distribution that regulates water flow across the land.It is a standard measure that assesses the potential of water accumulating in certain areas.A high index number implies that there is a lot of water that may be collected because of the low slope, and vice versa.As a result, places with a higher TWI are more susceptible to flooding.The TWI of the study area was derived from the SRTM DEM.Finally, the topographic wetness index (TWI) is computed using Eq. ( 1): where α is the catchment area and β is the slope.SPI, which is the product of the catchment area and slope, is used to calculate the erosive force of the overland flow.The SPI may be used to identify relevant locations for soil conservation measures, lowering the impact of concentrated surface runoff.Because of the increased stream power in the upstream stretch, the streams can erode and transfer a substantial amount of debris.When stream power is reduced, the channels become shallow and meandering, resulting in silt deposition on the overbank.On the lower plains, this is a major cause of floods.As a result, places with a lower SPI are more likely to flood.The SPI of the study area was derived from the (1) TWI = ln( ∕tan ) SRTM DEM.Finally, the steam power index is calculated by Eq. ( 2): where α is the catchment area and β is the slope.Following that, thematic maps were created for each criterion by performing spatial analysis in GIS.These maps were reclassified using the weights derived from AHP technique and are utilized in overlay analysis to generate the Flood Hazard Index (FHI) for the entire district region.

Analytical hierarchy process (AHP)
The AHP approach is used to determine the impact of the flood on each element (Daneshparvar et al. 2022;Samanta et al. 2018;Swain et al. 2020), and a weight value equal to its relative relevance is applied to it.Saaty's pairwise comparison approach shown in Table 1 was used to calculate this weight, which he invented in 1980 (Das 2020;Hammami et al. 2019;Naik et al. 2019).An n × n dimensional (2) SPI = × tab  pairwise comparison matrix is used to compare the conditioning factors.Each component in the matrix was assigned an arithmetic value between 1 and 9 based on its significance in relation to the other element.The arithmetic value 1 indicates that both components are equally important; however, value 9 indicates that the corresponding column factor is extremely important.
A pairwise comparison matrix is created, and the consistency ratio (CR) is determined for the selected factors affecting the flood susceptibility using Eq. ( 3).If the CR is more than 0.1 (i.e., 10%), the judgments may be too inconsistent to be relied on.However, if the consistency ratio (CR) is smaller than 0.1, we may infer that our matrix is perfectly consistent and can proceed with the decisionmaking process utilizing AHP (Table 2).
where CI is the consistency index and RI is the random index.
where n is number of factors and λ max is the average value of the consistency vector.
The 11 components were employed in this work for flood hazard mapping.Each parameter was separated into appropriate subclasses using reclassification, weights were assigned, and their normalized weights were applied to estimate the Flood Hazard Index using Eq. ( 5).The gathered data is analyzed in a GIS tool.
(3) CR = CI∕RI where n is number of parameters, w i is the weighting factor, and r i is the rating of factors.
where p r is the parameter rating, p w is the parameter weight, and V is the summation value of the applied index.
Using AHP, the flood susceptibility map was generated, to assess the relative relevance and get a ranking; the values in each row are compared to the values in each column.After filling up the pairwise comparison matrix shown in the Table 3, the criteria are normalized to get the normalized matrix.Normalized values are calculated by dividing the value in each column by the total of the values in each column.Later criterion weights are derived by averaging all of the row's components.Following that, a consistency check is performed to determine whether or not the computed value is valid.The ratio is calculated by dividing weighted sum by criterion weights and the final flood susceptibility map was generated with a CR value of 0.02 (< 0.1 valid).
Finally, all of the parameters are converted to a raster format, with each raster layer's spatial resolution set to 30 m × 30 m cell size, and it is overlaid using a raster calculator in ArcGIS based on the AHP technique's effective weights.This provides an integrated database with five flood susceptibility classes: very high, high, moderate, low, and very low susceptible, respectively, which may be used to estimate the rate of flooding probability. (5)

Results and discussion
The flood susceptibility map was generated using the effective flood susceptibility factors shown in Table 4 and the AHP criterion weights (%).The findings are addressed further below.

Slope map
The slope percent is a surface that is used to characterize the flood.A low-sloped region is more vulnerable since it is easily flooded by floodwaters.A surface with a steep slope, on the other hand, is less vulnerable to flooding since floodwaters may be swiftly drained down slope.As a result, this element is significant in determining the quantity of surface runoff and infiltration, and hence influences flood sensitivity.The research area is classified into five slope groups based on its slope shown in Table 4.The low slope (0-3°) indicates high vulnerable and the high slope indicates (> 25°) less high vulnerable to flood shown in Fig. 3.

Elevation map
The elevation has a considerable influence on the propagation of floods.This parameter plays a significant function in controlling the following direction movement as well as the depth of the flood.The generated map was divided into five classes shown in Table 4.The low elevation (< 15) is high susceptible to floods and the high elevation (> 66) implies less susceptible to floods shown in Fig. 3.

TWI map
Because of the low slope, a high index value indicates that there is a lot of water that can be collected, and vice versa.
As a result, areas with a higher TWI are more likely to get flooded.In this study, TWI with (> 8.2) is assigned with a high risk, whereas TWI with (< 2.8) is associated with a lower risk shown in Table 4.

SPI map
The steam power index has negative values for locations with topographic potential for deposition and positive values for potentially erosive zones, with the maximum values associated with a steep slope gradient.This terrain form greatly contributes to the aggressiveness of erosion and the likelihood of land degradation.The lowest values (< − 2.5) reflect generally flat terrain shown in Fig. 3, which influences high flood and sediment deposition and accumulation.

Drainage density
The length of rivers per unit of area is expressed as drainage density, which is an important component of flood control techniques.The terrain of the study area has a high drainage density (> 15).On the other side, the plains have a low drainage density (0-2.5)shown in Table 4.

Land use and land cover (LULC) map
LULC is the most important element in determining whether places are at risk of floods.The rate of infiltration is influenced by land usage.Forests and plants, for example, promote water infiltration.Storm runoff is increased in urban areas, which are largely composed of impermeable surfaces and barren land.Water body, trees, grass, flooded vegetation, scrub, built-up area, and barren terrain were identified as LULC classes shown in Fig. 4.

Distance from the river map
The flood severity was maximum in the areas adjacent to the catchment outlet.As a result, the distance from the river is a crucial factor in flood mapping.2000 m, 2000-4000 m, 4000-6000 m, 6000-8000 m, and > 8000 m are the distances from the river ranges, accordingly shown in Fig. 4.

Aspect map
Another contributing element in predicting flood occurrence is aspect, which is closely related to the convergence and direction of water flow toward which the slope of a mountain is facing.

Lithology map
This component is connected to the permeability and storage capacity of the rock, which varies depending on the

Soil map
In the current study, thematic soil types were presented in a GIS layer assessing soils according to their textures and structures.Effectively, each soil type's texture and structure may have a significant influence on its permeability and, as a result, its water storage capacity.Water, pellic vertisols, dystric regosols, eutric nitosols, chromic livosols, thionic livosols, calcaric fluvisols, and lithosols are among the soil types illustrated on the map shown in Fig. 5.

Precipitation map
Heavy precipitation and extreme weather events have the most direct effect of increasing the intensity of other causes The AHP approach was used to construct the final flood susceptibility map, which was divided into five broad groups with flood potentialities ranging from very low to very high, which are shown in Table 5.According to flood susceptibility studies, high-class outlined regions have a high risk of flooding.It has a lower slope and elevation, a higher TWI, a lower SPI, is near to streams, has poorly  drained soil and lithology, and receives extreme rain, all of which are critical conditions for flood mapping using AHP.According to this study, precipitation is the most important factor in floods.Since severe rain reduces the amount of infiltration.As a result, it flows into the river.
The faster water travels into the river, the more likely it will flood.Water that travels over the surface (surface runoff) reaches a river faster than water that travels through the subsurface.So, the precipitation parameter was given the most weight shown in Table 4.
In general, maps generated by integrated models addressed higher prediction accuracy and may be utilized for spatial prediction of flood hazard analysis in the research region.The results shown in Fig. 6 of the model is categorized as very low, low, moderate, high, and very highly susceptible, indicating that the majority of flood-affected regions fall into the moderate, high, and very high categories.Table 5 shows the flood susceptibility map developed from the AHP technique which indicates that 124,593.259ha of the study area is very highly susceptible, 242,944.039ha of area is highly susceptible, 261,035.858ha is moderately susceptible, 184,318.836ha of the area is less susceptible, and 59,807.7977ha of the area is very less susceptible to floods; for an area of 872,699.7 ha, the total area of Krishna District is 872,700 ha which is exactly matching.Finally, the current findings of the study maps may be utilized as a reference for managing flood vulnerability in the Krishna District.Provincial governments can know about high-risk zones in order to adequately control flooding and arrange the progression of necessary flood prevention systems.

Validation discussion
The validation is carried out by locating the past flood events in the prepared flood hazard map as shown in Fig. 6.Total 39 number of flood samples from 1962 to 2022 are collected from various sources such as newspapers, published literature, and disaster manual.The validation indicates good correlation between past flood events and mapped zones.Most of the preceding flood occurrences happened in the high, and very high potential zones, while moderate zone also experienced few instances of flood events.However, from the hazard potential map, the regions marked as very low did not experience any event from the flood history.Out of 39 total flood events 9, 17, and 13 number of flood events have occurred in moderate, high, and very high flood hazard zones, respectively.

Flood management plans
Particularly in the case of floods, proposing different mitigation strategies play a vital role in reducing or eliminating the consequences of hazards.No infrastructure should be permitted in high-floodplain regions, although structures with elevated platforms should be permitted in medium and low-floodplain areas.In the case of a flood, residents in low-floodplain areas should have access to community infrastructure.Retrofitting river banks will be critical in reducing flood damage and increasing flood storage capacity.People are made aware of the need to take essential preparations before a flood occurs.Local governments and communities might employ a disaster management approach based Fig. 6 Flood susceptibility mapping of the study area on geospatial technology to make better decisions before and after the flood, minimizing the loss of life and property.

Conclusions
GIS and remote sensing methods and analysis are useful in a variety of disciplines.This has been utilized for mapping, modeling, and analysis of a vast scope of disaster management applications at various levels and stages.For flood vulnerability mapping in the Krishna District, a GISbased spatial multi-criteria evaluation framework combined with the AHP approach has proven useful for demarcating flood hazard zoning.Elevation, slope, aspect, land use/land cover (LULC), drainage density, topographic wetness index, stream power index, lithology, soil, precipitation, and distance from streams were used as inputs to the AHP model for flood susceptibility mapping.
The use of remote sensing data in accordance with a GIS tool is extremely effective for preliminary flood hazard analysis.However, the availability of adequate datasets and data resolution is critical to the current method's reliability, performance, and applicability.The study employed an SRTM DEM with a spatial resolution of 30 m, which may be termed coarser for the floodplain region.Higher resolution data could be used to fine-tune the study.The accuracy of the current technique might be improved by incorporating more detailed field data such as groundwater depth, local drainage, and flood locations.Furthermore, the lack of hightemporal-scale remote sensing data is a key impediment to flood mapping.
The AHP model's flood vulnerability zone was classified into three groups (high, moderate, and low).The findings clearly show that regions with low elevation and slope, high TWI, low SPI, high drainage density, and high precipitation are primarily risk-prone due to the high potential of flooding in such areas.The AHP technique used expert judgment and expertise to determine the best weights for the components that contribute to flood risk.The ease of handling, flexibility, and low cost of the applicable technique make it possible to employ, particularly in such a location with relatively limited data and information.In the present preliminary assessment of flood hazards, the AHP technique serves its purpose.
Food risk management is a crucial segment of all social and environmental processes aimed at minimizing loss of life, injury, and property damage.The findings of this research will assist decision-makers in carrying out adequate flood management in the future.Finally, the present study's maps may be utilized as a reference for flood prevention and preparedness by planners, developers, and the government.Knowledge of high-risk zones is essential for local governments to adequately control flooding and plan the execution of necessary flood prevention systems.
The inclusion of socio-economic factor, resiliency index, and public opinions into the framework can result in more robust flood hazard potential map for operational purpose.

Fig. 1
Fig. 1 Study area location map

Fig. 2
Fig. 2 ArcGIS conceptual model for flood susceptibility mapping based on the analytical hierarchy process

Table 4
Flood susceptibility mapping parameters and subclasses of different categories based on weights

Table 5
Spatial distribution of flood susceptibility classes