Abstract
Pakistan’s geographic location makes it an important land hub between Central Asia, Middle East-North Africa, and China. However, the railways, roads, farmland, riverways, and residential quarters in the Piedmont plains of Baluchistan province in northwestern Pakistan are under serious threat of flooding in the summer of 2022. The urgency and severity of climate change’s impact on humanity are underscored by the significant threats posed to human life and property in Piedmont Plains environments through extreme flood events, which has garnered widespread concerns. In flood scenarios, accurately predicting the extent of flooding is crucial for disaster assessment, emergency response, and the efficient allocation of resources. Previous research has primarily predicted flooding likelihood based on topographical factors or integrated annual rainfall data, failing to account for the extent of flooding from short-term rainfall before and after an event. Flood disasters are not caused by a single factor but are influenced by a variety of elements, including terrain and climate. Therefore, current research still lacks a comprehensive consideration of these influencing factors to accurately predict both the range and severity of flood impacts. In this paper, in response to the inability to accurately predict the flood damage in the pre-hill plains region in previous studies, combined with the current Pakistan mega-flood disaster, will couple the impacts of various flood-inducing factors on flooding, construct a prediction model for the degree of inundation of the Pakistani pre-hill plains flood disaster, and combined with the distribution of regional bearers, analyze the risk-resistant capacity of different types of bearers, and draw a comprehensive risk map piece under the flooding disaster. This paper bridges the gap of not integrating various factors in previous studies. Our research results provide strong evidence for flood prediction in Pakistan and similar regions, which is of great significance in reducing the loss of life and property of people around the world.
Keywords: Pakistan, Piedmont plain, Flood, Submerged area, Risk assessment
Subject terms: Environmental sciences, Natural hazards
Introduction
The primary objective behind the establishment of the China-Pakistan Economic Corridor (CPEC) is to bolster bilateral exchanges and collaboration between China and Pakistan in the fields of transportation, energy, maritime affairs, and other relevant sectors. This initiative aims to enhance connectivity and facilitate joint development endeavors between both nations. The prevention of natural disasters has become the most critical area of cooperation between China and Pakistan1,2. The Emergency Events Database from the Disaster Epidemiology Research Center indicated that between 1980 and 2020, there were 4,845 flood disasters with an average yearly cost of more than $21 billion and 5,900 deaths3. The increasing frequency and magnitude of natural hazards across the CPEC in the past half-century (e.g., floods, rockfalls, landslides, and debris flows) necessitate strengthening hazard preparedness and effective monitoring and early warning, which is critical for building corridor disaster resilience4–6.
Floods are the most frequently occurring natural disaster in the CPEC. Over the past decade, approximately 80 to 90% of global natural disasters have consisted of flash floods, coastal floods, or river floods. In China and Pakistan, in particular, the floods have been devastating. The floods devastated everything in their path, and inadequately designed infrastructure was swiftly overwhelmed, leading to the unfortunate consequences of casualties and damage to both human lives and property. On July 6, 2016, the flood disaster affected 757,000 people in 12 districts of Wuhan, China. Additionally, crop damage encompassed 97,404 hectares, 5,848 houses were destroyed. On July 21, 2010, three villages in Pakistan’s North Frontier Province were destroyed by floods triggered by heavy rainfall during the night of the 20th, with the death toll estimated to exceed 100 people. Subsequently, torrential rains continued to fall across the country, and by August 7, 2010, the number of people affected by the floods nationwide in Pakistan had reached 12 million, with more than 1,600 fatalities7. Since July 2022, Pakistan has experienced heavy rainfall, leading to extensive flooding and causing significant impacts across various regions. Many water-damaged roads had to be repaired. Hence, it is crucial to conduct a comprehensive evaluation of the flood vulnerability associated with the CPEC.
The process of evaluating natural disasters for potential risks involves multiple components, such as hazard and vulnerability. Many studies have examined and utilized indices related to these aspects to determine the flood risk index8. The intrinsic and extrinsic variables must be considered by determining the flood disaster hazard9–12. Variables, including elevation, slope, lithology, drainage density, and normalized difference vegetation index, are intrinsic variables determining the hazard of flood disasters. Similarly, extrinsic factors such as the melting of ice and snow and heavy rainfall resulting from global climate change contribute to urban flooding disasters13,14. In flood disasters, emergency response and rescue action, the population and building density, the different land utilization types of spatial arrangement, and the gross domestic product at the municipal level are the essential factors in the vulnerability evaluation of disaster bodies15–17. Numerous scholars employ various approaches and methodologies to assess the flood hazard after intense precipitation. Geographic Information System (GIS) technology and weighted integrated assessment methods are widely used because of their user-friendly interface and low data requirements18–20. The integration of GIS and remote sensing (RS) techniques was utilized to carry out the assessment and mapping of social flood risk in the Lower Mono River basin in West Africa21. A univariate deterministic flood loss model based on depth is established to estimate the potential flood losses by studying the direct losses of flood disasters in different return periods22. A proposed method combines neural networks and numerical models to simulate and detect regions with a high likelihood of experiencing urban flooding, as well as provide quick predictions for water depth23. In 2022, an Analytic Hierarchy Process (AHP) based on GIS technology was employed to evaluate the risk of flood hazards in the province of Bitlis, located in Turkey24. The flood risk of Henan Province, China is analyzed by GIS technology and a weighted comprehensive assessment method, and a disaster prediction model is established by combining the Particle Swarm Optimization-support vector regression (PSO-SVR)_algorithm25.
Given the scientific and technological circumstances, developing nations face significant risks of extensive destruction caused by hazardous floods26. The Piedmont Plain in Pakistan is prone to various natural calamities, including excessive precipitation, floods, and debris flow. The Plain is liable to be threatened by topographic convergence-caused floods, in addition to heavy rainfall-induced floods. Hence, it is crucial to take into account both factors contributing to flood disasters and employ objective criteria for their characterization to conduct an accurate assessment of flood risks in the Piedmont Plain. For the current flood risk assessment system, the choice of evaluation indicators rarely considers the extent of inundation caused by topography and the negative influence of anti-risk capability in the flood threat area. There is a lack of sufficient research on evaluating flood risks in the Piedmont Plain region, which is also the main focus of this study. The primary objective of this study is to propose a comprehensive flood risk assessment methodology for the Piedmont Plain of Pakistan. The research primarily revolves around the development of a comprehensive system for evaluating flood risk for Pakistan Piedmont Plain combining the indicators for characterizing topographic inducible factors, flood-forming environment, coping capacity, and vulnerable features. The evaluation results have necessary guidance for flood disaster prevention and mitigation in the study area and even the territory of Pakistan.
Study area
Pakistan has tropical weather with generally high temperatures and little rainfall. But there were huge flood events in 2010 and 2022. In 2010, floods had an impact on a total of 12 million individuals in Pakistan and over 1,600 people died27. As a result of climate change, Pakistan has witnessed an unusually high amount of rainfall during this monsoon season, surpassing the national average of the past three decades by 2.87 times. Certain regions in Sindh and Balochistan have even experienced precipitation levels exceeding five times the usual amount. The magnitude of the disaster surpassed all expectations. The healthcare infrastructure, agricultural resources, irrigation system, information technology facilities, and transportation network have been severely damaged, leaving a significant number of individuals without shelter. The death toll from monsoon rains in Pakistan since mid-June has risen to 1,265, with 12,577 injured and more than 33 million individuals impacted, according to the National Disaster Management Authority of Pakistan on September 2.
After causing widespread destruction in the northern areas, floodwaters have converged in Baluchistan Province in the south, where the Piedmont Plain area has been particularly affected. Some news estimates that over 1.2 million people have been affected by rains and floods since July 7 in the district. In this study, the northwestern part of Baluchistan Province, which was badly affected, was selected as the research area, mainly including Sibi and Nasirabad (Fig. 1). Sibi is situated in the eastern part of Balochistan and is known for its hot climate, with temperatures often reaching extremely high levels in summer, sometimes exceeding 52 °C. The region is bordered by several mountain ranges including the Zen, Bambore, and Dungan, which influence its climate and topography. Nasirabad: Located centrally in Balochistan, Nasirabad’s capital is Dera Murad Jamali. The region is marked by its arid environment and is significantly influenced by its proximity to the Indus and Kacchi Plains, contributing to its agricultural activities despite its desert-like conditions. Geographically spanning 67°11′0.7″E-69°44′11.0″E, 27°53′30.2″N-29°41′7.8″N. The study area comprises approximately 300,000 individuals in terms of population, mainly in animal husbandry and agriculture, with a gross domestic product (GDP) of about 18.76 billion CNY.
Fig. 1.
The location map of study area. The basemap came from Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) digital elevation model (DEM) and was created using ArcGIS Pro®.
Data and methods
By August 28, 2022, rainfall in the region had more than doubled, according to Multi-source weighted ensemble precipitation (MSWEP) data. By comparing with the observed data of global rainfall stations, the weights of precipitation data from different sources of each grid cell were determined, and a kind of multi-source precipitation fusion data MSWEP was proposed. MSWEP integrates precipitation data from rain gauge measurements, satellite observations, and atmospheric reanalysis. It provides a more comprehensive study and objective assessment of precipitation patterns, while also meeting the requirements of high spatial and temporal resolution and long-term analyses. The data has been extensively verified with reliable accuracy in Latin America, Australia, India, Iran, and China. At the same time, the MSWEP analysis data showed that the Indian summer monsoon index in August exhibited a significant positive correlation with precipitation anomalies across the majority of regions located south of 32°N in Pakistan, and the significant correlation was mainly located in the Baluchistan province (Fig. 2).
Fig. 2.
Spatial distribution of monthly precipitation in the 2022 rainy season and its comparison with the historical rainfall in the same period.
A flood inundation database plays a crucial role in conducting comprehensive assessments of flood disaster susceptibility, hazard, vulnerability, and risk. By analyzing satellite imagery, and international meteorological data, and compiling information from previous publications, an elaborate flood inundation database can be established. The interpretation of objects at risk of flooding was conducted using satellite images from GF-1 and GF-6, which had spatial resolutions of 16 m and 8 m respectively. The normalized difference vegetation index (NDVI) and residential areas were mapped using Landsat-8 satellite imagery, which had a spatial resolution of 30 m. The flood threat region’s slope, elevation, stream density, surface roughness, and incise depth were calculated using a DEM featuring a 30 m spatial resolution. The GDP factors have been extracted from the Pakistan Statistical Yearbook 2021 (Table 1).
Table 1.
Remote sensing images source and other geographical information data.
| ID | Data type | Time | Data sources | Precision |
|---|---|---|---|---|
| 1 | Image source | 2020.07 | GF-1 | 16 m |
| 2020.08 | GF-6 | 8 m | ||
| 2022.09 | Landsat-8 | 30 m | ||
| 2 | DEM | 2013 | Advanced spaceborne thermal emission and reflection radiometer | 30 m |
| 3 | Gross domestic product | 2021 | Pakistan Statistical yearbook | / |
| 4 | Geological map | 1996 | M. A. Khan and M. P. Searle (1996)28; Khan et al. (2019)29 | 1:50000 |
Since Pakistan is not a developed country, it cannot cope with this emergency alone, especially in the piedmont plain. In many different ways, people suffer in floods. Suddenly, they are swept away with their belongings and, in most cases, their livelihoods. They are forced to leave their homes and communities. In this study, coupled with the main evaluation factors of flood disaster, AHP and mean square error weighting method were used to comprehensively analyze the weights of each evaluation factor, and establish a flood inundation range prediction and evaluation system in Pakistan. Combined with the distribution of regional disaster-bearing bodies, the flood damage under different rainstorm conditions was analyzed.
The evaluation index of flood inundation is taken as a random variable, and dimensionless processing is carried out on it by using extreme value standardization. The positive index processing method is used to analyze and study the favorable factors, and the negative index is used otherwise. The calculation formula is presented below:
 |
1 |
 |
2 |
Where 
- Normalized value of evaluation index of flood inundation; i - the number of evaluation factors; j- different interval grades in evaluation factors; 
- the maximum value of an evaluation index; 
- the minimum value of an evaluation index;
Many researchers have used AHP to assess the weights of flood hazard susceptibility factors. Combining geospatial technology and Multi-Criteria Decision-Making method, a flood risk index map of the basin was constructed based on the flood dataset of AHP30. To evaluate the potential for flooding in Kastamonu, Türkiye on August 11, 2021, we used 11 different variables and applied the AHP to construct flood hazard susceptibility maps31. The AHP has been utilized to produce flood risk maps for the identification of vulnerable areas that exhibit higher susceptibility to inundation during floods in Navsari City, Gujarat, India32. The comparison between the AHP, Weights of Evidence (WoE), and Frequency Ratio (FR) was employed to forecast flood susceptible areas within the Lower Kosi River Basin of the Ganga River Basin33. AHP is a multi-objective evaluation decision-making method for constructing, expressing, correlating and quantifying the elements of a problem. Its steps for calculating the weights of each indicator include creating a recursive hierarchical analysis model, constructing a judgment matrix using Saaty’s scaling method, and comparing the influencing factors two by two to determine the weights 
. The AHP approach involves the computation of the maximum eigenvalue, choice of the discriminant matrix, and determination of the normalized weight for evaluation indices. The consistency of the matrix was tested using formulas (3) to (6), and normalization was applied to the eigenvector.
 |
3 |
 |
4 |
 |
5 |
 |
6 |
Where 
- evaluation index weight of flood inundation;
- the significance comparison between factor 
and factor
; CI - the coincidence indicator; n - the order of the matrix; 
- the maximum eigenvalue of the judging matrix; RI - the average random consistency index; Based on the matrix of discrimination in construction (Table 2), the values of 
Moreover, CR is 10 and 0, respectively. Generally, If CR is less than 0.1, the test is satisfactory; otherwise, it does not have satisfactory consistency. The practicality of constructing the discriminant matrix is evident from the calculation results.
Table 2.
Random consistency index RI values.
| n | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| RI | 0 | 0 | 0.58 | 0.90 | 1.12 | 1.24 | 1.32 | 1.41 | 1.45 | 1.49 | 1.52 | 1.56 | 1.58 |
Mean-square error decision method has the characteristics of simple calculation and clear meaning. In this method, the dispersion level is indicated by the mean square error of random variables for each index. A model utilizing GIS technology was created to assess and appraise the vulnerability to flooding in the Accra Metropolitan Area. The computation of the influences of all parameters in the flood vulnerability assessment was performed using the Mean Square Error decision method34. In the Bahah region of Saudi Arabia, logistic regression (LR), random forest (RF), and support vector machine (SVM) were used to generate multi-hazard maps, identifying 17 influencing factors and employing the mean square error decision method35. In Nghe An province, Vietnam, flood susceptibility maps were constructed using machine learning techniques and remote sensing interpretation technology, integrating soil moisture content data collected from the Cyclone Global Navigation Satellite System (CyGNSS)36. By utilizing hydrological models and mean square error decision validation, early warning systems can be enhanced to accurately predict floods in the Himalayan region, which is of significant importance to the local population37. Based on the evaluation index system of flood inundation degree, this paper obtains the mean value of random variables of each index through dimensionless processing of each index and obtains the weight of each index by homogenizing the mean square error of random variables of each index. The calculation process can be summarized as follows:
 |
7 |
 |
8 |
 |
9 |
Where 
- the mean of all indicators;
- the indicators; 
- the standardized value of the non-dimensional i indicator corresponding to the j evaluation index; 
- the mean square error; 
- the weight value of the j evaluation index.
In conclusion, both methods described above have their advantages and disadvantages. The AHP relies on the expertise and judgment of experts and operators to determine weights, which can vary among individuals and undermine the authority of standards. Moreover, as the number of factors increases, users may find the workload tedious, leading to subjective errors in calculations. On the other hand, the Mean-square error decision method provides the quantitative weights based solely on existing data sets, eliminating bias from the decision-makers preferences. However, this approach may lack persuasiveness when input data quality is low. Additionally, it overlooks real system operation status and human experience, potentially resulting in discrepancies between actual outcomes and calculated results. Both methods will cause excessive errors in subjective and objective aspects38. To solve the shortcomings of these two methods, the combined weighting approach has been proposed and developed in recent years38. By comprehensively considering the objective data and human experience, the models are more scientific and practicable. Therefore, to avoid subjective and objective factors, the concept of distance function is used to represent the different degrees based on the two methods. Let the distance function of the weights of the above two is
, the formula is as follows:
 |
10 |
 |
11 |
 |
12 |
Where 
-combined weight value; α, β - the distribution coefficients of weights.
The weight value of each factor in the process of addressing flood inundation risk is determined using the combined weight method. Flood inundation risk will be composed of two evaluation factors, namely, inundation degree and risk resistance ability. The inundation depth of the flood water flow represents the degree of flood inundation. When facing the threat of a flood disaster, the difference in factors such as different land use types, diverse infrastructure functions, and uneven economic development levels determine the damage degree of the region after a flood disaster to a certain extent. These differences determine the ability to weaken risks and also reflect local resilience. The flood risk calculation formula in Pakistan is as follows:
 |
13 |
 |
14 |
 |
15 |
Where 
-comprehensive assessment of flood inundation; 
-combined weight value of flood flow; 
- the standardized value of the convergence index of flood flow; V- the ability to resist the risk of the flood; 
- the weight value of the index of flood resistance risk ability after standardization; 
-the weight of flood resistance risk ability index data; 
-The main weight of flood inundation; 
-The main weight of flood resistance risk ability; R- risk value of flood disaster. In the study, the main flow-producing pooling indicators of coupled floods will be analyzed using hierarchical analysis and the mean square value assignment method to comprehensively analyze the weights of the evaluation factors and establish an evaluation system for predicting the extent of flood inundation in Pakistan. Combined with the regional distribution of disaster-bearing bodies, the flood risk formed under heavy rainfall conditions is analyzed. The main research technical route is shown in Fig. 3.
Fig. 3.
The quantitative risk assessment technical route of flood disaster in the Piedmont Plain of Pakistan.
Results and discussion
Evaluation factor selection of flood inundation
Two factors affect the flood formation of the drainage basin, namely, the underlying surface parameter and the climate change parameter. Underlying surface factors are divided into topographic parameters, including elevation, slope, NDVI, and surface cutting depth, and geological parameters, including lithology, and surface roughness. Climate indicators include rainfall amount and heavy rain days. The aggregate index of flood flow adopted in this paper is shown in Fig. 4.
Fig. 4.
The aggregation index of flood flow adopted in the Piedmont Plain of Pakistan (H1: elevation; H2: slope; H3: cutting depth; H4: lithology; H5: normalized difference vegetation index; H6: drainage density; H7: surface roughness; H8: accumulated rainfall; H8: Heavy rain days). The basemap came from ASTER DEM and its hillshade and created using ArcGIS Pro®.
According to the study of the underlying surface parameter methods, this paper selects elevation, slope, cutting, NDVI, drainage, roughness, lithology, rainfall, and other variables as the evaluation criteria for assessing flood hazards. Additionally, the pertinence among various indexes guarantees the independence and objectivity of evaluation criteria. The graded-index matrix was later imported into SPSS for conducting Pearson analysis and verification The analysis outcomes indicate that all the indices fulfilled the criteria for independence (as shown in Fig. 5). The results show that the correlation of the selected indicators |R| ≤ 0.32, which means that the selected indicators are satisfied with the evaluation of the degree of flood inundation.
Fig. 5.
The correlation of the aggregation index of flood flow.
In addition, the distribution density of rivers and lakes also plays a crucial part in the generation and discharge of floods. The failure of recharge and drainage of rivers, lakes, and other water systems and local blockage and outbursts will directly affect the scale of the flood. The most immediate cause of instability in these water systems is rainfall. Rainfall and rainfall days are indicators to evaluate rainfall factors, and heavy rainfall is an important incentive for floods. Flood in certain Geological settings and Geomorphic environments requires a certain rainfall, rainfall intensity, or duration to promote the Piedmont Plain damage. The plot of daily rainfall statistics in Pakistan’s Piedmont plains during the 2022 rainy season is shown in Fig. 6. The figure shows that Pakistan experienced unusually heavy rainfall in all the months of July and August, and that the amount of rainfall and the number of days of rainfall are the indicators that affect the flood production and flow, where heavy rainfall is an important trigger for the occurrence of floods.
Fig. 6.
The plot of daily rainfall statistics in Pakistan’s Piedmont plains during the 2022 rainy season.
Analysis of flood inundation prediction results
The index weights of evaluation factors under different methods were calculated according to formulas 1 to 12. The weight values of the prediction indicators of flood inundation are shown in Table 3. Elevation and slope play a primary role in flood inundation, while NDVI plays a relatively minor role. The flood inundation classification map of Pakistan’s Piedmont plains was obtained by assigning weights to various factors through geographic information system software. The elevation and slope factors were negatively correlated with flood inundation, so they were inverted in the calculation.
Table 3.
Statistical tables of the weight values of the prediction indicators of flood inundation.
| Evaluation factor | Main weigh | Secondary evaluation parameter | Secondary evaluation weight | ||||
|---|---|---|---|---|---|---|---|
| Analytic hierarchy process | Mean square error | The combined weight | Analytic hierarchy process | Mean square error | The combined weight | ||
| Flood inundation | 0.685 | 0.502 | 0.586 | Elevation | 0.184 | 0.191 | 0.189 |
| Slope | 0.157 | 0.175 | 0.165 | ||||
| Surface cutting | 0.116 | 0.115 | 0.116 | ||||
| Lithology | 0.112 | 0.088 | 0.103 | ||||
| NDVI | 0.014 | 0.120 | 0.059 | ||||
| Drainage density | 0.116 | 0.086 | 0.094 | ||||
| Roughness | 0.122 | 0.042 | 0.092 | ||||
| Rainfall | 0.109 | 0.104 | 0.105 | ||||
| Rain days | 0.071 | 0.080 | 0.077 | ||||
The accuracy of the submergence degree is validated using the receiver operating characteristic curve (ROC curve)1. The accuracy of the flood inundation prediction model can be assessed by evaluating the area under the curve (AUC). The higher the AUC value, the more accurate the model’s calculated effect. The results show that the AUC values were 0.8761, proving the combined weight method is more reasonable (Fig. 7A). At the same time, this paper obtains the flood inundation range of this region through the Pakistan Data Service Sharing system (Fig. 7B). According to the calculation formula 16, combined with confusion matrix parameters (Table 4), the accuracy of predicting the inundation range is 0.80. The result also shows that the model has good applicability.
Fig. 7.
The map of prediction and verification inundation distribution in Piedmont plains, Pakistan. The basemap came from ASTER DEM and its hillshade and was created using ArcGIS Pro®.
Table 4.
Confusion matrix of accuracy test of flood inundation range.
| Inundated area / km² | TP | FP |
|---|---|---|
| FN | 3753.02 | 5676.99 |
| TN | 446.88 | 19866.48 |
 |
16 |
Where Accuracy - the proportion of area predicted correctly in the whole research area; TP - the region where the projected outcomes perfectly align with the real flood zone; TN - the region where the anticipated outcomes align flawlessly with the factual non-flooded territory; FP - the area that was predicted to be inundated but was non-inundated; FN - the area that was predicted to be non-inundated but was inundated.
According to the natural break point method, the submergence degree of the study area is separated into four levels, as shown in Fig. 7C. Among them, the actual inundated area with a high degree is 2,985.081 km², accounting for 71.08% of the total inundated area. To further analyze and verify the scientificity and feasibility of the prediction model of flood inundation scope in this paper, the actual inundation scope is divided into three categories according to the regional inundation degree, in which class I represent the overwhelming majority of ground objects in the region are submerged, Class II represents part of ground objects in the region are submerged, and class III represents scattered ground objects in the region are submerged (Fig. 7A). The predicted places with a high degree of inundation are mainly concentrated in the Class I area, which is low-lying and widely distributed with farmland, which is not conducive to flood drainage. The places with moderate inundation degree are mainly distributed in the Class II area, which is located in the transition zone from plain to mountain, with higher terrain and some channel distribution, which is convenient for drainage. The is as with low inundation degrees is mainly distributed in Class III areas, which are concentrated in mountainous areas with relatively steep terrain and strong flood discharge capacity. For the non-region is mainly concentrated in the mountainous areas on both sides, which feature relatively steep terrain and strong flood discharge capacity. The direction of water convergence is also primarily toward the central low-lying areas, which exacerbates the flooding convergence in the central region. The test area is consistent with the actual inundation situation, so the flood inundation range prediction model can be applied in the Piedmont Plain of Pakistan.
Analysis of flood risk resistance.
The damages resulting from the flood calamity in the Piedmont Plain of Pakistan are mainly reflected in the aspects of social vulnerability, economic vulnerability, population vulnerability, and material vulnerability. Therefore, GDP, population density, land use, building density, and POI density are selected as indicators of flood risk resistance in this paper (Fig. 8). The GDP refers to the overall monetary worth or market value of all goods and services manufactured within the geographical boundaries of a city during a specific timeframe. Serving as an extensive indicator of domestic production, it provides a comprehensive evaluation of the economic well-being of a particular urban area. The population density refers to the number of individuals residing within a given land area, serving as a crucial metric for assessing the spatial distribution of inhabitants in a particular country or region. Land use type pertains to a land resource unit characterized by the same mode of land utilization. It is categorized based on regional variations in land usage and serves as a fundamental regional entity that mirrors the characteristics of land use, its natural attributes, and distribution patterns. It is a land use classification that emerges during the transformation and utilization of land for production and construction, exhibiting distinct directions and features. Building density is the percentage (%) of the total footprint area occupied by a building within a specific range. It represents the extent to which buildings cover an area and specifically indicates the ratio of the combined base areas of all buildings on project land to the planned construction land (%). This measurement reflects both the amount of open space and building density within a designated land area. Point of Interest (POI) pertains to the geographical coordinates representing significant landmarks and entities that have a direct impact on individuals’ daily lives, encompassing educational institutions, medical facilities, commercial centers, and recreational areas. It serves as an indicator of the diverse urban activities to some extent. The combined weight model was used to calculate the weight value of the risk resistance index, as shown in Table 5. Then, combined with the risk resistance calculation model, the flood’s distribution map resistance ability of the Piedmont Plain in Pakistan was drawn (Fig. 9). The results show that in the southern region with intensive human activities, the risk of flood resistance is higher because of the developed economy and comprehensive infrastructure construction. The reason for this may be that in densely populated areas, there are usually more resources and services available for emergencies, including rescuers, emergency ambulance facilities and rapid response teams. These factors can speed up disaster response and mitigate the effects of flooding. Meanwhile well-developed infrastructure, such as good drainage systems, reinforced river embankments and floodwalls, can effectively manage and channel floodwaters, reducing their damage to urban areas. In addition, a good transportation network can ensure rapid evacuation of people and materials when floods strike.
Fig. 8.
The map of risk resistance indicators in Piedmont plains, Pakistan (V1: GDP; V2: Population; V3: POI; V4: Building; V5: Land use). The basemap came from ASTER DEM and its hillshade and was created using ArcGIS Pro®.
Table 5.
Statistical tables of the weight values of the risk resistance indicators.
| Evaluation factor | Main weigh | Secondary evaluation parameter | Secondary evaluation weight | ||||
|---|---|---|---|---|---|---|---|
| Analytic hierarchy process | Mean square error | The combined weight | Analytic hierarchy process | Mean square error | The combined weight | ||
| Risk resistance | 0.315 | 0.498 | 0.414 | Building | 0.127 | 0.271 | 0.235 |
| POI | 0.266 | 0.058 | 0.093 | ||||
| Landuse | 0.137 | 0.221 | 0.208 | ||||
| GDP | 0.091 | 0.187 | 0.151 | ||||
| Population | 0.379 | 0.263 | 0.313 | ||||
Fig. 9.
The Risk resistance classification map of the study area. The basemap came from ASTER DEM and its hillshade and was created using ArcGIS Pro®.
Flood risk assessment in Piedmont plain of Pakistan
According to the results of the inundated degree distribution of flood disaster and the classification of risk resistance ability, this paper uses the flood risk model to classify flood disaster risk levels in the Piedmont Plain of Pakistan, as shown in Fig. 10. The risk classification results indicated that around 20.08% of the entire area of study was categorized as areas with an extremely high or high level of risk, primarily situated in the southwestern region of Nasirabad and the southeastern region of Jhal Magsi. Due to the flat terrain, abundant water resources, and frequent human activities, the region had a relatively developed economy, high population density, and dense construction distribution. If these areas are hit by heavy rainfall, they are at risk of being completely flooded. Figure 11 shows the distribution of disaster-bearing bodies under the risk classification, where most of the forest and grassland are low-risk areas, and agricultural land is distributed in each classification, with a relatively large number of high risks. The rest of the bearers are relatively small in size, but are distributed in each risk class.
Fig. 10.
The risk classification map of the flood disaster in Piedmont plains, Pakistan. The basemap came from ASTER DEM and its hillshade and was created using ArcGIS Pro®.
Fig. 11.
The distribution map of disaster bearing body under risk classification.
Discussion
The paper suggests that the selection of a rainstorm-induced flood disaster risk assessment model facilitates strategic planning and mitigation measures to address potential disasters. in an actual inundated area of 2,985.081 km², which accounts for 71.08% of the total inundated area. The flood risk resistance shows that the southern region with intensive human activities is higher. Compared with the predicted inundation range of the foremountain flood and the actual inundation range, the accuracy of predicting the inundation range is 0.80. The findings from the verification results of the receiver operating characteristic curve indicate that the AUC values of the combined weight method were 0.8761. It meets the accuracy requirements of the flood inundation prediction, and then the submergence degree of the study area is separated into four levels based on the natural break point method. The actual inundated area with a high degree is accounting for 71.08% of the total inundated area. We selected five factors related to the ability to resist flooding to quantitatively judge the ability to resist flooding within different regions, and the results showed that the better resistance to flooding in the study area is the region with many people, well-developed infrastructures, and more reservoirs and ponds water systems, and through the calculation of the interactions with the degree of inundation, we showed that flooding caused damage to farmland, sparse people, and incomplete infrastructures in the region, and the economic losses were Maximum. For high-risk areas, local people’s awareness of flood control should be improved, and flood discharge projects should be built to increase flood discharge capacity. For extremely high risk areas, the local government should regulate the upstream river, and build flood protection projects such as levees and drainage channels.
The current evaluation of flood risk in the Piedmont Plain of Pakistan indicates that 25% of the regions remain within a high-risk category. In the affected areas of high-risk level, high-frequency flood events may occur, which will seriously affect the future development planning of the city. Early warning systems, long-term observation mechanisms, effective disaster mitigation measures, and convenient emergency evacuation plans should be set up in time. The results can provide a basis for the establishment of a more accurate risk assessment and early warning system; by identifying the main influencing factors of floods, the relevant departments can more accurately predict the potential disaster risks and take appropriate preventive and mitigation measures; and it is a guide for disaster prevention and mitigation in the Northwest Pakistan study area and the whole territory.
The paper suggests that the selection of a rainstorm-induced flood disaster risk assessment model facilitates strategic planning and mitigation measures to address potential disasters. However, there are still some difficult problems in the course of this study. For example, it is challenging to acquire high-precision data, such as rainfall data, building information, population density, etc. And the applicability of the research model in this paper is limited by geography and applies to non-urban areas such as mountainous and pre-mountain plains areas. In addition, there is a need for further amelioration in the course of regionalizing flood risk to improve the precision of evaluation outcomes. In the follow-up research process, we will actively try to obtain dynamic rainfall data in the study area and strive to carry out a refined dynamic flood risk assessment.
Acknowledgements
The research was supported by the National Natural Science Foundation of China (42477173, 42442060), Sichuan Province Special Project for Central Guidance on Local Technological Development (Basic Research in Free Exploration) (2024ZYD0121), Everest Scientific Research Program 2.0: Research on mechanism and control of glacial lake outburst chain catastrophe in Qinghai-Tibet Plateau based on man-earth coordination perspective, and China Scholarship Council (202309230008).
Author contributions
M.C., K.Z., and B.Y.: Conceptualization, Methodology, Software, Visualization, Investigation, Writing- Original draft preparation. K.Z., B.Y.: Data curation, Validation. M.C. and X.D.: Funding support and Writing—Review & Editing. M.C. and F.S.: Project administration, Supervision, Resources. All authors reviewed the manuscript.
Data availability
The Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) DEM can be downloaded at https://search.earthdata.nasa.gov/search. The GF-1&6 satellite imagery is not publicly available at this time, please contact the corresponding author. The Landsat-8 satellite imagery can be downloaded at https://earthexplorer.usgs.gov/. The Gross Domestic Product (GDP) data of Pakistan can be obtained from https://www.macrotrends.net/global-metrics/countries/PAK/pakistan/gdp-gross-domestic-product#google_vignette. The Geological Map are not available online.
Declarations
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Chang, M., Cui, P., Dou, X. Y. & Su, F. H. Quantitative risk assessment of landslides over the China–Pakistan economic corridor. Int. J. Disaster Risk Reduct.63, 102441. 10.1016/j.ijdrr.2021.102441 (2021). [Google Scholar]
- 2.Kanwal, S., Atif, S. & Shafiq, M. GIS based landslide susceptibility mapping of northern areas of Pakistan, a case study of Shigar and Shyok basins. Geomat. Nat. Hazards Risk8, 348–366. 10.1080/19475705.2016.1220023 (2017). [Google Scholar]
- 3.Duan, Y. et al. Assessment and spatiotemporal analysis of global flood vulnerability in 2005–2020. Int. J. Disaster Risk Reduct.80, 103201. 10.1016/j.ijdrr.2022.103201 (2022). [Google Scholar]
- 4.Du, G. L. et al. Landslide susceptibility mapping in the region of eastern himalayan syntaxis, Tibetan Plateau, China: A comparison between analytical hierarchy process information value and logistic regression-information value methods. B Eng. Geol. Environ.78, 4201–4215. 10.1007/s10064-018-1393-4 (2019). [Google Scholar]
- 5.Sun, D. L., Xu, J. H., Wen, H. J. & Wang, D. Z. Assessment of landslide susceptibility mapping based on bayesian hyperparameter optimization: A comparison between logistic regression and random forest. Eng. Geol.281, 105972. 10.1016/j.enggeo.2020.105972 (2021). [Google Scholar]
- 6.Yang, S. Y., Jan, C. D., Yen, H. & Wang, J. S. Characterization of landslide distribution and sediment yield in the TsengWen River Watershed, Taiwan. Catena 174, 184–198 (2019). 10.1016/j.catena.2018.11.011 [Google Scholar]
- 7.Webster, P. J., Toma, V. E. & Kim, H. M. Were the 2010 Pakistan floods predictable? Geophys. Res. Lett.38, L04806. 10.1029/2010gl046346 (2011). [Google Scholar]
- 8.Wang, L., Chang, M., Le, J., Xiang, L. L. & Ni, Z. Two multi-temporal datasets to track debris flow after the 2008 Wenchuan earthquake. Sci. Data9, 525. 10.1038/s41597-022-01658-y (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Cao, J. et al. Multi-geohazards susceptibility mapping based on machine learning-a case study in Jiuzhaigou, China. Nat. Hazards102, 851–871. 10.1007/s11069-020-03927-8 (2020). [Google Scholar]
- 10.Haynes, H., Haynes, R. & Pender, G. Integrating socio-economic analysis into decision-support methodology for flood risk management at the development scale (Scotland). Water Environ. J.22, 117–124. 10.1111/j.1747-6593.2007.00086.x (2008). [Google Scholar]
- 11.Tang, C. et al. Catastrophic debris flows on 13 August 2010 in the Qingping area, southwestern China: The combined effects of a strong earthquake and subsequent rainstorms. Geomorphology139, 559–576. 10.1016/j.geomorph.2011.12.021 (2012). [Google Scholar]
- 12.Zhou, Q., Mikkelsen, P. S., Halsnaes, K. & Arnbjerg-Nielsen, K. Framework for economic pluvial flood risk assessment considering climate change effects and adaptation benefits. J. Hydrol.414, 539–549. 10.1016/j.jhydrol.2011.11.031 (2012). [Google Scholar]
- 13.Xiao, Y. F., Yi, S. Z. & Tang, Z. Q. A spatially explicit multi-criteria analysis method on solving spatial heterogeneity problems for flood hazard assessment. Water Resour. Manag.3210.1007/s11269-018-1993-6 (2018).
- 14.Yang, J., Yang, Y. C. E., Chang, J. X., Zhang, J. R. & Yao, J. Impact of dam development and climate change on hydroecological conditions and natural hazard risk in the Mekong River Basin. J. Hydrol.579, 124177. 10.1016/j.jhydrol.2019.124177 (2019). [Google Scholar]
- 15.de Brito, M. M., Evers, M. & Höllermann, B. Prioritization of flood vulnerability, coping capacity and exposure indicators through the Delphi technique: A case study in Taquari-Antas basin, Brazil. Int. J. Disast Risk re. 24, 119–128. 10.1016/j.ijdrr.2017.05.027 (2017). [Google Scholar]
- 16.Luo, H. Y., Zhang, L., Zhang, L., He, J. & Yin, K. Vulnerability of buildings to landslides: The state of the art and future needs. Earth Sci. Rev.238, 104329. 10.1016/j.earscirev.2023.104329 (2023). [Google Scholar]
- 17.Silva, M. & Pereira, S. Assessment of physical vulnerability and potential losses of buildings due to shallow slides. Nat. Hazards. 72, 1029–1050. 10.1007/s11069-014-1052-4 (2014). [Google Scholar]
- 18.Pham, B. T. et al. Flood risk assessment using hybrid artificial intelligence models integrated with multi-criteria decision analysis in Quang Nam Province, Vietnam. J. Hydrol.592, 125815. 10.1016/j.jhydrol.2020.125815 (2021). [Google Scholar]
- 19.Toosi, A. S., Calbimonte, G. H., Nouri, H. & Alaghmand, S. River basin-scale flood hazard assessment using a modified multi-criteria decision analysis approach: A case study. J. Hydrol.574, 660–671. 10.1016/j.jhydrol.2019.04.072 (2019). [Google Scholar]
- 20.Wu, Z. N., Shen, Y. X., Wang, H. L. & Wu, M. M. Urban flood disaster risk evaluation based on ontology and bayesian network. J. Hydrol.583, 124596. 10.1016/j.jhydrol.2020.124596 (2020). [Google Scholar]
- 21.Ntajal, J., Lamptey, B. L., Mahamadou, I. B. & Nyarko, B. K. Flood disaster risk mapping in the lower Mono River Basin in Togo, West Africa. Int. J. Disaster Risk Reduct.23, 93–103. 10.1016/j.ijdrr.2017.03.015 (2017). [Google Scholar]
- 22.Pinos, J., Orellana, D. & Timbe, L. Assessment of microscale economic flood losses in urban and agricultural areas: Case study of the Santa Barbara River, Ecuador. Nat. Hazards103, 2323–2337. 10.1007/s11069-020-04084-8 (2020). [Google Scholar]
- 23.Yan, X. Y., Xu, K., Feng, W. Q. & Chen, J. A. Model of urban flood inundation in a high-risk area coupling machine learning and numerical simulation approaches. Int. J. Disaster Risk Sci.12, 903–918. 10.1007/s13753-021-00384-0 (2021). [Google Scholar]
- 24.Aydin, M. C. & Birincioglu, E. S. Flood risk analysis using gis-based analytical hierarchy process: A case study of Bitlis Province. Appl. Water Sci.12, 122. 10.1007/s13201-022-01655-x (2022). [Google Scholar]
- 25.Han, D. et al. Mega Flood Inundation Analysis and the selection of Optimal shelters. Water-Sui14, 1031. 10.3390/w14071031 (2022). [Google Scholar]
- 26.Lacasse, S. & Nadim, F. Landslide risk assessment and mitigation strategy. Landslides Disaster Risk Reduct. 31–61. 10.1007/978-3-540-69970-5_3 (2009).
- 27.Mahmood, S., Sajjad, A. & Rahman, A. U. Cause and damage analysis of 2010 flood disaster in district Muzaffar Garh, Pakistan. Nat. Hazards107, 1681–1692. 10.1007/s11069-021-04652-6 (2021). [Google Scholar]
- 28.Khan, M. A. & Searle, M. P. Geological map of North Pakistan and Adjacent Areas of Northern Ladakh and Western Tibet: (Western Himalaya, Salt Ranges, Kohistan, Karakoram, Hindu Kush) (Shell International Exploration and Production, 1997).
- 29.Khan, H. et al. Landslide susceptibility assessment using frequency ratio, a case study of northern Pakistan. Egypt. J. Remote Sens.22, 11–24. 10.1016/j.ejrs.2018.03.004 (2019). [Google Scholar]
- 30.Ramkar, P. & Yadav, S. M. Flood risk index in data-scarce river basins using the AHP and GIS approach. Nat. Hazards109, 1119–1140. 10.1007/s11069-021-04871-x (2021). [Google Scholar]
- 31.Yilmaz, O. S. Flood hazard susceptibility areas mapping using analytical hierarchical process (AHP), frequency ratio (FR) and AHP-FR ensemble based on geographic information systems (GIS): A case study for Kastamonu, Turkiye. Acta Geophys.70, 2747–2769. 10.1007/s11600-022-00882-9 (2022). [Google Scholar]
- 32.Pathan, A. I., Agnihotri, P. G., Said, S. & Patel, D. AHP and TOPSIS based flood risk assessment—A case study of the Navsari City, Gujarat, India. Environ. Monit. Assess.194, 509. 10.1007/s10661-022-10111-x (2022). [DOI] [PubMed] [Google Scholar]
- 33.Arora, A. Flood susceptibility prediction using multi criteria decision analysis and bivariate statistical models: A case study of Lower Kosi River Basin, Ganga River Basin, India. Stoch. Environ. Res. Risk Assess.37, 1855–1875. 10.1007/s00477-022-02370-4 (2023). [Google Scholar]
- 34.Nkonu, R. S., Antwi, M., Amo-Boateng, M. & Dekongmen, B. W. GIS-based multi-criteria analytical hierarchy process modelling for urban flood vulnerability analysis, Accra Metropolis. Nat. Hazards117, 1541–1568. 10.1007/s11069-023-05915-0 (2023). [Google Scholar]
- 35.Youssef, A. M., Mahdi, A. M. & Pourghasemi, H. R. Landslides and flood multi-hazard assessment using machine learning techniques. B Eng. Geol. Environ.81, 370. 10.1007/s10064-022-02874-x (2022). [Google Scholar]
- 36.Nguyen, H. D. et al. Flood susceptibility mapping using advanced hybrid machine learning and CyGNSS: A case study of Nghe An province, Vietnam. Acta Geophys.70, 2785–2803. 10.1007/s11600-022-00940-2 (2022). [Google Scholar]
- 37.Tsering, K. et al. Verification of two hydrological models for real-time flood forecasting in the Hindu Kush Himalaya (HKH) region. Nat. Hazards110, 1821–1845. 10.1007/s11069-021-05014-y (2022). [Google Scholar]
- 38.Li, Q., Lu, C. J. & Zhao, H. Risk assessment of floor water inrush based on TOPSIS combined weighting model: A case study in a coal mine, China. Earth Sci. Inf.16, 565–578. 10.1007/s12145-022-00898-1 (2023). [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) DEM can be downloaded at https://search.earthdata.nasa.gov/search. The GF-1&6 satellite imagery is not publicly available at this time, please contact the corresponding author. The Landsat-8 satellite imagery can be downloaded at https://earthexplorer.usgs.gov/. The Gross Domestic Product (GDP) data of Pakistan can be obtained from https://www.macrotrends.net/global-metrics/countries/PAK/pakistan/gdp-gross-domestic-product#google_vignette. The Geological Map are not available online.