Escalating rainstorm-induced flood risks in the Yellow River Basin, China

The warming climate-induced intensification of hydrological cycle is amplifying extreme precipitation and increasing flood risk at regional and global scales. The evaluation of flood risk, which depends on assessment indicators, weights, as well as data quality, is the first step toward mitigation flood disasters. In this study, we accepted ten risk assessment indicators concerning hazard of disaster-causing factors, sensitivity of hazard-forming environments, and vulnerability of disaster-bearing bodies. We used a combined weighting method based on the analytic hierarchy process and entropy weight (AHP-EW) technique to evaluate rainstorm-induced flood risks across the Yellow River Basin (YRB) from 2000 to 2018. We observed flood hazards are intensifying across the YRB. Specifically, areas with medium flood hazards expanded from the lower to the middle and upper YRB. The sensitivity to floods exhibited a spatial pattern of increasing from southeast to northwest (lower to upper YRB). The increase in vegetation coverage in the middle and upper reaches of the YRB reduces the sensitivity to flood disasters. Flood vulnerability shows an increasing trend, with higher vulnerability mainly observed in the middle and lower YRB. The overall flood risk in the YRB shows an increasing trend, with a 9-fold increase in flood risk from 2000 to 2018. Medium to high flood risk and vulnerability can mainly be identified in the middle and lower YRB, where population and gross domestic product are concentrated. The intensifying rainstorm-induced flood risks over urban areas in these regions should arouse public concern.


Introduction
The warming climate has been driving the intensification of the hydrological cycle leading to an alleged increase in the frequency and severity of extreme precipitation events and torrential rainstorm flood disasters [1,2].Model predictions have shown significant augmentation of flood risk in the Andes Mountains, East Africa, the Indian Peninsula, and Southeast Asia [3].Due to the growing population and booming economy in flood-prone areas, a threefold increase in global exposure to flood disasters by 2050 is anticipated [4].A similar trend has been projected in China that warming climate-driven intensifying precipitation extremes can be expected to push up frequency and intensity of flood disasters [1].Rapid urbanization and the expansion of urbanized areas are rendering China highly vulnerable to climate change and flood hazards [5].Many cities in China, such as Beijing, Wuhan, and Guangzhou, have been experiencing frequent flood hazards and disasters [6].Furthermore, under the influence of future climate change and urbanization, the flood disaster risk in China is expected to further intensify [7,8].
Under the 1.5 • C warming scenario, human casualties caused by floods could increase by 70%-83%, and direct flood losses might increase by 160%-240% with the largest projected losses in Asia [9].China is one of the countries suffering from frequent devastating flood disasters.Flood disasters that occurred in 1991, 1994, 1996, and 1998 inflicted losses of 3%-4% of the gross domestic product (GDP) of China [10].Liu et al developed a disaster risk assessment model based on information content to evaluate the risk of flood and drought disasters in the Yangtze River basin [11].Chen et al analyzed flood disaster risks and losses in South China based on the flood disaster risk model [12].However, relevant scholars have done little on flood disasters in the YRB.The Yellow River is the second largest river in China, home to 12% of the national population, feeding 15% of the irrigation area, and contributing to 9% of China's GDP [13,14].The YRB has been viewed as 'China's Sorrow' due to frequent disastrous flood disasters [15].Severe flood disasters have led to over 1500 breaches in the YRB and at least 26 major channel diversions [16].The Central Chinese Government issued the 'Ecological Protection and High-Quality Development Plan for the Yellow River Basin' on 8 October 2021 [17], attaching considerable importance to the development of the YRB.In this context, it is of utmost significance to evaluate flood risk aiming to enhance mitigation and prevention, the reduction of flood disasters over the YRB [18].
Flood risk management can map flood risk for socioeconomic development planning and can greatly reduce flood disasters [19].Huge body of researches have been done on flood modeling and risk prevention.For instance, Liu et al proposed a spatial framework integrating Bayes and Geographic Information System (GIS) to evaluate flood hazard in the Bowen basin in Australia [20].Similarly, Sheng et al evaluated the relationship between rainfall and flood in the Wei River Basin by combining statistical methods and hydrological model, resulting in a series of risk zoning maps for the impact of mountain torrents at different return periods [21].Meanwhile, recent years have witnessed the development of a body of flood risk evaluation techniques, including statistical analysis of historical disaster data [22], indicator-based approaches [23], GIS analysis [24], and scenario simulation methods [25].The indicatorbased approaches have been widely used in flood risk analyses, such as analytic hierarchy process (AHP) [26], entropy weight (EW) method [27], neural networks [28], along with 3S techniques.Efficient flood monitoring and assessment call for a combination of the indicator-based approach and GIS-based spatial analysis [10].For example, Chakraborty and Mukhopadhyay developed a model combining flood hazard and vulnerability indicators to identify highrisk areas for regional flood disasters using GIS-based methods [29].Similarly, Zhang et al employed a risk assessment model that incorporates hazard, vulnerability, and exposure indicators to delineate flood risk zones in the Yangtze River Basin [10].GIS-based multi-indicator models have been widely recognized as effective methods for flood risk evaluation [30].
However, flood risk evaluation involves several steps, including the selection of evaluation indicators, determination of weights, standardization of data, and precision, which can introduce considerable uncertainties into the results of flood risk evaluation [31].The determination of the disaster risk assessment framework forms the basis for disaster risk assessment.However, there existed different definitions of disaster risk assessment framework.For example, Wilson and Crouch believed that risk is the uncertainty about the extent of loss and damage that may be caused by a disaster [32].Merz and Thieken classified risks as the degree of danger of disasters, with greater danger implying greater risk [33].Subsequently, the definition of the risk assessment framework was further developed.Liu and Chen believed that risk depends on three factors: hazard, vulnerability, and exposure [34].Wu et al showed that disaster-causing factors, disaster-bearing bodies, and disaster-forming environment work together to form risks [35].Furthermore, the selection of flood risk evaluation indicators often involves the integration of information from various fields such as hydrology, meteorology, topography, ecology, society, and environment [23,36,37].The AHP method can exhibit considerable subjectivity, while methods like entropy weighting can be influenced by data objectivity [38].Therefore, relying solely on a single method for weight determination can introduce uncertainty into disaster risk assessment.Recent years have witnessed the adoption of a combination of subjective and objective methods [38,39], as well as improved weight determination techniques [40,41], for conducting flood and other natural disasters risk zoning for regions.Wu et al used the least squares method to synthesize the weights determined by the AHP method and the EW method to evaluate flood disaster risks in the Huai River Basin [38].Lyu et al used the improved AHP (I-AHP) method to analyze the flood risk of the Guangzhou subway system [40].These studies showed that a combination of subjective and objective methods or an improved method can mask the uncertainty introduced by a single method.Therefore, to overcome the uncertainty introduced by the single weighting method into the research results, we used the AHP-EW technique to evaluate rainstorm-induced flood risks.Meanwhile, using precise raster data [10], especially within the same region, often leads to more accurate identification of flood risk zones compared to data at broader scales, such as city or provincial levels [42].Additionally, as a result of warming climate and rapid urbanization processes, disaster vulnerability and risk are also undergoing dynamic changes [7,43].Generally, studies have typically selected natural and socio-economic data during a specific historical period to assess flood disaster risk, and limited studies have been done analyzing variable flood risks across a series of historical periods [12,44].
The Yellow River, regarded as China's mother river and the cradle of Chinese civilization, is wellknown for its frequent floods and droughts [43,45].Meanwhile, The YRB holds significant importance as an essential grain-producing area in China, playing an indispensable role in socioeconomic development and ecological conservation within the country.Recognizing its importance, the ecological conservation and high-quality development of the YRB has been adopted as a major national strategy of China [46].However, the YRB is still afflicted by floods.In this study, we applied ten flood risk assessment indicators from the perspective of hazard of disastercausing factors, sensitivity of hazard-forming environments, and vulnerability of disaster-bearing bodies to assess the risk of rainstorm flood disasters.These indicators encompass natural conditions and human socio-economic activities.Thus, study can provide a theoretical ground for the mitigation of floods in the YRB and help support the socioeconomic development and ecological conservation of the YRB.

Study region and data
The YRB (95 • -119 • E, 32 • -41 • N) is highly sensitive to climate changes and is ecologically fragile [47] (figure 1).The Chinese government attaches profound importance to the sustainability and ecological conservation of the YRB and has initiated ecological protection and high-quality development of the YRB.However, recent years have witnessed an intensification of extreme precipitation under a warming climate [48], consequently escalating flood disaster risks in the YRB.Therefore, flood risk evaluation is of utmost significance for the mitigation of flood disasters across the YRB.
The rainstorm-induced floods are attributed to drivers, such as climate changes, land surface processes and human activities [49].In this study, we considered ten flood risk assessment indicators related to meteorology, hydrology, topography, vegetation, as well as socio-economic activities (economy, population, urbanization, industry, and agriculture) (table 1).The introduction of data that drive the flood risk assessment model can be found in section 1 of supplementary information.To enhance accuracy and ensure data consistency, ArcGIS 10.8 was used to correct all indicator data and uniformly resample them into raster data with a resolution of 30 m × 30 m.

Flood risk assessment model
Flood risk assessment model for the YRB included: (1) development of a flood risk evaluation indicator framework involving hazard, sensitivity, and vulnerability factors [35,50].(2) Determination of the weights for each assessment indicator using AHP-EW based on indicator data from five different years [38].(3) Application of standardization to each of the indicators.( 4) Identification of spatiotemporal pattern of flood hazard, sensitivity, and vulnerability.( 5) Visualization of the dynamic spatial distribution of flood risk in the YRB from 2000 to 2018 (figure 2).
In this study, flood risk is a comprehensive function of hazard of disaster-causing factors, sensitivity of hazard-forming environments, and vulnerability of disaster-bearing bodies.The flood risk index can be calculated using equation ( 4): Sensitivity Index : where H ji(x) , S ji(x) and V ji(x) are the standardized values of each indicator.H (x) , S (x) , V (x) and R (x) denote the hazard index, sensitivity index, vulnerability index, and flood risk index, respectively.w i is the AHP-WE weights assigned to each assessment factor.

Selection and processing of indicators
In this study, we considered the availability of indicator data and selected ten flood risk assessment indicators from three aspects: hazard, sensitivity, and vulnerability [20,35].The hazard factors included the rainstorm days and precipitation amount of the rainy season (July to October).For the sensitivity of the hazard-forming environment, factors such as DEM,  Precipitation is identified as the dominant driver behind flood occurrences [51].Using the GPM L3 daily precipitation data for the period from 2000 to 2018, we analyzed the rainstorm days (days with precipitation amount of ⩾25 mm) and the precipitation amount during the rainy season (July to October) as hazard indicators [31,52,53].Additionally, factors influencing the environment conducive to flooding included hydrological conditions and underlying surface characteristics.Assessing the sensitivity of hazard-forming environments required variables, such as DEM, slope, river networks, and vegetation coverage [51,54,55].The vulnerability of disasterbearing bodies referred to all potential losses caused by flood disasters during a specific time interval in a specific region.The higher the population density and GDP, the higher the population vulnerability to flood disasters [56,57].In this study, the built-up area and electricity consumption were selected as indicators to assess economic vulnerability.During flood events, different land use types experienced varying degrees of damage and therefore had different levels of vulnerability [35,50].We overlaid the land use types with GDP date and extracted GDP related to nine land use types in 2000, 2005, 2010, 2015, and 2018, evaluating the vulnerability levels of different land use types.

Determination of indicator weights-AHP-WE 2.2.3.1. AHP method
We used AHP [40] to calculate the subjective weight of the evaluation index.The judgment matrix in the flood risk assessment model reflects the relative importance of each factor (equation ( 5)).In the judgment matrix, two factors a i and a j were chosen at each time step.When a i is significantly more important than a j , a ij was set to 9, and a ji was set to 1/9.
where A u is the judgment matrix, a ij the relative importance of factor i to factor j, which ranges from 1 to 9, and its reciprocal.Then, the subjective weights (w ′ i ) of the factors can be calculated from equations ( 6) and ( 7): The value of the consistency ratio (CR), which can be calculated by equation (8), was used to evaluate the sensitivity and consistency of the judgment matrix.If CR of <0.1 indicated that the consistency of the judgment matrices was reasonable, and vice versa.
where CI = (λmax−n)/(n−1) and λmax is the largest eigenvalue of the judgment matrix, which can be calculated from equation (9).RI is the average random consistency index.

Entropy weight method
The EW method has been widely used in objective weighting [27].The smaller the entropy value of the indicator, the smaller the confusion of the indicator information and the smaller the uncertainty.This results in a higher amount of information.Consequently, the weight of that indicator becomes larger.We assumed that the number of pixels in the study area was n and the number of indicators used was m.The indicator matrix was represented as: In equation (10), m is the number of indicators and n is the number of data in the study area.In equation (11), H i is the entropy of the ith indicator; n is the number of data.f (i,j) denotes the frequency of factor i, and u (i,j) refers to the standardized index value of the jth evaluation object under the ith index in equation (12).However, when f (i,j) = 0, ln f (i,j) is not allowed mathematically.Therefore, we supposed f (i,j) ln f (i,j) = 0 when f (i,j) = 0. Thus, the objective weight of factor i is calculated using equation (13).

AHP-WE
The AHP-WE method is an improved method based on the AHP-EW method.We used the linear combination method to determine the combined weights for the assessment of rainstorm flood risk (table 2).Combination of the subjective weight w ′ i and the objective weight w ′′ i can help derive the composite weight w i .To undermine the interference from highly-variable data and to ensure the consistency between the disparity of w ′ i and w ′′ i , and α and β, we introduced the concept of a distance function, and the composite weight was computed as: where α and β are the allocation coefficients of weights, and α + β = 1.The distance function d(w ′ i , w ′′ i ) for the subjective and objective weights was presented as: In order to make it consistent the degree of difference between different weights and the degree of difference between distribution coefficients, we constructed the following system of equation ( 16): By solving equation ( 16), α and β for each weight were obtained.Substituting the allocation coefficients into equation ( 14) yielded the combination weights.After calculating the weights for 5 years, the average of the combination weights for these 5 years was computed to obtain the combination weights (table 2).

Hazard assessment of rainstorm flood in the YRB
The indicators for the hazard of flood disasters included precipitation amount and rainstorm days during the rainy season [52].In order to eliminate We observed medium-high flood hazard in the lower YRB, specifically Shandong, Henan, Shanxi, and Shaanxi provinces [58].In 2015, due to the El Niño event, the precipitation amount during June to September was about 24% lower than the normal levels, triggering severe droughts in parts of the middle and lower YRB [59,60].Therefore, the flood hazard was reduced by approximately13% when compared to 2010 (figures 3(a)-(e)).
We calculated the average flood hazard index for each province within the YRB (figure 3  intensifying flood hazard in the upper YRB.However, we detected low flood hazard index for Ningxia and Inner Mongolia in the upper YRB due to the lower precipitation amount and rainstorm days of the rainy season.Medium flood hazard was found in Shanxi and Shaanxi provinces over the middle YRB, and flood hazard in Shanxi and Shannxi in 2018 was 1.5 times that in 2000.Additionally, due to the El Niño event in 2015, the flood disaster hazard levels in Henan and Shandong provinces in the lower reaches of the YRB dropped slightly.Generally, the flood hazard in the YRB followed a decreasing trend from the southeast to the northwest (lower to upper YRB), with regions exhibiting medium to high flood hazard spreading from the lower to middle and upper YRB.

Sensitivity assessment of rainstorm flood in the YRB
In this study, we analyzed the impact of NDVI on the flood hazard of the YRB during the years 2000, 2005, 2010, 2015, and 2018.We utilized the AHP-EW method [38] to calculate the weighted average of flood sensitivity index over the period of 2000-2018.The flood sensitivity index was categorized into five levels: low (0.01-0.05), medium-low (0.05-0.10), medium (0.10-0.20), medium-high (0.20-0.30), and high (0.30-0.71) (figures 4(a)-(e)).We identified increased flood sensitivity index from southeast to northwest (lower to upper YRB) during the period of 2000-2018.High flood sensitivity was found in the following regions: (1) parts of the Inner Mongolia in the upper YRB, such as Ningxia and Gansu provinces, where the river network was dense and the slope was gentle.Besides, additional sedimentation due to low vegetation covered also exacerbated flood [61].(2) Provincial capital cities in the middle and lower YRB are low-lying with relatively dense river networks as well as low vegetation cover in urban areas, and these drivers also intensified flood hazard in the middle and lower YRB [31].
Figure 4(f) shows the average flood sensitivity index for each province within the YRB.We found that the average sensitivity index of flood disasters in the YRB was 0.1926, 0.1841, 0.1768, 0.1767, and 0.1628 for the years 2000, 2005, 2010, 2015, and 2018, respectively (figure 4(f)).Overall, the sensitivity of flood disasters in the YRB exhibited a gradual decreasing trend.Meanwhile, from a provincial perspective, we detected lower sensitivity of flood disasters in the upper and middle YRB, while relative moderate changes in the lower YRB and specifically Shandong and Henan provinces (figure 4(f)).At the interannual scale, we identified shrinking regions with medium to high sensitivity of flood disasters in the upper and middle YRB, attributed to an improved ecological environment in the YRB [62].Therefore, increased vegetation cover greatly reduced flood sensitivity in the upper and middle YRB.The enhanced vegetation cover had led to a reduction in flood sensitivity in 2018, down to 85% of the sensitivity observed in the year 2000.

Vulnerability assessment of rainstorm flood in the YRB
The vulnerability indicators of flood disasters in the YRB included GDP, rural, built-up area, electricity consumption, and land use.Due to the unavailability of the 2018 GDP raster data for the study area, here we adjusted the 2019 GDP raster data based on provincial-level statistical data to obtain the 2018 GDP raster data for the YRB.Among them, the vulnerability classification of land use is in the attachment material (section 2 of supplementary information).The AHP-EW method was utilized to calculate the weighted average of flood vulnerability index for the YRB from 2000 to 2018.The flood vulnerability index was categorized into five levels: low (0.00-0.01), medium-low (0.02-0.05), medium (0.05-0.15), medium-high (0.15-0.25), and high (0.25-0.69) (figures 5(a)-(e)).From 2000 to 2018, areas with medium-high vulnerability to flood in the YRB were mainly concentrated in the lower YRB, such as Shandong and Henan provinces, particularly in the provincial capitals and relatively developed urban areas like Zhengzhou in Henan and Xi'an in Shaanxi.These areas had high population density and significant urban, industrial, and economic development compared to western regions.This directly amplified the direct losses caused by flood disasters [31,63].
The average flood vulnerability index for each province within the basin was calculated and presented in zoning in this study (figure 5(f)).The results indicated that the average vulnerability index of flood in the YRB for the years 2000, 2005, 2010, 2015, and 2018 were 0.0027, 0.0056, 0.0080, 0.0126, and 0.0154, respectively.Looking at the interannual variations across the basin, the vulnerability to flood in the YRB had shown an overall increasing trend, especially in provincial capital cities.The high vulnerability of the exposed entities in these cities would significantly amplify the direct losses and degree of damage caused by flood disasters [31,64].From a provincial perspective, the lower YRB was dominated by a sharp increase in flood vulnerability.In 2018, the vulnerability to flood in Henan Province increased by 6.4 times more than 2000.Similarly, in Shandong Province, it increased 5.5 times more in 2018 than 2000.Conversely, the middle and upper YRB was dominated by a gradual increase in flood vulnerability.

Risk assessment of rainstorm flood disasters in the YRB
Due to the relatively small range of the risk index obtained by multiplying various indices, the risk index was normalized for ongoing analysis.The normalized flood risk index for the period 2000-2018 in the YRB was calculated.In this study, areas with a normalized index less than 0.01 were categorized as risk-free zones.Areas with a normalized risk index greater than 0.01 were divided into five levels: low (0.01-0.05), medium-low (0.05-0.10), medium (0.10-0.15), medium-high (0.15-0.20), and high (0.20-1.00) (figures 6(a)-(e)).The results indicated that from 2000 to 2018, the rainstorm flood risk was mainly concentrated in the provincial capitals and relatively developed urban areas of the middle and lower YRB.The average normalized flood risk indices for the YRB in 2000, 2005, 2010, 2015, and 2018 were 0.0008, 0.0020, 0.0032, 0.0042, and 0.0068, respectively.This indicated an overall increasing trend in flood risk for the YRB.The high vulnerability of disaster-bearing bodies had led to an expanding trend in the area of medium to high risk zones over time.In 2018, the overall flood risk in the YRB increased nine times larger than 2000 (figure 6(f)).
From a provincial perspective, the flood risk index in the upper YRB, including Qinghai Province, Sichuan Province, Gansu Province, and Inner Mongolia, had been relatively low.However, in recent years, there had been a significant increase in the flood risk index in these areas.In the middle YRB, the Shanxi and Shaanxi provinces were characterized by increased flood risk.In the Henan and Shandong provinces in the lower YRB, there was a rapid increase in high flood risk values.In 2018, the flood risk in Henan Province increased 7.5 times larger than in 2000, while in Shandong Province, it increased 9.2 times larger than 2000 (figure 6(f)).Song et al identified the lower and middle YRB as high-risk areas for future floods, which is in good line with the findings of this current study [46].In summary, the increased vulnerability of disaster-bearing bodies as well as increased exposure to floods due to socio-economic development had led to an intensification of urban flash flood and flood risk in the YRB.

Discussion
Based on the results regarding the aggravation of urban rainstorm and flood disasters, we stepped further into the average changes in the normalized flood risk index of provincial capitals in the YRB during 2000, 2005, 2010, 2015, and 2018.We observed that the flood risk in each provincial capital city showed an increasing trend, and the area of regions with medium to high risk showed a spreading trend in the interannual variation (figure 7).Xining and Lanzhou, the provincial capitals in the upper YRB, were dominated by relatively low risk values for flood disasters due to the low hazard of disastercausing factors and the low vulnerability of disasterbearing bodies.However, provincial capitals in the middle and lower YRB showed high and rapidly increasing flood risk values, particularly cities like Xi'an in Shaanxi Province and Zhengzhou in Henan Province [63,65].These cities have well-developed economies, and their high vulnerability exacerbates the disaster risk associated with the events of similar magnitude.For instance, in 2018, the flood risk in Zhengzhou increased nearly nine-fold compared to 2000 (figure 7).Meanwhile, rapid urbanization aroused more exposure of increasing number of people and economic activities to natural disasters, which would further contribute to the vulnerability and exposure of the disaster-bearing bodies [64].Recent flood disasters in the lower YRB, such as the rainstorm flood disasters that occurred in Zhengzhou and surrounding areas on 20 July 2021, corroborate the findings of this study [66].Therefore, under the backdrop of complex climate change, the flood risk in China is expected to further increase.Cities like the Beijing-Tianjin-Hebei region and the Pearl River Delta, were facing frequent flood disasters [67,68].In addition, this article discusses the uncertainty issues in flood disaster risk assessment and the enlightenment that research brings to us (section 3 of supplementary information).

Conclusions
The conclusions drawn from the aforementioned results are as follows: (1) The overall flood hazard in the YRB showed a decreasing trend from southeast to northwest (lower to upper YRB), with medium to high hazard levels concentrated in the middle and lower YRB.When considering the inter-annual variations, the flood hazard in the YRB exhibited a gradual upward trend.Additionally, the impact of climate change was causing the regions with medium to high flood hazard risk to expand from the lower to the middle and upper YRB.
(2) The sensitivity to floods exhibited a spatial pattern of increasing from southeast to northwest (lower to upper YRB).While areas with medium to high sensitivity were initially concentrated in upstream regions and provincial capital cities, there is a shrinking trend in regions with medium to high sensitivity in the middle and upper YRB due to increased vegetation coverage.(3) The vulnerability of impermeable areas (urban areas) was significantly higher than other land use types and showed a rapid increasing trend.On an annual basis, the overall vulnerability to floods in the YRB was on the rise.Medium to high vulnerability regions are primarily found in the middle and lower YRB, particularly in provincial capital cities and developed urban areas.From a provincial perspective, Vulnerability is sharply increasing in downstream areas, while gradually increasing in the middle and upper YRB.(4) The overall flood risk in the YRB was showing an upward trend, with the flood risk in 2018 being nine times higher than that in 2000.Regions with medium to high risk were mainly concentrated in the middle and lower YRB.The areas with medium to high risk were expanding over time due to the high vulnerability of the exposed disaster-bearing bodies.The flood risk index in the upstream areas of the YRB, including Qinghai Province, Sichuan Province, Gansu Province, and Inner Mongolia, was relatively low.However, in recent years, there has been a steep increase in flood risk in these regions.In the middle YRB, Shanxi Province and Shaanxi Province were experiencing a gradual increase in flood risk.In the downstream areas, particularly in Henan and Shandong Provinces, there was a rapid increase in risk.(5) Furthermore, urban flood risk, especially in provincial capital cities, demands attention and effective risk mitigation measures.Flood risks in these cities show significant growth, with an expanding area of medium to high-risk regions.
Cities like Xi'an and Zhengzhou exhibit high vulnerability due to economic development, exacerbating the risk of disaster occurrence and impact.Zhengzhou, in particular, has the highest flood risk among provincial capital cities, with a nine-fold increase in risk from 2000 to 2018.Therefore, effective risk management measures are crucial for mitigating urban flood risk in these areas.

Figure 2 .
Figure 2. Analysis procedure of this study.
(f)).The results indicated that the average flood hazard index for the YRB was 0.2023 in 2000, 0.2569 in 2005, 0.2953 in 2010, 0.2575 in 2015, and 0.3613 in 2018.The flood hazard index in the YRB showed a gradual upward trend.From a provincial perspective, we identified

Table 1 .
Descriptions of the datasets considered in this current study.

Table 2 .
Weights of flood risk assessment indicators.