Risk Assessment of Yellow Muddy Water in High-Construction-Intensity Cities Based on the GIS Analytic Hierarchy Process Method: A Case Study of Guangzhou City

Abstract

During urbanisation, extensive production and construction activities encroach on ecological spaces, leading to changes in environmental structures and soil erosion. The issue of yellow muddy water caused by rainfall in cities with high construction intensity has garnered significant attention. Taking Guangzhou City as the research area, this study is the first to propose a risk assessment model for yellow muddy water in cities with high construction intensity, and the influence of construction sites on yellow muddy water was fully considered. Rainfall and construction sites were used as indicators to assess the hazards of yellow muddy water. Elevation, slope, normalised difference vegetation index (NDVI), soil erosion modulus, stream power index (SPI), surface permeability, and roads represent the exposure evaluation indicators. Population number and GDP (Gross Domestic Product) were used as vulnerability evaluation indicators. Based on the analytic hierarchy process (AHP) method, the weights of each evaluation indicator were determined, and a risk assessment system for yellow muddy water was established. By overlaying the weighted layers of different evaluation indicators on the geographic information system (GIS) platform, a risk degree distribution map of yellow muddy water disasters was generated. The evaluation results demonstrated that the disaster risk levels within the study area exhibited spatial differentiation, with areas of higher risk accounting for 14.76% of the total. The evaluation results were compared with historical yellow muddy water event information from Guangzhou, and the effectiveness of the model was verified by the receiver operating characteristic (ROC) curve. The validation results indicate that this model provides high accuracy in assessing the degree of risk of yellow muddy water in high-construction-intensity cities, offering effective technical support for precise disaster prevention and mitigation.

1. Introduction

In the rapid urbanisation process in China, intense and high-density production and construction activities have occupied ecological space, leading to changes in the structure of the original regional environment as well as soil and water loss [1]. The phenomenon of yellow muddy water caused by rainfall in high-construction-intensity cities has become a widely studied water environment issue. Yellow muddy water refers to water bodies with high turbidity and suspended sediment concentrations, primarily composed of fine-grained soil particles mobilised by rainfall-runoff processes in areas with exposed or disturbed soils.

Studies have shown that increased urban land use is one of the largest contributors of pollutants and sediments to global freshwater ecosystems [2]. Infrastructure construction (such as impervious pavements and rooftops) provides additional pathways for nonpoint source pollutants (such as urban surface runoff, lake-filling pollution, and atmospheric pollutant deposition) [3].

Because of the large amount of land in high-construction-intensity cities covered by hard surfaces, the soil erosion mechanism, as well as soil and water loss pathways, have become extremely complex. Moreover, most rainwater cannot enter the soil directly, resulting in reduced precipitation infiltration and increased surface runoff. Therefore, when rainfall exceeds the discharge capacity of an urban drainage system, it can easily cause surface waterlogging and urban flooding [4], accompanied by yellow muddy water.

Furthermore, during urban development and construction, pollutants such as sludge and sediment are easily eroded by natural rainfall and washed into urban surfaces and water bodies by surface runoff, causing soil and water loss, as well as water pollution issues. The abandoned soil and slag generated by construction projects and industrial production hinder rainstorms and flood discharge, resulting in the coexistence and aggravation of urban flood disasters, as well as water and soil loss. Sludge and silt can block urban drainage networks, resulting in issues such as the overflow of yellow muddy water. The entry of muddy water into rivers can deteriorate the water quality, negatively affecting aquatic organisms, ecosystems, and the quality of life of residents.

Previous studies have primarily concentrated on the causes and prevention of urban rainstorms and flood disasters [5,6,7,8], and the spatial differences in complex land use patterns and building environments in high-construction-intensity cities are often neglected. The application of existing research results to the risk assessment of yellow muddy water disasters is limited. Risk assessment of yellow muddy water disasters can provide a reference for disaster prevention and is of great significance in reducing disaster losses. It is, therefore, necessary to study the risk assessment methods for yellow muddy water disasters in cities with high construction intensity.

Three methods are commonly used for disaster risk assessment [5]. The historical disaster mathematical statistics method involves risk assessment and prediction based on the statistics and analysis of historical disaster events [9]. This method is primarily suitable for regions with adequate and reliable historical data and records (such as well-documented floodplains or earthquake-prone zones with long-term monitoring), which limits its application to areas with insufficient data [10]. The coupled analysis of hydrological and hydraulic models has a high computational accuracy, which is effective for simulating dynamic processes such as flood propagation, sediment transport, and pollutant dispersion [11]. However, achieving high-precision simulation analysis requires high-resolution data and expensive computational resources and is difficult to apply on a larger scale [12], which presents challenges for its practical application. Multicriteria decision analysis (MCDA) has relatively low data requirements and is economically efficient [13], allowing for flexible applications based on the specific conditions of the study area and the availability of data [14]. This method is suitable for areas that are easily affected by disasters but it is difficult to collect sufficient data. MCDA has significant advantages in dealing with mutual conflicts and competitive relationships among multiple factors, providing powerful decision-making support for policymakers [7]. MCDA was prioritised here due to the study area’s data availability and the need to balance competing factors (such as ecological exposure and economic vulnerability).

In recent years, numerous MCDA methods have been proposed to address various types of problems, including the analytic hierarchy process (AHP) [15,16,17], fuzzy AHP [18,19,20], analytic network process (ANP) [21,22,23], fuzzy logic [24,25,26], and artificial neural network (ANN) methods [27,28,29], among others. AHP is a commonly used MCDA method, which is a semi-quantitative approach that combines subjective and objective methods, making the calculation results more realistic. The AHP method utilises historical or field survey data based on the experience and expertise of researchers to analyse the correlation between disasters and their contributing factors. This breaks down complex problems into multiple simple evaluation indicators. By constructing comparison matrices, the relative importance of each evaluation indicator was determined, and weights were assigned. Additionally, the consistency of the evaluation indicators was validated through mathematical relationships [30,31,32].

The geographic information system (GIS) platform and remote sensing (RS) have been extensively applied in combination with the AHP method for the spatial analysis and geospatial mapping of disaster assessments [33,34]. The GIS-based analytic hierarchy process (GIS-AHP) technology can effectively integrate and analyse multi-source spatiotemporal data across large spatial scales, enabling the visualisation of risk maps, providing disaster risk maps for various scenarios, and rapidly determining the extent and severity of disasters [8,30]. Applications include urban flood risk assessment [5,7,8], flash flood susceptibility assessment [35], landslide hazard assessment [30,36], drought sensitivity assessment [37], and forest fire risk analysis [31,32], among others.

Using Guangzhou City as an example, this study constructed a risk assessment system for yellow muddy water and established a disaster risk assessment model for yellow muddy water in high-construction-intensity cities based on GIS-AHP. The research findings are intended to provide a scientific basis for the assessment and mitigation of yellow muddy water disasters, which hold significant theoretical value and practical importance for urban soil and water conservation, pollution control, and sustainable economic development.

2. Materials and Methods

2.1. Study Area

This study takes Guangzhou City, Guangdong Province, one of the high-construction-intensity cities in China’s Guangdong–Hong Kong–Macao Greater Bay Area, as the research object (Figure 1). Guangzhou has a very high population, high building density, intensive traffic, frequent economic activities, and abundant public facilities. However, the unique climatic and geographical characteristics of Guangzhou provide rich conditions for the occurrence of yellow muddy water.

Guangzhou is located at 112°57′~114°3′ E longitudes and 22°26′~23°56′ N latitudes, with a total area of 7434.4 km² [38]. The topography of Guangzhou is characterised by hilly terrain, with the elevation decreasing from northeast to southwest. The northeastern part is mountainous with medium and low mountains, the central part is a hilly basin, and the southern part is a coastal alluvial plain that is part of the Pearl River Delta.

Guangzhou is situated downstream of the Pearl River Basin and serves as a major national hub connecting the inland and oceans. The area has a well-developed river system with a dense network of rivers, with a river density of 0.75 km/km², providing complex natural terrain conditions for yellow muddy water.

Guangzhou is located in the subtropical monsoon climate zone [39], with an annual average temperature of 21.9 °C and an average relative humidity of 77%. The region receives abundant rainfall, with an annual precipitation of approximately 1800 mm, where the flood season is primarily concentrated from April to September [40]. In particular, unique hydrological conditions such as frontal rain, tropical cyclone rain, and convective rain, as well as extreme precipitation under the background of climate change, provide a strong rainfall erosion force for regional yellow muddy water.

Intensive development and construction activities in Guangzhou have led to fragmented terrain and complex underlying surfaces within the area. The development of underground spaces (such as underground garages and subways, etc.) has disrupted the natural water cycle. Moreover, the replacement of vegetated land with a large number of hard road surfaces has reduced the permeability of rainwater through the ground surface. There are piles of abandoned soil and slag at construction, roadwork, and municipal engineering sites, providing abundant material conditions for yellow muddy water.

2.2. Data Sources

For the purpose of yellow muddy water risk mapping, open-source spatial data for the study area, as well as information from local monitoring agencies, were collected (

Table 1. Data descriptions.

Data Type	Description	Source
Digital elevation model (DEM)	ASTER GDEM (30 m resolution) ASTGTM_N23E113 (14 July 2010)	Geospatial Data Cloud (http://www.gscloud.cn/, accessed on 1 July 2024)
Satellite imagery	Landsat 8 OLI_TIRS satellite (30 m resolution) LC81220432021339LGN00 (12 December 2021) LC81220442021339LGN00 (5 December 2021)	Geospatial Data Cloud (http://www.gscloud.cn/, accessed on 1 July 2024)
Precipitation data	2022 annual precipitation data in China (1 km resolution)	National Earth System Science Data Center (https://www.geodata.cn/, accessed on 1 July 2024)
Soil data	2015 soil erosion modulus in China (250 m resolution)	Resource and Environmental Science Data Platform (https://www.resdc.cn/, accessed on 1 July 2024)
Land use land cover data	2020 global land cover fine classification data (30 m resolution)	Earth big data science engineering data sharing service system (https://data.casearth.cn/, accessed on 1 July 2024)
Road network data	Various levels of road network data (vector data)	Open Street Map (https://openstreetmap.us/, accessed on 1 July 2024)
Construction site image data	2023 research area construction disturbance image data manually drawn using a visual interpretation method (vector data)	Soil and Water Conservation Monitoring Center of Pearl River Basin
Population data	2020 China population spatial distribution kilometre grid dataset (1 km resolution)	Resource and Environmental Science Data Platform (https://www.resdc.cn/, accessed on 2 July 2024)
Gross Domestic Product (GDP) data	2020 China GDP spatial distribution kilometre grid dataset (1 km resolution)	Resource and Environmental Science Data Platform (https://www.resdc.cn/, accessed on 2 July 2024)

).

2.3. Methodology

The occurrence of yellow muddy water is influenced by multiple factors, and its risk assessment in high-construction-intensity cities is a complex issue that intersects natural and social disciplines. Therefore, a comprehensive understanding of the causes and factors influencing yellow muddy water disasters is essential.

This study focuses on the risk assessment of yellow muddy water disasters in high-construction-intensity cities, considering yellow muddy water disasters to be the result of the combined effects of disaster-causing factors, disaster-pregnant environments, and disaster-bearing entities. Therefore, three evaluation criteria—hazard, exposure, and vulnerability—have been established [41]. Based on the evaluation criteria and characteristics of the study area, 11 evaluation indicators were identified—rainfall, construction site, elevation, slope, normalised difference vegetation index (NDVI), soil erosion modulus, stream power index (SPI), surface permeability, road spatial distribution, population spatial distribution, and GDP (Gross Domestic Product) spatial distribution. The evaluation indicator variables were rasterised in ArcMap 10.7 (Environmental Systems Research Institute, Redlands, CA, USA, 2019), with each indicator layer created as a raster layer. Considering the scale of the study area and to maintain the accuracy of the data for each indicator as much as possible, a raster spatial resolution of 30 m × 30 m was uniformly applied to each layer.

The importance of each evaluation indicator varied according to the risk assessment results. The AHP method was used to assign weights to 11 evaluation indicators. These indicators were standardised to address the order-of-magnitude differences among the various evaluation indicators. Natural break and binary classification methods were utilised to segment the range of variable values and assign ratings to each evaluation indicator. Through a weighted overlay of the raster layer of evaluation indicators in ArcMap 10.7 software, a risk degree distribution map of yellow muddy water disasters was generated, categorising disaster risk into five levels—very low, low, medium, high, and very high—so as to show the risk level distribution of yellow muddy water in the study area. The receiver operating characteristic (ROC) curve combined with historical yellow muddy water incident event points was used to verify the accuracy of the risk degree distribution map. The framework for assessing the risk of yellow muddy water in cities with high construction intensity is presented in Figure 2.

2.4. Evaluation Indicators

2.4.1. Hazard

The primary sources of yellow muddy water in cities with a high construction intensity are rainfall and sediment produced from construction activities. Therefore, the risk of disaster-causing factors was represented by rainfall and construction sites.

(1)

Rainfall

Against the backdrop of global warming, the frequency and intensity of extreme rainfall events have increased in recent years [42]. In addition, the urban heat island effect causes atmospheric instability in cities, thereby enhancing air convection. Under certain weather conditions, convection can trigger the formation of convective clouds and precipitation, promoting showers and heavy rainfall events. The phenomenon in which the urban precipitation intensity is higher than that in the surrounding suburban areas is particularly noticeable. Such short durations or sustained heavy rainfall often lead to urban flooding and soil erosion, accompanied by the occurrence of yellow muddy water. The greater the rainfall, the more likely it is for yellow muddy water to occur. The precipitation data used in this study were extracted from the 2022 annual precipitation data in China (National Earth System Science Data Center, https://www.geodata.cn/, accessed on 1 July 2024). Rainfall in the study area increases from southwest to northeast, as presented in Figure 3a.

(2)

Construction site

Rapid urban development has led to the conversion and destruction of large areas of land resources such as farmland, forests, and green belts. This process alters the original geographical environment and disrupts the natural surface structure, exposing a significant amount of soil to the air and making it vulnerable to wind and water erosion, thereby exacerbating soil erosion [43]. Moreover, during the urbanisation process, the shapes of river networks are often artificially altered, with non-mainstream rivers in highly urbanised areas gradually decreasing, simplifying river network structures. If certain sections along rivers are occupied by a large number of illegal constructions, it can restrict the development of river tributaries and further contribute to the shrinkage of river flows.

In addition, as the scale of urban construction continues to expand, infrastructure development, such as buildings, roads, and subways, has become routine. Earthwork operations and material stockpiling at construction sites expose large amounts of sediment and aggregates, making them susceptible to being blown away and dispersed into the surrounding environment by the wind [44]. Piles of discarded soil and debris can obstruct flood paths during heavy rainfall events. Pollutants generated during construction can enter urban surfaces and water bodies through wind erosion, rainwater runoff, and surface flows. Notably, the illegal discharge of slurry from construction sites can directly and adversely affect urban drainage systems, leading to clogged drainage networks, yellow muddy water overflow, and urban flooding.

Images of the construction sites within the study area were provided by local soil and water conservation regulatory authorities (Soil and Water Conservation Monitoring Center of Pearl River Basin), as presented in Figure 3b.

2.4.2. Exposure

Generally, topographic factors, hydrological environments, and vegetation are used to comprehensively reflect disaster-pregnancy conditions. In this study, elevation, slope, NDVI, soil erosion modulus, SPI, surface permeability, and road spatial distribution were selected to represent exposure to yellow muddy water disaster-pregnant environments.

(1)

Elevation

Topographic factors significantly influenced the development of yellow muddy water. Variations in elevation directly affect the water flow velocity. In areas with lower elevations, the flow rate is slower, facilitating the deposition of yellow muddy water [45]. Conversely, in higher-elevation areas, the flow rate increases, enhancing erosion and promoting the transport of yellow muddy water to lower-elevation regions. Therefore, low-altitude areas are more susceptible to yellow muddy water disasters than high-altitude areas. Elevation raster data for the study area were extracted using a DEM (Geospatial Data Cloud, http://www.gscloud.cn/, accessed on 1 July 2024). As presented in Figure 4a, the elevation of the study area decreases from northeast to southwest.

(2)

Slope

The smaller the slope, the stronger the water convergence ability of the surface. Gentle slopes can slow the flow of water [46], allowing rainwater to remain on the surface for a longer time, and facilitating the infiltration and formation of runoff on the surface. Under such conditions, sediments are more easily carried and accumulated by the water flow. Therefore, areas with relatively flat terrain were more prone to yellow muddy water. The slope of the study area was calculated using DEM (Geospatial Data Cloud, http://www.gscloud.cn/, accessed on 1 July 2024). As presented in Figure 4b, the terrain in the southern region is flatter than that in the northern region.

(3)

NDVI

Vegetation has a certain ability to conserve soil and water. The root systems of vegetation help stabilise the soil structure, reduce the effects of wind and water erosion, and mitigate soil erosion. Urban green spaces can effectively store rainwater, and the impact of vegetation on urban flooding largely depends on the area and biophysical parameters [47]. Healthier or denser vegetation canopies and root systems are better at intercepting surface runoff [48,49]. Studies have shown that an increase of 7.5% in vegetation cover within a watershed can reduce surface runoff by more than 90% [50].

NDVI is commonly used to reflect vegetation cover and growth status [51,52]. Mathematically, NDVI is expressed as follows:

$$NDVI={\displaystyle \frac{NIR-RED}{NIR+RED}}$$

(1)

where NDVI is the normalised difference vegetation index. NIR and RED are the spectral radiance (or reflectance) measurements recorded with sensors in the near-infrared and red (visible) regions, respectively [53]. The NDVI value ranges from −1 to 1, with values greater than 0 indicating vegetation-covered surfaces. The higher the NDVI value, the better the vegetation growth conditions, which are conducive to reducing the risk of yellow muddy water disasters. In ENVI 5.6 (Harris Geospatial Solutions, Boulder, CO, USA, 2020), the NDVI values of the study area were calculated and analysed based on satellite imagery from Landsat 8 (Geospatial Data Cloud, http://www.gscloud.cn/, accessed on 1 July 2024). As presented in Figure 4c, vegetation in the northern part of the study area was more abundant than that in the southern part.

(4)

Soil erosion modulus

The characteristics of the underlying surface influences that lead to runoff development are described in [54]. When soil is eroded by rain or other forces, the sediment is detached and transported by the water flow, forming yellow muddy water. The soil erosion modulus, which represents the soil and water loss per unit area, is calculated based on the Revised Universal Soil Loss Equation (RUSLE) [55]. Mathematically, the RUSLE is written as follows:

$$A=R\times K\times L\times S\times C\times P$$

(2)

where A is the annual soil loss (t ha⁻¹ y⁻¹), R is the rainfall erosivity factor (MJ mm ha⁻¹ h⁻¹ y⁻¹), K is the soil erodibility factor (t h MJ⁻¹ mm⁻¹), L is the slope length factor (m), S is the slope gradient factor (%), C is the cover factor, and P is the conservation practice factor [55].

The soil erosion modulus directly reflects the intensity and scale of soil erosion and indirectly affects the formation and severity of yellow muddy water. Areas with higher soil erosion moduli had a greater probability of yellow muddy water occurring during rainfall events. The soil data within the study area were extracted from the 2015 water erosion modulus data of soil in China (250 m resolution), as presented in Figure 4d (Resource and Environmental Science Data Platform, https://www.resdc.cn/, accessed on 1 July 2024).

(5)

SPI

SPI reflects the impact of topography on hydrodynamic processes. As a quantitative indicator of the erosion capacity of surface water flow, SPI can directly influence surface runoff and soil erosion processes in a region under specific rainfall conditions. A higher SPI value indicates stronger water flow dynamics in the area, which have a greater impact on soil erosion and are more likely to lead to the yellow muddy water phenomenon. SPI is typically calculated based on the terrain slope and specific catchment area (SCA) [56]. The SPI was calculated as follows:

$$SPI=ln(A_s\tan \beta )$$

(3)

where SPI is the stream power index, A_s is the local upslope contributing area from the flow accumulation raster (m²), and β is the local slope angle (degree) [56]. Using the hydrology tools in ArcMap 10.7, the SPI for the study area was calculated based on the DEM (Geospatial Data Cloud, http://www.gscloud.cn/, accessed on 1 July 2024), as presented in Figure 4e.

(6)

Surface permeability

Surface permeability directly limits water infiltration, which, in turn, determines the generation of rainfall runoff [57]. The promotion of urbanisation has a significant negative impact on the natural hydrological cycle. In natural environments, rainwater can be absorbed by the soil and enter the natural water cycle through vegetation transpiration or underground infiltration. However, because of the widespread presence of buildings and impervious surfaces in urban areas, most rainwater cannot directly infiltrate the soil, leading to increased urban surface runoff [58].

The land use types in the study area (reservoir, river, lake, stream, farmland, herbaceous cover, shrub cover, broadleaf forest, coniferous forest, wetland, bare land, and urban construction land, obtained from the Earth big data science engineering data sharing service system, https://data.casearth.cn/, accessed on 1 July 2024) were classified into three categories: water bodies, permeable surfaces, and impermeable surfaces. Water bodies include reservoirs, rivers, lakes, and streams of various sizes, permeable surfaces include farmland, herbaceous cover, shrub cover, broadleaf forests, coniferous forests, wetlands, and bare land. The remaining urban construction land is classified as impermeable surfaces. Greater surface permeability helps to reduce the risk of yellow muddy water disasters. As presented in Figure 4f, the southern part of the study area has more impermeable surfaces and water systems than the northern part.

(7)

Road

Urban road rainwater is typically directed towards roadside drains and drainage networks. Poor drainage can occur if the drainage system is not smooth or the road surface is uneven during heavy rain, causing yellow muddy water accumulation and flooding disasters on the road surface or under bridge culverts. Therefore, roads are high-incidence areas for yellow muddy water. Road network data within the study area were obtained from Open Street Map (https://openstreetmap.us/, accessed on 1 July 2024). As presented in Figure 4g, the roads in the southwest region of the study area were denser than those in the northeast.

2.4.3. Vulnerability

The primary disaster-bearing entities of urban yellow muddy water are the social population and economy; hence, the spatial distributions of the population and GDP are used to represent the vulnerability of disaster-bearing entities.

(1)

Population

Yellow muddy water sullies the surface of urban blocks and pollutes the surrounding water bodies and environments, damaging urban landscapes, degrading city images, and impacting residents’ quality of life. Sediment-slippery road surfaces reduce visibility, increasing the risk of pedestrians and cyclists falling, while also reducing vehicle speed and safety, making it more difficult for residents to travel, and increasing the risk of traffic accidents [59,60,61].

Moreover, yellow muddy water may carry large amounts of suspended solids and organic matter as well as various toxic and hazardous substances (such as heavy metals, petroleum products, chemical waste, and pathogens, including viruses and bacteria). These substances can enter reservoirs, rivers, lakes, and groundwater through runoff, contaminating water supply systems, reducing the availability of water sources, increasing the difficulty of water purification at treatment plants, threatening the safety of residents’ water use, and exacerbating water scarcity and supply–demand conflicts [62].

Although yellow muddy water does not directly threaten residents’ lives, such as through landslides or floods, its negative impacts on daily travel activities and water safety cannot be ignored, particularly in densely populated areas where such impacts are more significant. We extracted the population spatial distribution data within the study area from the 2020 China population spatial distribution kilometre grid dataset (Resource and Environmental Science Data Platform, https://www.resdc.cn/, accessed on 2 July 2024). As presented in Figure 5a, the southwestern part of the study area has a higher population density.

(2)

GDP

The suspension, transportation, and deposition of sediment particles in water bodies due to soil erosion can reduce water levels and water volume, thereby affecting the growth of aquatic plants and habitat space for aquatic animals. Suspended pollutants impede light transmission, reduce water transparency, and affect the photosynthesis and development of aquatic plants. The entry of large amounts of nutrients (such as nitrogen and phosphorus) into rivers and lakes can lead to eutrophication, resulting in excessive algal growth and blooms, which, in turn, cause oxygen depletion in aquatic organisms. Pollutants can degrade the water quality and impair the normal ecological balance of aquatic ecosystems, including aquatic animals, plants, and microorganisms [63]. Moreover, solid particles and chemicals in muddy water can degrade soil quality and fertility, damage soil environments, and lead to soil degradation and farmland yield reduction [64].

Sediments and pollutants entering urban drainage systems can accumulate and block pipes and channels, thereby hindering rainwater discharge. When drainage systems fail to effectively remove rainwater, it accumulates on the surface, causing surface waterlogging and urban flooding [65]. The deposition of sediment into river systems can also lead to an increase in river and lake water levels, heightening flood risks, and causing hydrological changes. In severe cases, it can damage urban infrastructure, such as vehicles, houses, roads, and bridges, causing economic losses and disrupting other economic activities, thereby hindering sustainable urban development.

It is evident that urban yellow muddy water can cause considerable losses to both the socioeconomic and ecological environments, as well as adverse effects on property safety, including infrastructure facilities and transportation tools. The higher the production value in a region, the greater the economic losses incurred. The GDP spatial distribution data within the study area were extracted from the 2020 China GDP spatial distribution kilometre grid dataset (Resource and Environmental Science Data Platform, https://www.resdc.cn/, accessed on 2 July 2024). As presented in Figure 5b, the GDP in the southwestern part of the study area was higher than that in other regions.

2.5. Weighting

2.5.1. Establish AHP Comparison Matrix

The importance of each evaluation criterion was pairwise compared to the evaluation objective and the importance of each evaluation indicator to the evaluation criteria. The importance rating criteria were obtained using the nine-integer value scale suggested by Saaty [15,16], as presented in

Table 2. Scales for pairwise comparisons.

Importance Scale	Definition
1	Equal importance
3	Moderate importance
5	Essential importance
7	Demonstrated importance
9	Extreme importance
2, 4, 6, 8	Intermediate value

, to obtain the comparison matrices.

2.5.2. Calculate the Weight

The eigenvector w corresponding to the maximum eigenvalue of the comparison matrix was calculated, where each component of w represents the importance of each evaluation indicator, and the weight value is the normalised result of the eigenvector.

Table 3. Each evaluation criteria’s comparison matrix and their weights.

Evaluation Criteria	Hazard	Exposure	Vulnerability	Weight Value (%)
Hazard	1	2	1	41.26
Exposure	1/2	1	1	25.992
Vulnerability	1	1	1	32.748

presents the comparison matrices for each evaluation criterion and their weights.

Table 4. Each hazard evaluation indicator’s comparison matrix and its weights.

Hazard Evaluation Index	Rainfall	Construction Site	Weight Value (%)
Rainfall	1	1	50
Construction site	1	1	50

,

Table 5. Each exposure evaluation indicator’s comparison matrix and its weights.

Exposure Evaluation Index	Elevation	Slope	NDVI	Soil Erosion Modulus	SPI	Surface Permeability	Road	Weight Value (%)
Elevation	1	1/2	1/2	1/3	1/2	1/3	1/3	5.744
Slope	2	1	1/2	1/3	1/2	1/3	1/3	7.002
NDVI	2	2	1	1/3	2	1/2	1/2	11.683
Soil erosion modulus	3	3	3	1	2	1/2	1/2	17.956
SPI	2	2	1/2	1/2	1	1/3	1/3	9.045
Surface permeability	3	3	2	2	3	1	2	26.682
Road	3	3	2	2	3	1/2	1	21.888

and

Table 6. Each vulnerability evaluation index’s comparison matrix and its weights.

Vulnerability Evaluation Index	People	GDP	Weight Value (%)
People	1	1	50
GDP	1	1	50

present the comparison matrix and weights of each evaluation indicator for hazard, exposure, and vulnerability, respectively.

2.5.3. Consistency Check

To verify whether the weight allocation is reasonable, a consistency check was performed on the comparison matrix as follows:

$$CR={\displaystyle \frac{CI}{RI}}$$

(4)

$$CI={\displaystyle \frac{1}{n-1}}\left({\lambda }_{\max }-n\right)$$

(5)

where CR is the random consistency ratio of the comparison matrix, CI represents the consistency index, and RI is the average random consistency index, which depends on the number of criteria compared [15,16], as presented in

Table 7. RI of the comparison matrix.

Number of Criteria (n)	1	2	3	4	5	6	7	8	9	10
RI	0.00	0.00	0.58	0.90	1.12	1.24	1.32	1.41	1.45	1.49

. Additionally, λ_max is the maximum eigenvalue of the comparison matrix and n is the number of evaluation indicators.

When CR was less than 0.1, the consistency of the comparison matrix was considered satisfactory, indicating that the distribution of the weight coefficients was reasonable. Otherwise, the comparison matrix must be adjusted until satisfactory consistency is achieved. The CR threshold of 0.1 in the AHP method is a widely accepted criterion established by Thomas L. Saaty, the creator of AHP [15,16].

In Table 3, the λ_max of the evaluation indicators is 3.054 and n is 3. According to Table 7, the corresponding RI value is 0.58 and is calculated using Equations (3) and (4); CR = 0.047 < 0.1, passing the consistency test.

In Table 5, λ_max of the evaluation indicators is 7.313, and n is 7. According to the corresponding RI value of 1.32, CR = 0.040 < 0.1, as listed in Table 7, which passes the consistency test.

2.6. Yellow Muddy Water Risk

2.6.1. Standardisation of Evaluation Indicators

The evaluation indicators were standardised to eliminate differences in magnitude, dimension, and nature among the various indicators [66]. Each of the 11 evaluation indicators was assessed using the binary classification method or the natural break method for assigning values. The natural break method maximises the difference between classes based on the distribution patterns of numerical statistics.

Table 8. Standardisation and rating assignment for evaluation indicators.

Evaluation Indicators	Attribute	Unit	Class Ranges	Class	Class Ratings
Rainfall	Positive	mm	1710~1780	Very low	1
			1780~1832	Low	2
			1832~1878	Moderate	3
			1878~1945	High	4
			1945~2106	Very high	5
Construction sites	Positive		Non-construction land	Very low	1
			Construction land	Very high	5
Elevation	Negative	m	620~1162	Very low	1
			373~620	Low	2
			209~373	Moderate	3
			78~209	High	4
			0~78	Very high	5
Slope	Negative	degree	24.34~62.08	Very low	1
			16.07~24.34	Low	2
			9.25~16.07	Moderate	3
			3.41~9.25	High	4
			0~3.41	Very high	5
NDVI	Negative		0.70~1	Very low	1
			0.47~−0.70	Low	2
			0.22~0.47	Moderate	3
			−0.23~0.22	High	4
			−1~−0.23	Very high	5
Soil erosion modulus	Positive	t ha−1 y−1	0~19.14	Very low	1
			19.14~61.52	Low	2
			61.52~129.88	Moderate	3
			129.88~233.79	High	4
			233.79~350	Very high	5
SPI	Positive		−8.46~−3.05	Very low	1
			−3.05~0.68	Low	2
			0.68~2.54	Moderate	3
			2.54~5.06	High	4
			5.06~15.40	Very high	5
Surface permeability	Negative		Waterbody	Very low	1
			Pervious surface	Low	2
			Impervious surface	Very high	5
Road	Positive		Non-road	Very low	1
			Road	High	4
Population	Positive	persons	0~1105	Very low	1
			1105~2943	Low	2
			2943~6649	Moderate	3
			6649~15,690	High	4
			15,690~35,070	Very high	5
GDP	Positive	10,000 RMB	0~25,716	Very low	1
			25,716~94,389	Low	2
			94,389~209,393	Moderate	3
			209,393~407,706	High	4
			407,706~1,140,174	Very high	5

presents the standardised value assignment table for the evaluation indicators, with values ranging from 1 to 5, which are proportional to the increase in the degree of risk of yellow muddy water associated with each indicator.

2.6.2. Risk Degree

The risk of yellow muddy water disasters is the result of the combined effects of disaster-causing factors, exposure to disaster-pregnant environments, and the vulnerability of disaster-bearing entities. The impact capacity of the evaluation indicators for each criterion on the disaster varied. Therefore, a risk assessment model for yellow muddy water was constructed as follows [5,67]:

$$H=f\left(h\right)=\sum _{i=1}^n\left({\omega }_i{h}_i\right)$$

(6)

$$S=f\left(s\right)=\sum _{j=1}^n\left({\omega }_j{s}_j\right)$$

(7)

$$V=f\left(v\right)=\sum _{k=1}^n\left({\omega }_k{v}_k\right)$$

(8)

$$R=f\left(H,S,V\right)={\omega }_{H}H+{\omega }_{S}S+{\omega }_{V}V$$

(9)

where H, S, V, and R represent the hazard, exposure, vulnerability, and degree of risk, respectively, of the yellow muddy water; $h_i$, $s_j$, $v_k$ and $w_i$, $w_j$, and ${\omega }_k$ represent the standardised variable ratings for each evaluation indicator of hazard, exposure, and vulnerability, and their corresponding weights, respectively; n denotes the number of evaluation indicators; and ${\omega }_H$, ${\omega }_S$, and ${\omega }_V$ are the weights of the hazard, exposure, and vulnerability, respectively. Using the raster calculator function in the ArcMap 10.7 software, a risk distribution map of yellow muddy water disasters in the study area was generated based on the weighted values of each evaluation indicator for each pixel.

3. Results

3.1. Assessment Results

Based on the proposed risk assessment index system and index weights for yellow muddy water in high-construction-intensity cities, the weighted sums of each evaluation criterion layer and target layer were calculated using the ArcMap 10.7 software. A hazard distribution map (Figure 6a) was obtained by overlaying the hazard evaluation indicators of rainfall and construction sites in the study area. The exposure distribution map (Figure 6b) was generated by overlaying the exposure evaluation indicators, including the elevation, slope, NDVI, soil erosion modulus, SPI, surface permeability, and road spatial distribution. A vulnerability distribution map (Figure 6c) was created by overlaying the vulnerability evaluation indicators of the population and GDP spatial distributions. Finally, the risk degree distribution map for yellow muddy water disasters in the target layer (Figure 6d) was obtained by overlaying all evaluation indicator layers according to their respective weights. The natural break method was adopted to classify evaluation levels into five categories: very low, low, moderate, high, and very high.

Figure 6a shows that for the hazard evaluation criteria, the area proportions of regions classified as very low (1–1.49), low (1.49–1.99), moderate (1.99–2.49), high (2.49–3.49), and very high (3.49–5) hazard levels were 7.91%, 24.29%, 28.05%, 36.99%, and 2.76%, respectively. The combined proportion of high and very high hazard areas was 39.75%. These areas are primarily concentrated in regions with high rainfall and frequent construction activities.

Figure 6b shows that for the exposure evaluation criteria, the area proportions of regions classified as having very low (1.32–1.95), low (1.95–2.45), moderate (2.45–2.95), high (2.95–3.37), and very high (3.37–4.67) exposure levels were 40.03%, 33.73%, 11.54%, 10.79%, and 3.91%, respectively. Areas with very low and low exposure levels account for 73.76% of the total. The proportion of areas with high and very high exposure levels was 14.7%, which was relatively small.

Based on Figure 6c, it may be noted that for the vulnerability evaluation criteria, the area proportions of regions classified as very low (1–1.01), low (1.01–1.99), moderate (1.99–2.99), high (2.99–4), and very high (4–5) vulnerability levels are 48.56%, 15.88%, 25.25%, 6.63%, and 3.68%, respectively. Very low vulnerability areas occupy the largest proportion (48.56%) and are characterised by lower population density and dispersed economic activities. The combined proportion of high and very high vulnerability areas was relatively small, accounting for only 10.31%.

Figure 6d shows that the disaster risk levels of yellow muddy water in the study area exhibit spatial differentiation. The proportions of regions classified as very low (1.3–1.73), low (1.73–1.97), moderate (1.97–2.27), high (2.27–2.71), and very high (2.71–4.16) risk were 23.11%, 33.06%, 29.07%, 10.17%, and 4.59%, respectively. The disaster risk assessment results indicated that areas with lower risk levels dominated, accounting for 56.17% of the total area. Areas with higher risk levels accounted for 14.76% of the total area and require particular attention.

3.2. Validation

To verify the effectiveness of the proposed risk assessment model for yellow muddy water in high-construction-intensity cities, data were collected from 81 historical yellow muddy water events reported in Guangzhou news in 2023. Figure 7 shows the situation of yellow muddy water incidents in some areas of Guangzhou in 2023. Figure 7a displays the illegal discharge of yellow muddy water from construction sites, while Figure 7b depicts the appearance of yellow muddy water on streets near construction sites after rainfall. As shown in Figure 7, the phenomenon of yellow muddy water pollution in Guangzhou is particularly severe.

Figure 8a presents the spatial distribution of these 81 historical yellow muddy water events on the risk degree map presented in Figure 6d. The numbers of historical yellow muddy water events that overlapped with very high-, high-, moderate-, and low-risk degrees were 45, 27, 7, and 2, respectively, and none of these points were located in the very low-risk region. Among the historical yellow muddy water locations, 88.89% were classified as being within the very high- or high-risk areas.

Additionally, 8.64% of the historical yellow muddy water locations were distributed in moderate-risk areas and 2.47% were located in low-risk areas. This may be due to the model’s incomplete capture of specific influencing factors in certain regions when considering the spatial heterogeneity and non-equilibrium development of high-construction-intensity cities, leading to deviations in the risk degree classification. Moreover, the accuracy and timeliness of the data sources for the evaluation indicators may have affected the evaluation results.

The ROC curve was used to verify the reliability of the proposed model. The historical yellow muddy water event points were designated as positive samples, while 82 randomly generated points within the study area (created using ArcMap 10.7) served as negative samples. In the ROC curve analysis, the area under the curve (AUC) represents the model accuracy. At AUC = 1.0, the assessment model is considered a perfect evaluation method under ideal conditions. At AUC = 0.5–1.0, the assessment model is considered a reasonable model superior to random evaluation; the closer the value is to 1.0, the more accurate the classification is considered to be [5,30]. In this study, the AUC of the classification model was 0.879 for the study area, which indicated high accuracy of the assessment results. The generated ROC curve is shown in Figure 8b.

Overall, this model demonstrates high accuracy in assessing the degree of risk of yellow muddy water in high-construction-intensity cities, providing potent technical support for precise disaster prevention, mitigation, and post-disaster recovery.

4. Discussion

4.1. Disaster Characteristics

According to the results of hazard evaluation, the yellow muddy water disaster-prone areas are primarily concentrated in regions with high rainfall and frequent construction activities where risk management and preventive measures are crucial. Therefore, in future development and planning, it is essential to comprehensively consider the influence of natural factors and construction activities on the spatial distribution of yellow muddy water.

The exposure evaluation reveals that areas with lower exposure levels should focus on maintaining the advantages of their ecological bases and reinforcing the connectivity of their green infrastructure [68]. For regions with moderate exposure levels, it is necessary to optimise the land use structure and ensure surface permeability through low-impact development strategies. The regions with higher exposure levels pose significant potential environmental risks and should be prioritised for prevention and control efforts, including the implementation of interception and purification projects, improvements to road drainage systems, and enhanced ecological restoration.

The vulnerability evaluation results indicate that areas with higher vulnerability levels have higher population and building densities as well as relatively frequent socioeconomic activities, necessitating stricter disaster prevention measures.

According to the results of the disaster risk degree assessment, the spatial differentiation distribution provides a scientific basis for the differentiated deployment of disaster prevention efforts, facilitating the concentration of limited resources in higher-risk areas, while also considering the overall disaster prevention system construction in the region. Current disaster prevention measures should be maintained alongside the routine monitoring of areas with very low- and low-risk levels. Areas with moderate risk levels require dynamic risk assessments and targeted preventive measures to prevent the escalation of risk levels. An in-depth analysis of disaster-causing factors is necessary for areas with high and very high levels, prioritising the allocation of disaster-prevention resources and implementing precise management measures.

4.2. Prevention and Control Measures

To reduce the occurrence of yellow muddy water in cities with high construction intensity, a series of measures must be taken to regulate construction behaviours and strengthen construction environmental management. It is necessary to enhance dust suppression at various construction sites by increasing water spraying, setting up temporary or permanent dust barriers, and utilising covering techniques to mitigate dust pollution and soil erosion caused by flying particles such as soil and gravel [69]. Rainwater and sewage separation systems should be established to ensure that wastewater and sewage generated at construction sites are collected and treated before discharge. Dedicated sewage disposal facilities, such as settlement tanks and filtration equipment, should be installed to effectively filter out sediment and contaminants from water. The Guangzhou Local Standard for Construction Site Drainage stipulates that the suspended solids concentration in construction site discharge water must not exceed 50 mg/L.

With urban expansion, traditional drainage systems in Guangzhou face significant challenges, leading to frequent flooding and pollution issues [70]. Optimisation and improvement measures should be implemented when designing, constructing, and managing urban drainage systems. The efficiency of drainage systems can be improved by increasing the number of drainage facilities in highly vulnerable areas, enhancing the density of drainage networks, expanding pipe diameters, and increasing the lift height and flow rate of pumping stations. A rational layout and capacity design of facilities should be ensured to adapt to future urban development and climate change. It is vital to strengthen the routine management and monitoring of drainage facilities and conduct regular cleaning, dredging, and checking. Specifically, periodic cleaning can prevent blockages caused by debris accumulation, while dredging ensures that the drainage capacity is maintained, particularly during heavy rainfall events. The research and selection of pipeline materials may be intensified to improve the durability and service life of pipe networks, thereby preventing accidents such as leaks and ruptures. Normal operation of urban drainage systems should be ensured to reduce the risk of waterlogging.

Urban planning should prioritise the ecological balance by maintaining sufficient natural water absorption areas and reducing impermeable surfaces to minimise the disruption of natural waterways caused by urban construction. The Guangzhou Ecological Civilization Construction Plan proposes that by the end of 2025, over 45% of the city’s built-up areas will meet the requirements of sponge city construction. The rational planning of urban green infrastructure, including the composition and spatial arrangement of urban parks, grasslands, wetlands, and forests, should be emphasised. Initiatives such as green roofs, permeable pavements, and similar eco-friendly construction should be promoted to enhance urban rainwater infiltration and storage capacity. The adoption of green buildings, low-carbon technologies, and low-impact development strategies along with the use of sustainable building materials [71] should be encouraged to mitigate the environmental impacts of urban construction and development.

5. Conclusions

During the process of urbanisation, the yellow muddy water problem caused by rainfall in high-construction-intensity cities has attracted much attention. In this study, the risk assessment model of yellow muddy water in high-construction-intensity cities was proposed for the first time, and the impact of construction sites on yellow muddy water was fully considered. Based on the GIS-based analytic hierarchy process (GIS-AHP) technology, taking Guangzhou, China, as the research object, three evaluation criteria were set: the hazard of yellow muddy water disaster-causing factors, the exposure of disaster-pregnant environments, and the vulnerability of disaster-bearing entities. The evaluation indicators for hazards are based on rainfall and construction sites. The exposure indicators include elevation, slope, normalised difference vegetation index (NDVI), soil erosion modulus, stream power index (SPI), surface permeability, and roads. The vulnerability indicators include population number and GDP (Gross Domestic Product). The analytic hierarchy process (AHP) method was used to determine the weights of each evaluation indicator, and standardisation was applied to the indicators, along with the classification of their variable value ranges and assignment ratings, to construct a risk assessment system for yellow muddy water in high-construction-intensity cities. A weighted overlay of the raster layers of the evaluation indicators was conducted on a GIS platform to generate distribution maps for the hazard, exposure, vulnerability, and degree of risk of yellow muddy water disasters in Guangzhou. The natural break classification method was adopted to divide the degree of evaluation into five categories: very low, low, moderate, high, and very high.

The evaluation results indicated that the proportion of higher hazard level areas in the study region was 39.75% and was primarily concentrated in areas with high rainfall and frequent construction activities. The area of higher exposure regions accounts for 14.7%, which, due to their prominent potential environmental risks, should be prioritised for prevention and control efforts. The proportion of areas with higher vulnerability was 10.31%, characterised by high population and building density as well as frequent socioeconomic activities. Disaster risk levels exhibited spatial differentiation, with higher-risk areas accounting for 14.76%. In these areas, it is essential to conduct in-depth analyses of disaster-causing factors, prioritise the allocation of disaster prevention resources, and implement targeted governance measures.

The historical yellow muddy water events reported in Guangzhou news in 2023 combined with the ROC curve were used to verify the risk assessment model. The AUC value was determined to be 0.879, indicating that the assessment model has high accuracy and can provide robust technical support for precise disaster prevention and mitigation of yellow muddy water in cities with high construction intensity. Additionally, this paper proposes prevention and control measures for yellow muddy water in high-construction-intensity cities, which primarily focus on standardising construction behaviours and enhancing supervision of the construction environment, optimising and improving the design, construction, and management of drainage systems, and emphasising ecological balance in urban planning to reasonably plan urban green infrastructure.