Risk Assessment of Geological Hazards Based on Multi-Condition Development Scenarios: A Case Study of Huangshi Town, Guangdong Province

Abstract

This study focuses on Huangshi Town, Longchuan County, Guangdong Province, as the research area. By utilizing existing data and field surveys, the study identifies geological hazards and risks in the area, deeply analyzes the formation mechanisms and disaster-causing patterns, and systematically summarizes the developmental characteristics and spatial distribution of these hazards. Methodologically, the research combines regular grids and slope units, selecting ten evaluation factors for correlation analysis. The information theory model is used to assess disaster susceptibility. The study further evaluates the impact of three rainfall scenarios—100 mm of rainfall over 24 h, 250 mm of rainfall over 24 h, and 240 mm of effective rainfall over 72 h—on geological disasters in Huangshi Town. As a result, a comprehensive hazard assessment under multiple rainfall scenarios is produced. The findings show that the Receiver Operating Characteristic (ROC) accuracy of the disaster-susceptibility evaluation reaches 0.8739, ensuring high data quality for the geological hazard assessment in Huangshi Town. The zoning results align closely with field survey observations. In conclusion, incorporating rainfall as a triggering factor enhances the accuracy of the susceptibility analysis by better capturing the temporal and spatial patterns of landslide occurrences, thereby offering a more comprehensive understanding of the geological hazard development in the study area.

1. Introduction

Landslide geological disasters, triggered by natural or anthropogenic factors under complex geological conditions, pose serious threats to human life and property and constrain local socio-economic development [1]. Longchuan County, located in the mountainous and hilly area of northeastern Guangdong Province, features complex geological structures and a harsh climate, making it a high-susceptibility area for geological hazards [2]. In 2022, Guangdong Province experienced the most intense “Dragon Boat Water” rainfall in nearly two decades. Due to its prolonged duration and high cumulative rainfall, Longchuan County suffered multiple landslides and collapses. The disaster points were widespread, causing casualties, missing persons, and serious damage to local production and livelihoods. The risk of geological disasters in the region remains extremely severe [3,4]. To implement national and provincial directives on geological disaster prevention and control, the Guangdong Provincial Department of Natural Resources launched a pilot project for a 1:10,000-scale detailed geological hazard investigation in Longchuan County. This project aims to identify hidden hazards, enhance monitoring and early warning systems, support rural planning, and promote comprehensive disaster mitigation efforts [5].

This study focuses on rainfall as a primary triggering factor and evaluates the impact of different rainfall conditions on geological hazard risk in the region, particularly in Huangshi Town, a key surveyed area. In line with national and provincial guidelines, the objective is to conduct a comprehensive risk assessment, improve early warning capabilities, guide rural development, and provide scientific support for disaster reduction [6]. In recent years, with the advancement of GIS technology, regional geological hazard assessment has rapidly developed. Models such as the Information Value (IV), Frequency Ratio, and Weight of Evidence have been widely applied, alongside machine learning methods such as SVMs, Decision Trees, Random Forests, and Logistic Regression [7].

Based on relatively complete evaluation data and evenly distributed hazard labels in Huangshi Town [8], this study adopts the IV model due to its simplicity, objectivity, and ability to avoid subjective bias [9]. To ensure reliability, the accuracy of susceptibility assessment results is verified using the Receiver Operating Characteristic (ROC) curve [10,11]. Furthermore, the study selects three representative rainfall conditions (24-h 100 mm, 24-h 250 mm, and 72-h 240 mm) to simulate risk scenarios [12]. A risk index method is used for quantitative evaluation, and slope unit zoning is applied to classify the area into different risk levels [13,14]. The results are expected to provide critical references for disaster mitigation planning and emergency response.

2. Study Area and Data

Huangshi Town is in the northeastern hilly region of Guangdong Province, under the jurisdiction of Longchuan County in Heyuan City. Geographical coordinates: East Longitude ${115}^{\circ }{10}^{\prime }{14}^{″}$–${115}^{\circ }{18}^{\prime }{07}^{″}$, North Latitude ${24}^{\circ }{15}^{\prime }{50}^{″}$–${24}^{\circ }{23}^{\prime }{23}^{″}$ (Figure 1). with a total area of 109.81 km², characterized by mountainous and hilly terrain, including approximately 10,700 mu of cultivated land and 140,000 mu of forest land.

This geomorphology forms the basis of the regional hydrogeological conditions, while the complex geological environment, featuring a well-developed weathered crust and dense fault systems, further shapes the hydrological processes. In terms of hydrology, surface water and groundwater in Huangshi Town achieve bidirectional recharge through natural infiltration. Rainwater infiltration serves as the primary source of groundwater recharge, while during dry seasons, groundwater discharges into surface water in the form of diffuse flow or spring discharge. In the wet season, the rising river water levels recharge the groundwater system [15]. This recharge mechanism benefits from the fracture networks in the bedrock mountainous areas and the hilly terrain, with shallow fractured water providing an important water source for the region. However, the area is significantly influenced by extreme weather events, particularly during the “Dragon Boat Water” period, when short-duration intense rainfall (with 24-h rainfall exceeding 250 mm) and prolonged rainfall events occur frequently. Coupled with the complex topography and hydrogeological conditions, extreme rainfall can rapidly activate the groundwater dynamic system, leading to abrupt changes in the moisture content of slope bodies and greatly increasing the risk of geological hazards. Currently, geological hazards in Huangshi Town are predominantly collapses and landslides, with landslide disasters highly concentrated in areas with slope heights of 5–20 m (accounting for 90.79%) and slope gradients greater than 50° (accounting for 95.03%) [16].

Moreover, human engineering activities, such as slope cutting for housing, construction of transportation infrastructure, and hillside cultivation, further exacerbate the potential for geological hazards [17]. However, current studies on geological hazard susceptibility in Huangshi Town mostly rely on static models and evaluate risks based solely on single rainfall indicators (e.g., annual average rainfall). These studies fail to adequately account for the coupled mechanisms between dynamic rainfall processes and geological hazards under extreme precipitation conditions and often neglect the combined effects of human activities and natural environmental factors on disaster occurrences [18].

This study, based on the actual conditions in Huangshi Town, integrates multi-source data including topography, geology, meteorology, and human activities to systematically assess geological hazard risks. It aims to fill the research gap in disaster prediction under extreme rainfall conditions and to provide theoretical support and practical references for geological hazard prevention and control in subtropical hilly regions.

3. Research Methods

3.1. Correlation of Evaluation Factors

Based on the disaster-forming mechanisms of landslides and collapses, this study selects ten factors—slope, aspect, slope shape, terrain relief, cover layer thickness, engineering geological lithology, vegetation coverage, land use type, slope structure, and cut-slope height [19]. From four dimensions: topography and geomorphology, geological background, surface cover, and hazard characteristics (Figure 2).

The high correlation between evaluation factors may affect the accuracy of the disaster-susceptibility assessment model for geological hazards, leading to factor data redundancy and subsequently reducing the model’s evaluation precision. Therefore, it is necessary to conduct a correlation analysis of the factors before establishing the assessment model [20]. Factor correlation refers to using statistical methods to assess the strength of relationships between different factors to identify redundant information and improve model accuracy. Advanced correlation analysis methods, such as Pearson, Spearman, and Cramér’s correlation coefficients, can more comprehensively reveal complex relationships between factors, thereby optimizing model performance and enhancing decision-making support.

In this study, the Pearson correlation coefficient was employed for the correlation analysis of evaluation factors, based on a comprehensive consideration of data characteristics, methodological suitability, and research objectives. The evaluation factors involved in the study (such as slope, terrain relief, and cover thickness) are primarily continuous variables, and the Pearson correlation coefficient can directly quantify such linear associations. Although the Spearman coefficient can measure nonlinear monotonic relationships, the data were preprocessed to approximate a normal distribution (e.g., the Pearson coefficient between slope and terrain relief is 0.48, indicating a moderate linear relationship), and introducing Spearman analysis would increase redundant calculations (due to data ranking) without providing additional useful information [21]. Moreover, categorical variables (such as land use type and engineering geological lithology) were incorporated into the continuous analysis framework through numerical encoding (e.g., agricultural land = 1), and their linear independence was validated by the Pearson coefficient (e.g., the coefficient between “land use type” and vegetation coverage is −0.25). Since Cramér’s V coefficient is specifically designed for pure categorical variable analysis, it is not compatible with the research objective (verifying linear independence), and applying mixed methods would compromise the consistency of model inputs [22].

The Pearson correlation coefficient measures the strength and direction of the linear relationship between two continuous variables. It has strong data compatibility and is minimally affected by the random distribution of the original variables [23]. This approach avoids redundant analysis while ensuring methodological simplicity and efficiency, delivering reliable and comparable results that comprehensively balance data characteristics and model requirements. The correlation coefficient is usually denoted by r. Given two random variables X and Y, the calculation formula for r is shown in Equation (1):

$$r={\displaystyle \frac{{\sum }_{i=1}^n(X_i-\overline{X})(Y_i-\overline{Y})}{\sqrt{({\sum }_{i=1}^n{(X_i-\overline{X})}^2)({\sum }_{i=1}^n(Y_i-{\overline{Y}}^2))}}}$$

(1)

In Equation (1), n represents the number of samples; $X_i$ and $Y_i$ are the observed values of variables X and Y at the i point, while $\overline{X}$ and $\overline{Y}$ represent the means of the X and Y samples, respectively.

The value of the correlation coefficient r ranges from −1 to 1, where a value of 1 indicates a perfect positive correlation, −1 indicates a perfect negative correlation, and 0 indicates no correlation. Typically, the Pearson correlation coefficient method focuses on the absolute value of the correlation coefficient to represent the strength of the relationship between variables, denoted as $\left|R\right|$.

3.2. Information Entropy Calculation

The influencing factors of geological hazards are numerous, and the degree of impact of different factors on hazards varies. It is necessary to use the information value method to determine the weights of each region, as shown in Equation (2):

$$I(Y,x_1,x_2,x_3\dots x_n)=ln{\displaystyle \frac{P(Y,x_1,x_2,x_3\dots x_n)}{P(Y)}}$$

(2)

Based on conditional probability operations, Equation (2) can be rewritten as Equation (3):

$$r={\displaystyle \frac{{\sum }_{i=1}^n(X_i-\overline{X})(Y_i-\overline{Y})}{\sqrt{({\sum }_{i=1}^n{(X_i-\overline{X})}^2)({\sum }_{i=1}^n(Y_i-{\overline{Y}}^2))}}}$$

(3)

Specifically, $I(Y,x_1,x_2,x_3\dots x_n)$ represents the information contribution of the combination of influencing factors $x_1,x_2,x_3\dots x_n$ to geological disaster Y; $P(Y,x_1,x_2,x_3\dots x_n)$ denotes the probability of geological disaster Y occurring under the combination of influencing factors $x_1,x_2,x_3\dots x_n$; $P(Y)$ represents the probability of occurrence of geological disaster Y; $I_{x_1}(Y,x_2)$ signifies the information contribution of evaluation metric $x_2$ to geological disaster Y under the condition of influencing factor $x_1$ [24].

This equation illustrates that the total information provided by the combination of influencing factors $I(Y,x_1,x_2,x_3\dots x_n)$ on the occurrence of geological hazards is equal to the total information contributed by $x_1$, the information contributed by $x_2$ under the condition of $x_1$, and so on, up to the sum of the information contributed by $x_n$ under the condition of the presence of $I(Y,x_1,x_2,x_3\dots x_{n-1})$ This indicates that the disaster-susceptibility assessment based on the information value method fully considers the impact and influence of the combined conditions of various contributing factors on the occurrence of geological hazards.

Taking this study as an example, the information value method based on Geographic Information System (GIS) technology is used to assess the disaster susceptibility of geological hazards, requiring a reliable standard model. The specific establishment method is as follows [25]:

(1) Calculate the information amount contributed by the evaluation factors using Equation (4):

$$I(x_i,D)=ln{\displaystyle \frac{P(x_i|D)}{P(x_i)}}$$

(4)

In practical scenarios, frequency is typically calculated using Equation (5): N represents the frequency of geological disasters in the analyzed region, S denotes the number of assessed units, $N_i$ is the number of disasters, and $S_i$ represents the number of evaluation factors [26].

$$I(x_i,D)=ln({\displaystyle \frac{N_i/N}{S_i/S}})$$

(5)

(2) The total information amount of geological disasters under the conditions of n-type influencing factors is calculated using Equation (6):

$$I_i=\sum _{i=1}^{n}I(x_i,D)=\sum _{i=1}^{n}ln({\displaystyle \frac{N_i/N}{S_i/S}})$$

(6)

(3) The total information amount is denoted as $\sum I_i$, serving as the evaluation criterion. An increase in this value indicates a higher probability of geological disaster occurrence in the evaluation unit. Conversely, a decrease suggests a lower probability of geological disaster occurrence. When the value is less than zero, it indicates that the probability of a geological disaster occurring in the evaluation unit is below the average level of the study area during the same period. If the value equals zero, it signifies that the probability matches the average level. When the value is greater than zero, it implies that the probability of geological disaster occurrence in the unit exceeds the average level of the study area during the same period.

(4) Based on the final ranking results, classification is performed, and each level is analyzed in depth to determine the different levels of disaster susceptibility for geological hazards.

3.3. Hazard Index Method

Geological hazard risk refers to the likelihood of a specific scale and type of geological hazard occurring in a particular area within a certain period under specific triggering factors [27]. The hazard assessment of geological disasters is conducted based on the disaster-susceptibility evaluation, considering three rainfall scenarios: 24-h rainfall of 100 mm, 24-h rainfall of 250 mm, and 72-h effective rainfall of 240 mm. Combined with the developmental characteristics of geological disasters, the hazard assessment classifies the geological disaster risk in Huangshi Town into four levels: extremely high-risk zones, high-risk zones, medium-risk zones, and low-risk zones [28].

The risk evaluation of geological hazards is conducted using the risk index method, as expressed in Equation (7):

$$H_i={\displaystyle \frac{Y_i}{Y_{max}{P}_i}}$$

(7)

where $H_i$ represents the risk index (hazard probability) of the i-th evaluation unit under a specific scenario; $Y_i$ is the susceptibility index of the i-th evaluation slope unit; $max$ is the maximum susceptibility index, with ${\mathit{Y}}_{\mathit{max}}=5$; and $P_i$ is the instability probability of the i-th evaluation unit within a given time period under a specific scenario. From Equation (7), it can be seen that the hazard index of an evaluation unit is closely related to its susceptibility index and instability probability. The susceptibility index of each slope unit and the maximum susceptibility index are derived from the geological disaster-susceptibility evaluation indices of the previously surveyed key areas. The instability probability of slope units is typically calculated using two methods: one is based on the extreme rainfall hypothesis, and the other is based on slope stability analysis. In order to assess the regional geological disaster hazard, this investigation uses the calculation method based on the extreme rainfall hypothesis. The specific calculation process is shown in Equation (8):

$$P_i={\displaystyle \frac{L}{L_{imax}}}$$

(8)

In Equation (8): $P_i$ represents the instability probability of the i-th evaluation unit under a specific scenario; L is the rainfall value under a specific scenario; and $L_{\mathrm{imax}}$ is the 24-h maximum rainfall value at the i-th evaluation unit.

This principle is based on the historical occurrence of geological disasters in the evaluation area. It assumes that the maximum rainfall within 24 h, $L_{\mathrm{imax}}$, has been the main factor triggering disasters since the monitoring records began. Under different rainfall conditions, the probability of instability can be expressed as the ratio of the three rainfall conditions to the 24-h maximum rainfall. When the ratio of rainfall conditions to the 24-h maximum rainfall exceeds 1, the value of $P_i$ is fixed at 1, indicating that as $P_i$ increases from 0 to 1, the slope stability gradually decreases. When $P_i=1$, the slope becomes unstable; thus, the maximum value of the instability probability $P_i$ is 1. After calculating the hazard index $H_i$ for each slope unit, the hazard level is classified into four categories according to the standard recommended geological disaster hazard classification table: extremely high-risk areas, high-risk areas, moderate-risk areas, and low-risk areas.

4. Establishment and Grading of Evaluation Indicators

4.1. Slope Unit Division

In this experiment, a combination of regular grids and slope units is used as the evaluation unit. The risk assessment employs a 10 m × 10 m regular grid as the initial evaluation unit. After standardizing the data of each evaluation factor to a uniform grid unit and calculating the disaster susceptibility, spatial statistical tools are used to summarize the evaluation results within the range of slope units and express them in the attribute table of the slope units [29].

First, areas with flat terrain and extremely low geological hazard potential were excluded to focus on regions with potential risks. For small slope units that were difficult to identify by the naked eye at a 1:2000 scale, adjacent units were merged to ensure the practical relevance of the division. In addition, slope units crossing multiple administrative villages (or communities) were subdivided based on administrative boundaries to facilitate management. Each slope unit was required to possess similar geological environmental conditions, disaster-inducing mechanisms, and disaster modes to improve the evaluation accuracy. Watershed divides or valleys served as the upper and lateral boundaries of the slope units, while the lower boundaries were defined by the extent of the potentially affected disaster-bearing bodies or river exteriors. Micro-topographic features and the connectivity between units were also taken into account to achieve scientifically reasonable boundary delineation. In regions that had experienced or were prone to gully-type debris flow disasters, relevant slope units were merged, and exclusive evaluation units were delineated according to debris flow evaluation methods. Furthermore, the scale of slope units was controlled according to practical conditions to ensure the evaluation precision and to facilitate risk management efforts by government authorities.

The division of slope units was specifically carried out in three steps. The first step involved preliminary division based on DEM data. Using the ArcGIS platform, hydrological analysis was conducted, including operations such as generating depressionless DEMs, extracting flow directions, calculating flow accumulation, generating river networks, and delineating catchment areas. Ridgelines and valley lines were extracted, and catchment and reverse catchment areas were integrated. Combined with terrain, lithology, vegetation coverage, and residential distribution characteristics, the preliminary slope units for Huangshi Town were divided, and the preliminary division map was generated (Figure 3) [30,31].

The second step is the refined division. With the aid of high-precision optical images and DEM digital elevation base maps, and by comprehensively considering the distribution characteristics of disaster-bearing bodies, the preliminarily divided slope units are optimized. The refinement and adjustment are mainly focused on the areas with threatened objects. Finally, 387 slope units are divided, among which there are 350 units in the key investigation area and 37 units in the general investigation area, and a refined division map of the slope units is drawn (Figure 4).

The last is the field correction link. During the ground comprehensive investigation stage, field investigations are carried out according to the finely divided slope units, and corrections are made in combination with the actual situations, such as the distribution of disaster-bearing bodies, topography and geomorphology, and slope structures. For slope units where disaster-bearing bodies are concentrated, have a large scope and obvious topographic changes, they are split; for units where disaster-bearing bodies are scattered, adjacent and have a small area, they are merged to reasonably control the number of disaster-bearing bodies within the unit, which is convenient for risk prevention and control. Supplementary division is carried out for areas where there are disaster-bearing bodies but lack corresponding slope units, and local boundary adjustments are made for slope units that do not completely cover the first row of houses at the foot of the slope to ensure that the division results are in line with the actual topography and the distribution of disaster-bearing bodies.

4.2. Selection and Correlation Analysis of Evaluation Factors

The analysis of the conditions, structures, and states of geological disaster formation shows that the occurrence of geological disasters is not determined by a single factor. Therefore, the assessment of geological disasters must comprehensively consider multiple influencing factors. Scientifically, comprehensively, and reasonably selecting evaluation indicators forms the foundation for disaster-susceptibility assessment [32]. Based on the Technical Requirements for Geological Hazard Risk Survey and Evaluation in Towns (Streets) of Guangdong Province (1:10,000) (Trial) and the actual geological environment and disaster types of Huangshi Town [33]. This study selected the evaluation factors as follows. Topographic factors such as slope, aspect, slope shape, and terrain undulation were extracted from Digital Elevation Model (DEM) data; engineering geological rock groups and cover layer thickness were obtained based on geological survey data; vegetation coverage was calculated through remote sensing image inversion; while land use type, slope structure, and cut-slope height were primarily determined through field investigation and surveying [34].

The influence mechanisms of each factor on slope stability are as follows: greater slope steepness increases the tangential component of gravity, significantly raising instability risk; aspect affects solar radiation and weathering, indirectly altering the physical and mechanical properties of rock and soil; different slope shapes cause stress variations—convex slopes are prone to tensile stress at the top, and concave slopes to shear stress at the base—both weakening stability; greater terrain undulation indicates stronger tectonic activity and erosion, leading to higher rock fragmentation and instability risk. In terms of geological background, thicker cover layers and weak lithologies or well-developed structural planes reduce shear strength and increase deformation and failure risks. Surface coverage conditions regulate hydrological processes and erosion resistance; higher vegetation coverage strengthens soil and reduces water content, lowering disaster risk, while human activities, reflected by different land use types, further destabilize slopes. Regarding hazard characteristics, weak structural planes within slopes tend to form potential sliding surfaces, significantly undermining stability; increased cut-slope height weakens toe support, causing stress imbalance, and is a key factor in the frequent geological disasters in Huangshi Town [35].

To perform the correlation analysis of the geological disaster-susceptibility evaluation factors for Huangshi Town, the Pearson correlation coefficient matrix between the susceptibility evaluation factors is obtained (

Table 1. Pearson phase relation table of geological disaster-susceptibility evaluation factors.

Factor	Slope	Aspect	SM	TR	EGRG	CLT	VC	LUT	SS	SH
Slope	1.00	0.01	0.03	0.48	0.11	−0.04	0.39	−0.09	0.05	0.21
Aspect	0.01	1.00	0.00	0.02	0.06	−0.01	0.00	−0.02	0.08	0.09
SM	0.03	0.00	1.00	0.03	0.00	−0.01	0.00	−0.03	0.10	0.15
TR	0.48	0.02	0.03	1.00	0.11	−0.05	0.43	−0.10	0.04	0.23
EGRG	0.11	0.06	0.00	0.11	1.00	0.08	0.22	−0.01	0.46	0.09
CLT	−0.04	−0.01	−0.01	−0.05	0.08	1.00	−0.07	0.08	−0.06	−0.13
VC	0.39	0.00	0.00	0.43	0.22	−0.07	1.00	−0.25	0.16	0.18
LUT	−0.09	−0.02	−0.03	−0.10	−0.01	0.08	−0.25	1.00	−0.10	−0.11
SS	0.05	0.08	0.10	0.04	0.46	−0.06	0.16	−0.10	1.00	−0.02
SH	0.21	0.09	0.15	0.23	0.09	−0.13	0.18	−0.11	−0.02	1.00

The abbreviations in the table are as follows: SM represents Slope Morphology, TR represents Topographic Relief, EGRG represents Engineering Geological Rock Group, CLT represents Cover Layer Thickness, VC represents Vegetation Coverage, LUT represents Land Use Type, SS represents Slope Structure, and SH represents Slope Height.

).

According to the correlation coefficient classification table, a coefficient smaller than 0.5 indicates a low correlation between factors, meaning no collinearity exists, and the factors can be used as predictive variables. The correlation coefficients between the 10 evaluation factors are all smaller than 0.5, indicating that the evaluation factors are mutually independent and can all be used for disaster-susceptibility assessment [36].

5. Results

5.1. Information Content Calculation

Referring to the Technical Requirements for Geological Hazard Risk Survey and Evaluation of Townships (Streets) in Guangdong Province (1:10,000) (Trial), the classification criteria of evaluation factors in this study are based on a combination of multidisciplinary theories and empirical data. Terrain slope is classified using the natural breaks method and engineering experience thresholds, reflecting the nonlinear influence of slope gradient on landslide occurrence. Terrain aspect follows the standard geographic azimuth classification, incorporating the impact of water and heat differences between sunny and shady slopes on slope stability. Slope morphology is classified based on the theory of terrain curvature, quantifying the control of different curvatures on slope stress distribution. Terrain relief is related to structural activity intensity through elevation range and classified using the natural breaks method. Engineering geological rock groups are classified based on differences in lithological strength. The thickness of the cover layer is categorized according to geo-mechanical principles and borehole data, considering the threshold effects of thickness on slope stability. Vegetation coverage is quantified using NDVI remote sensing to reflect the slope protection effects of vegetation. Land use types are classified in relation to the intensity of human activity disturbance. Slope structure is categorized according to geological survey standards, based on the relationship between slope direction and stratum dip. The height of artificial cut slopes is evaluated using free-face failure models and validated by aerial survey data [37,38]. The final quantified results are presented in

Table 2. Geological disaster-susceptibility evaluation factor information.

Factor Indicator	Status Indicator	Information Quantity
Topographic Slope	0–10°	−0.4030
	10–25°	0.2383
	25–40°	−0.0958
	>40°	0.4657
Topographic Aspect	Horizontal (−1)	−1.2071
	North (337.5°–22.5°)	−0.5177
	Northeast (22.5°–67.5°)	−0.3595
	East (67.5°–112.5°)	−0.1046
	Southeast (112.5°–157.5°)	0.3230
	South (157.5°–202.5°)	0.3546
	Southwest (202.5°–247.5°)	0.2904
	West (247.5°–292.5°)	−0.1036
	Northwest (292.5°–337.5°)	−0.1003
Slope Morphology	Concave Slope	0.4197
	Zigzag Slope, Straight Slope°	−0.7646
	Convex Slope	−0.1037
Topographic Relief	0–10 m	−1.4132
	10–20 m	0.3037
	20–30 m	0.3521
	>30 m	−0.7748
Engineering Geological Rock Group	Sandy soils, clayey soils, and other sedimentary soil bodies	0.4648
	Layered relatively soft metamorphic rock groups	−0.0725
	Layered relatively soft red bed rock groups	−0.5125
	Blocky, relatively hard to hard intrusive rock groups	0.0269
Vegetation Cover	<0.1	0.6956
	0.1–0.3	2.4445
	0.3–0.5	−0.0047
	>0.5	−2.9056
Cover Layer Thickness	<1 m	0.1161
	1–3 m	−0.2501
	3–6 m	0.6221
	6–9 m	0.5256
	9–12 m	0.2853
	>12 m	0.3925
Land Use Type	Agricultural Land	−0.8750
	Construction Land	2.4287
	Forest Land	−0.4141
	Water Bodies	−1.5377
	Grassland	0.1356
	Barren Land	2.3691
Slope Structure	Transverse Slope	−0.5750
	Oblique Slope	−0.1173
	Blocky Rock Slope	−0.0202
	Downhill Slope	0.8197
	Uphill Slope	−0.1511
Slope Height	<3 m	−2.2672
	3–6 m	1.2183
	6–12 m	2.2297
	12–25 m	2.4369
	>25 m	2.46917

.

The evaluation factors are classified according to the aforementioned quantification standards of the evaluation indicators. Using the information value model method, the information value of each evaluation factor and its corresponding classification is calculated. The resulting information values are assigned to the classified indicator states, yielding the information value table for the slope-type geological disaster-susceptibility evaluation factors in Huangshi Town (Table 2).

Based on the weights of the disaster-susceptibility evaluation factors mentioned above, these weights are assigned to the classified factor raster map. Using spatial overlay analysis tools, the total information value of the raster cells at each spatial location in Huangshi Town is calculated, generating the disaster-susceptibility evaluation-index map for Huangshi Town (Figure 5).

The information value model is a spatial prediction model, and the landslide susceptibility zoning results it produces are also predictive. The quality of susceptibility evaluation directly determines the accuracy of subsequent hazard and risk assessments. Therefore, in order to ensure the reliability of risk assessment results, it is necessary to evaluate the accuracy of the susceptibility analysis.

The Receiver Operating Characteristic (ROC) curve is employed to validate the model’s predictive performance. The predicted susceptibility values are divided into several classes, and within each class, the cumulative area of susceptible zones and the actual landslide areas are calculated. In the context of landslide susceptibility assessment, the true positive rate (TPR) on the y-axis represents the proportion of actual landslide areas, while the false positive rate (FPR) on the x-axis indicates the proportion of predicted susceptible areas.

The area under the ROC curve (AUC) is commonly used as an objective and quantitative metric to evaluate model prediction performance. The AUC value ranges from 0 to 1, with values closer to 1 indicating higher predictive accuracy.

In this study, the accuracy of the susceptibility evaluation results was tested using the ROC curve. The information values of the susceptibility index map were statistically analyzed, and the ROC curve was generated using Python 3 programming, with the map area ratio plotted on the x-axis and the landslide area ratio on the y-axis (Figure 6). The resulting AUC value of 0.8739 demonstrates that the susceptibility results derived from the information value model are of high accuracy, thus providing data quality assurance for subsequent hazard and risk assessments of geological disasters in Huangshi Town.

5.2. Susceptibility Mapping

After obtaining the disaster-susceptibility evaluation-index raster map validated for accuracy, the results of the geological disaster-susceptibility evaluation for Huangshi Town are displayed in the form of slope units, according to the division requirements of the evaluation units. Based on the classification of slope units, the average susceptibility index within each slope unit is calculated using the ArcGIS spatial statistics function and assigned to the corresponding attribute table of the slope units. Using the natural breaks method combined with field survey data, the susceptibility index of the slope units is classified into three levels: high susceptibility, medium susceptibility, and low susceptibility. Subsequently, adjacent slope units with the same susceptibility level are merged, zoned, and numbered, generating the disaster-susceptibility zoning map for Huangshi Town (Figure 7).

5.3. Hazard Assessment Under Different Rainfall Conditions

In the existing risk assessment framework, this study simplifies the rainfall-triggering mechanism into an immediate response relationship based on statistical analysis of historical landslide events and rainfall data. Based on the likelihood of geological hazards being triggered by specific factors in certain periods within the survey area, the hazard levels for different segments were determined, and the boundaries of hazardous zones were calculated. The hazard zones were then classified into four levels: extremely high hazard zone, high hazard zone, moderate hazard zone, and low hazard zone. In Huangshi Town, the primary geological hazards are slope-type disasters. Referring to the “Geological Hazard Risk Survey and Evaluation Technical Requirements for Townships (Streets) in Guangdong Province (1:10,000) (Trial)”, which specifies the triggering factors for slope-type geological hazards, and combining field surveys with expert experience, rainfall was chosen as the triggering factor for geological hazards in Huangshi Town. A triggering factor refers to the internal and external dynamic forces that cause the geological environment system to evolve towards a geological disaster.

According to the calculation method for the instability probability of geological hazards in the previous evaluation approach, spatial distribution maps of the instability probability for geological hazards were generated under three rainfall conditions: 24-h 100 mm rainfall, 24-h 250 mm rainfall, and 72-h 240 mm effective rainfall (Figure 8).

5.4. Risk Assessment Under Different Rainfall Conditions

Based on the slope unit classification, the average hazard index for each slope unit was calculated using ArcGIS spatial statistics. This value was then assigned to the corresponding slope unit attribute table. Using the natural break method combined with field survey data, the hazard index for each slope unit was divided into four levels: extremely high hazard zone, high hazard zone, moderate hazard zone, and low hazard zone. The slope units with the same hazard level were then merged, subdivided, and numbered, generating the hazard zoning maps for Huangshi Town under different rainfall conditions (Figure 9).

(1) Hazard Assessment Under 24-h 100 mm Rainfall Condition: Using the geological hazard risk assessment method for the key investigation area, a geological hazard risk zoning map under the 24-h 100 mm rainfall condition was generated for the key investigation area. The key investigation area includes 64 slope units, with a total area of 1.596 km², and a total of 76 geological hazard risk points. Among these, the high-risk area contains 22 slope units, covering an area of 0.556 km², which accounts for 34.86% of the total area of slope units in the key investigation area. There are 38 geological hazard risk points in this area. The medium-risk area contains 33 slope units, covering an area of 0.828 km², which accounts for 51.88% of the total area of slope units. There are 38 geological hazard risk points in this area. The low-risk area contains 9 slope units, covering an area of 0.212 km², which accounts for 13.27% of the total area of slope units. There are no geological hazard risk points in this area.

(2) Hazard Assessment Under 24-h 250 mm Rainfall Condition: For the key survey area containing 64 slope units and a total area of 1.596 km², with 76 geological hazard risk points. Extremely High Hazard Zone: 33 slope units, covering 0.884 km², accounting for 55.36% of the total area. There are 49 geological hazard risk points. High Hazard Zone: 28 slope units, covering 0.653 km², accounting for 40.90% of the total area. There are 26 geological hazard risk points. Moderate Hazard Zone: 2 slope units, covering 0.034 km², accounting for 2.14% of the total area. There is 1 geological hazard risk point. Low Hazard Zone: 1 slope unit, covering 0.026 km², accounting for 1.61% of the total area. No geological hazard risk points are identified.

(3) Hazard Assessment Under 72-h 240 mm Effective Rainfall Condition: For the key survey area containing 64 slope units and a total area of 1.596 km², with 76 geological hazard risk points. Extremely High Hazard Zone: 33 slope units, covering 0.884 km², accounting for 55.36% of the total area. There are 49 geological hazard risk points. High Hazard Zone: 28 slope units, covering 0.653 km², accounting for 40.90% of the total area. There are 26 geological hazard risk points. Moderate Hazard Zone: 2 slope units, covering 0.034 km², accounting for 2.14% of the total area. There is 1 geological hazard risk point. Low Hazard Zone: 1 slope unit, covering 0.026 km², accounting for 1.61% of the total area. No geological hazard risk points are identified. These evaluations provide a clear risk classification based on different rainfall scenarios, supporting the effective mitigation and planning for geological hazard management in Huangshi Town.

The evaluation results show that there are a total of 349 residential slope units in the town, of which 37 are high-risk, 174 are medium-risk, and 138 are low-risk. Based on the geological hazard risk assessment, a geological hazard prevention zoning plan was carried out for Huangshi Town, dividing the entire area into 4 key prevention zones with a total area of 18.55 km², 6 secondary key prevention zones with a total area of 25.78 km², and 3 general prevention zones with a total area of 65.48 km². Additionally, based on different rainfall conditions, prevention zoning work was conducted for the key investigation areas. In the key investigation areas, 33 key prevention zones were delineated, covering an area of 0.884 km², 176 secondary key prevention slope units were identified, covering an area of 4.917 km², and 146 general prevention slope units were defined.

The analysis results under different rainfall scenarios indicate a clear trend of risk concentration with increasing rainfall intensity. Under the 24-h 100 mm rainfall condition, the high-risk zone includes 22 slope units, covering an area of 0.556 km², which accounts for 34.86% of the total area of the key investigation area; the medium- and low-risk zones together account for 65.14% of the area. In contrast, under the 24-h 250 mm and 72-h 240 mm rainfall conditions, the extremely high and high-risk zones combined include 61 slope units, covering a total area of 1.537 km², representing over 96% of the total area. This demonstrates that as rainfall intensity increases, slope units originally classified as moderate-risk are extensively transformed into high-risk zones. At the same time, the number of geological hazard risk points increases significantly under high rainfall scenarios, reflecting a heightened level of hazard activity. Spatially, high-risk zones are significantly concentrated in the southeastern and southwestern parts of Huangshi Town, indicating a localized sensitivity to extreme rainfall. This trend highlights the importance of dynamic zoning and emergency planning tailored to different rainfall scenarios in geological hazard prevention and mitigation efforts.

6. Discussion

In the detailed geological hazard survey of Huangshi Town, the project innovatively integrated remote sensing and UAV technologies—including orthophotography, oblique photogrammetry, and airborne LiDAR—to establish a multi-dimensional data acquisition system combining “satellite–air–ground” observations. High-resolution orthophotos, three-dimensional oblique models, and laser measurements were implemented in key areas to generate precise real-world models, providing rich and accurate data support for hazard research. By integrating remote sensing interpretation with ground surveys, the project accurately identified risk locations, monitored slope changes, delineated threatened areas, and achieved a comprehensive and systematic investigation of geological hazards. This approach significantly enhanced early warning and prevention capabilities, effectively overcoming the limitations of traditional ground surveys, such as low efficiency, limited coverage, and insufficient precision of single remote sensing techniques, thereby offering more reliable foundations for hazard mitigation.

However, the integrated approach also faces practical challenges. The high costs of equipment acquisition and maintenance impose significant financial burdens, limiting its promotion in underfunded regions. Data processing and analysis heavily rely on skilled professionals and advanced software platforms, raising the technical threshold and posing challenges for talent cultivation and technology dissemination. Additionally, UAV operations are notably constrained by weather conditions, with adverse weather severely impacting the timeliness and completeness of data collection.

Previous geological hazard risk assessments generally considered only single rainfall conditions, neglecting the variations in hazard probability and risk under different rainfall scenarios [39]. Furthermore, the long-term effects of rainfall on the physical and mechanical properties of rock and soil masses were often overlooked, resulting in considerable assessment inaccuracies [40,41]. This study comprehensively analyzed three rainfall scenarios: 24-h rainfall of 100 mm, 24-h rainfall of 250 mm, and 72-h effective rainfall of 240 mm, thoroughly accounting for the effects of prolonged rainfall. A combined evaluation unit approach of regular grids and slope units was adopted, selecting ten evaluation factors and deeply analyzing their correlations to achieve more accurate and reliable results. By calculating slope instability probabilities under different rainfall conditions, the study provided a more comprehensive reflection of geological hazard risks across varied scenarios, offering more practical guidance for disaster prevention and mitigation efforts. Through the risk assessment of geological hazards under multi-scenario rainfall conditions in Huangshi Town, Longchuan County, Guangdong Province, the study elucidated the impact mechanisms of prolonged rainfall on hazard risks and proposed corresponding improvement measures. The results indicate that evaluating disaster susceptibility with rainfall as a triggering factor better reflects the overall characteristics of geological hazard development in the study area.

Future research could focus on several key directions: first, exploring the mechanisms of disaster occurrence under complex geological conditions to improve and refine evaluation models, thereby enhancing prediction accuracy under extreme scenarios; second, advancing multi-source data fusion technologies to further boost the efficiency and precision of data acquisition and analysis; and third, leveraging emerging technologies such as artificial intelligence and big data to achieve real-time dynamic monitoring and early warning of geological hazards, providing more effective technical support for hazard prevention and mitigation.

By integrating new technologies with traditional methods, this study thoroughly identified the development characteristics of geological hazards in Huangshi Town, including types, scales, and spatial distribution patterns. It clarified the internal disaster-forming geological conditions, encompassing geomorphology, geological structures, and rock-soil types, as well as external triggering factors such as rainfall. Using scientific methodologies, geological hazard risk zones and prevention zones were delineated, offering critical references for regional planning and disaster control. Additionally, a geological hazard risk survey and assessment information database was established, integrating extensive data resources and laying a solid foundation for the subsequent construction of municipal, provincial, and national-level risk survey databases, thereby strongly promoting the advancement of geological hazard prevention and control efforts.