Optimization of Emergency Stockpiles Site Selection for Major Disasters in the Qinghai Plateau, China

Abstract

The Qinghai Plateau has a complex geographical environment and vast amounts of land with a sparse population, dispersed settlements, and a low traffic density. In the face of major disasters, the rational layout of emergency material reserve warehouses is crucial for reducing disaster losses, ensuring regional stability, and quickly restoring production and life. This paper starts by considering the rationality and timeliness of the location selection of provincial emergency material reserve warehouses, considering the distance costs of emergency material transportation on the Qinghai Plateau. By using a traffic accessibility analysis model combined with a location–allocation model and an L-A maximum coverage model, this study optimizes the location selection of emergency material reserve warehouses on the Qinghai Plateau. The research results show the following: (1) On the basis of the existing Golmud Depot and Chengxi Depot in Qinghai Province, it is necessary to add four more depots, i.e., the Yushu Depot, Gande Depot, Ping’an Depot, and Tongde Depot, to achieve the timely and efficient supply of emergency materials. (2) After the optimization, the layout of the six provincial emergency material reserve warehouses can achieve full coverage of Qinghai Province within 8 h in the event of major disasters, increasing the coverage by 20% compared to the original layout; the new plan allows for emergency material transportation to cover 87% of Qinghai Province within 4 h, an increase of 28% compared to before. (3) The optimized location selection plan for emergency material reserve warehouses saves 139 min of time costs, and the transportation efficiency is increased by 46% compared to the previous plan. The optimized location selection plan for emergency material reserve warehouses is instructive for the construction of emergency material reserve warehouses on the Qinghai–Tibet Plateau.

1. Introduction

Major disasters, such as earthquakes, landslides, and floods, are characterized by their suddenness, uncertainty, and destructiveness. These characteristics frequently cause severe disruptions to infrastructure, supply chains, and essential services, thereby increasing the vulnerability of affected populations. For example, on 23 November 1980, the Irpinia–Basilicata region in southern Italy was struck by a devastating earthquake. The 6.7-magnitude quake caused extensive destruction, resulting in over 2700 fatalities and injuries. The earthquake affected 687 municipalities, with 37 being completely destroyed, 314 suffering severe damage, and 336 experiencing varying degrees of destruction. The scale of destruction highlighted the vulnerability of communities and the necessity for timely rescue and emergency material support [1]. Another example is the landslide that occurred on 22 March 2014, in Oso, WA, USA. This disaster resulted in the instant death of 43 people. The devastation covered an area of 200 acres, with a runout distance exceeding 6000 feet and an elevation drop of approximately 600 feet. Over 20 homes were destroyed in an instant [2]. In particular, the basic survival and living conditions of affected individuals may not be guaranteed due to the insufficient supply of critical materials, such as medicines and emergency supplies. This can directly or indirectly lead to casualties. For example, in February 2021, a series of winter storms caused severe energy infrastructure failures in Texas, leading to shortages of water, food, and heat for several days. The lack of timely emergency supply distribution made it difficult for 75% of the affected residents to obtain food, and vulnerable groups like children with epilepsy (CWE) were particularly at risk of life-threatening situations. This underscores the need to optimize the location of emergency relief supply depots to enhance disaster preparedness and improve the efficiency of emergency material delivery, ensuring that emergency supplies can be accessed promptly during sudden natural disasters [3].

The location and allocation of emergency supplies are an important part of emergency rescue work [4], and the timely and efficient supply of emergency materials is key to reducing disaster losses [5]. An emergency material stockpile is the infrastructure of the emergency response system, which can provide basic life protection for the emergency relocation of people and for those who are in urgent need of immediate emergency life assistance [6]. Local emergency material reserve depots emphasize the timeliness of emergency responses, and they have strict requirements for the arrival time of disaster relief materials [7]. Provincial emergency material reserve depot materials, characterized by large storage volumes and a high emergency value, play a pivotal role in coping with the occurrence of serious and large-scale disasters. For example, after the 2023 Jishishan earthquake, the Qinghai Provincial Emergency Management Department urgently organized troops, volunteer teams, and an agreement loading team, and more than 240 people and more than 200 transport vehicles (including loading and unloading vehicles) arrived at the emergency materials reserve depot for the emergency shipment of materials to provide disaster relief, the resettlement of disaster victims, on-site staff equipment, and the command center of the emergency material needs in the disaster area. Therefore, the accessibility efficiency of the vehicles and the accessibility of the transportation of materials to the disaster area play an extremely important role in the location of an emergency stockpile [8].

The function of an emergency material reserve depot determines its non-economic attributes, and it has fixed requirements for spatial location, coverage, and other factors, so there are more factors affecting the decision of site selection. At present, most studies on the siting of emergency material reserve depots start from the perspective of coverage and multi-objective demand, considering the coverage capacity of the reserve depots and multi-objective factors such as efficiency, fairness, and satisfaction of distribution, and they solve the problem of the siting and distribution of emergency materials by constructing a coverage model [9]. Vatsa et al. established a multi-stage maximum coverage siting mixed-integer planning model based on the uncertainty of demand. They used the max–min robust optimization method to solve the problem of siting emergency supplies reserves [10]. Vicencio-Medina et al. presents a variant of the Maximum Covering Location Problem, considering multiple accessibility indicators and optimizing the generalized coverage function [11]. Li et al. proposed a multi-objective optimization framework for the EMS facility layout, integrating multi-source data to support urban planning and public safety risk management [12]. Hu et al. proposed a multi-period emergency facility location-routing problem, considering uncertainties in demand and travel time [13]. Rahman et al. developed models to optimize the spatial distribution of emergency evacuation centers (EECs) in the northeastern Sylhet region of Bangladesh. They used location-allocation models (LAMs) in GIS, including the Maximum Covering Location Problem (MCLP), to enhance flood emergency planning and minimize disaster risks [14]. Some scholars have studied the siting problem from the perspective of multi-objective demand; Zhuo et al. emphasized the importance of considering multiple planning objectives and integrating the planning of emergency warehouse locations with the distribution volumes of emergency materials. They proposed a multi-objective emergency material planning model aimed at maximizing the efficiency and fairness of emergency material distribution while deploying strategic emergency warehouses and allocating each potential demand point to the nearest deployed strategic emergency warehouse [15]. Praneetpholkrang et al. proposed a multi-objective optimization model for determining the location-allocation of shelters in humanitarian relief logistics. The model includes three objective functions: minimizing cost, minimizing allocation time, and minimizing the number of shelters, aiming to improve efficiency and effectiveness [16]. Ershadi et al. proposed a multi-objective optimization model for logistics planning during the crisis response phase which includes minimizing costs, maximizing coverage, and improving response speed. This model helps decision-makers make more effective logistics planning decisions in emergency situations [17]. Liu et al. proposed a multi-objective decision-based optimization method for emergency plans, aiming to improve the timeliness of emergency response and the efficiency of resource allocation [18]; and Wang et al. examined the emergency logistics network in disaster-prone areas, addressing demand and supply uncertainties. They proposed a robust location-allocation model to minimize costs and delivery time while enhancing network reliability [19].

In summary, most scholars’ research on site selection focuses on the impact of factors such as coverage, cost, fairness, efficiency, and demand satisfaction on site selection from the perspective of the needs of the disaster area, and the research results focus on the analysis of the methodology of site selection for emergency reserve bases; there are fewer research studies that combine the actual situation of the region and the real needs, and there is a lack of comprehensive analysis of a variety of influencing factors in the modeling [20]. A region needs to have how many emergency supplies reserve base, emergency supplies reserve base location, not only related to population size of the region, the degree of regional influence, the degree of dependence on the facilities of the socio-economic environment factors related to [21], but also to the region’s natural geography, reserve base transportation accessibility affect the efficiency of the emergency supplies shipment out of the warehouse. The location and layout of the emergency material reserve is the key link, and the basic problem of emergency management and resource allocation [22], how to determine a reasonable number of material reserves, and whether the large-scale construction of material reserves is economical are questions which are worth thinking about [23]. Comprehensive site selection considers the maximum coverage of the emergency material reserve depot, the efficiency of emergency material out of the depot, the timeliness of transportation to the point of demand, and the minimum number of built depots and other issues. This paper combines the natural geographic characteristics of the Qinghai Plateau; the spatial distribution of the disaster-bearing body; and the transportation status quo of the provincial emergency supplies reserve depot, which mainly serves for major disasters, large emergency supplies reserves, the high value of supplies, and the reserve of scarce supplies, with the goal of “the smallest cost, the shortest time, the most extensive coverage, and in line with the actual”, and combines the transportation accessibility, relief zoning, and the maximum coverage of emergency supplies. With the objectives of “minimum cost, shortest time, maximum coverage and practicality”, we analyze the location of the Qinghai Plateau Emergency Reserve Depot through a dual-objective model. This model integrates the accessibility model of the transportation road network, the location-configuration maximum coverage model, and the P-median siting model. We also propose optimization plans for the existing layout.

2. Study Areas

The Qinghai Plateau is located within the territory of Qinghai Province in the northeast of the Qinghai Tibet Plateau, with a total area of approximately 722,300 km² and an average elevation of more than 3000 m above sea level, with more than 4/5 of the area being a plateau. The distribution of disaster-bearing bodies in the plateau region is uneven, and the population has the characteristics of small aggregations and large dispersion, with long transportation routes, low road network density, and scattered settlements. By the end of 2023, the province had a resident population of 5.92 million, of which 2.943 million were ethnic minorities, accounting for 49.47% of the resident population. The population was mainly concentrated in the eastern region, which accounted for about 70% of the total population of the study area. Qinghai Province comprises 45 counties and districts in its administrative region, and the road traffic is dense in the east and sparse in the west. Road transport is the primary mode of transportation [24] (Figure 1). Due to the unique disaster-conceiving environment, which makes the plateau disasters a frequent issue, the risk of major disasters is high, and the amount of hazards has grown since 2010. Qinghai Plateau was hit by the M7.1 Yushu Earthquake on 14 April 2010 [25]; the M6.2 Zaduo County Earthquake on 17 October 2016, the M7.4 Maduo County Earthquake on 22 May 2021, and the M6.0 Delingha City Earthquake on 26 March 2022 [26]. Additionally, the debris flow disaster in Datong County on 17 August 2022 [27] is believed to have caused serious casualties and property losses. At present, there are only two emergency supply depots in Qinghai Plateau (the Golmud Depot and the Chengxi Depot), which cannot meet the requirement of full coverage of emergency supplies within eight hours in the event of a major disaster.

3. Materials and Methods

3.1. Data Sources

The data used in this paper are divided into five parts:

①

The first part of the spatial data of administrative divisions of Qinghai Province is taken from the map of Qinghai Province with 1:3.5 million review number Qing S(2024)015, downloaded from the website of Standard Map Service of the Ministry of Natural Resources (https://qinghai.tianditu.gov.cn/qhbzmap (accessed on 7 November 2024));

②

The second part of the data of the transportation road network is from the Open Street Map road dataset (2024) (https://www.openstreetmap.org (accessed on 1 December 2024));

③

The third part of the digital elevation data is from the SRTM global 30 m resolution DEM data (https://srtm.csi.cgiar.org/srtmdata (accessed on 12 October 2024));

④

The fourth part of the data on existing provincial emergency material stockpiles in Qinghai Province is from the data statistics of the Qinghai Provincial Department of Emergency Management;

⑤

The fifth part of the population data is from the LandScan population dataset (https://landscan.ornl.gov (accessed on 2 December 2024)).

3.2. Assumptions

Assumption 1.

The categories and quantities of materials in the reserve are sufficient to meet the requirements for the categories and quantities of materials at the point of need.

Assumption 2.

It is assumed that transportation roads are accessible and have not suffered damage during the disaster.

Assumption 3.

Due to the special location of Tanggula Mountain Township, this paper analyzes it as an equal status with 45 counties and municipalities.

Assumption 4.

The mobilization of materials relies mainly on a flexible and convenient road transport system, taking into account the reality of high transport volumes and the large number of vehicles being mobilized and disregarding the construction of low-grade township and rural roads with weak road infrastructures.

3.3. Research Methods

Based on the road vector data, the road level is divided into four categories: level 1 (highways), level 2 (national highways), level 3 (provincial highways), and level 4 (some county roads). Using the Network Analyst interpolation method in ArcMap 10.8.1 software, we analyze traffic accessibility to identify potential sites for provincial and above-level emergency stockpile depots. In this analysis, the demand points (county and district governmental station) are set as destination points, while the provincial emergency stockpile depot site candidates (screened through the traffic accessibility analysis) are set as starting points. Specifically, we use the Network Analyst tool in ArcMap software to create a new location-allocation model, focusing on maximum coverage and integrating it with the P-median model to calculate and solve the optimal locations.

3.3.1. Transportation Accessibility Analysis

Accessibility can be described as how easy it is to get from one place to another [28]. Roads are the foundation of transportation trips, and the better the road accessibility, the more convenient the transportation becomes. Transportation accessibility is a key factor in socio-economic development [29]. Therefore, accessibility is an effective indicator for evaluating the efficiency of road transportation networks, reflecting the ease of traveling from one place to another [30]. Emergency stockpiles are ideally located in economically developed areas with convenient transportation and advanced logistics [31]. Transportation accessibility not only affects the efficiency of deliveries of emergency supplies but also directly impacts the efficiency of shipments of emergency supplies out of the warehouse. In emergency situations, time is life. Therefore, site selection must take into account the transportation accessibility between the reserve depot and the demand area. The layout of emergency supplies reserve bases is a key link and fundamental issue in macro emergency management and resource allocation [32]. The construction of the emergency material reserve system should take into account the actual transportation in the region, ensuring rapid mobilization and strong guarantees [33]. After receiving the mobilization instruction, except for remote areas, road interruptions, and other special circumstances, it should be ensured that the first batch of emergency materials will be transported to the disaster area within 10 h at the provincial level [34]. Since the demand points in this study are the administrative centers of counties and municipalities, it takes time for emergency supplies to be distributed from these points to the actual disaster areas. Therefore, it is stipulated that the areas where emergency supplies can arrive within 8 h are designated as “basic relief areas” and the areas where supplies can arrive within 4 h are designated as “easy relief areas”. After a disaster occurs, the more “easy relief areas” that are covered by the emergency supplies stockpile, the higher the efficiency of the rescue operations in the short period following the disaster [35]. At present, there are only two emergency material stockpiles in Qinghai Plateau (Golmud Depot and Chengxi Depot). The current emergency material stockpile covers 37 “basic relief areas” with a coverage rate of 80% and 27 “easy relief areas” with a coverage rate of 59%. However, there are still 9 demand points that are not within the 8 h delivery range of the “easy relief areas”. Additionally, there are nine demand points that are not covered by the “basic relief areas” within the eight-hour delivery range (Figure 2). This situation fails to meet the requirement that emergency supplies be delivered within eight hours in the event of a major disaster.

This paper adopts the spatial analysis method and selects four levels of roads in Qinghai Plateau (Level 1 (highways), Level 2 (national highways), Level 3 (provincial highways), and Level 4 (some county roads)) and calculates the required time on the basis of the lengths of the roads at all levels and design speeds, taking into full consideration the limitation of the high undulation and the long climbing (descending) gradient of the roads on the Qinghai Plateau on the speed of the vehicles, as well as the physiological reaction of the drivers caused by the high elevation and the low oxygen content. When calculating the accessibility, the transportation speed was fine-tuned on the basis of the 2014 Technical Standards for Highway Engineering (JTG B01-2014) [36] (

Table 1. Road classification.

Road Grade	Maximum Speed (Speed/(km/h))
Primary roads	100
Secondary roads	80
Tertiary roads	60
Class IV road	40

).

Based on the road transportation network dataset, Network Analyst was used to create a new OD cost matrix. The impedance was set to time (in minutes) to import time attributes, generate nodes, and construct an OD cost matrix model. This model measures the accessibility of the 46 counties and districts (demand points in this paper), using their locations as both origin and destination points. It reflects the degree of accessibility between any demand points in the study area. Using the shortest travel time as impedance, the transportation accessibility between demand points in the study area was analyzed. This analysis provides basic theoretical and data support for the selection of alternative sites for the emergency stockpile.

The specific formula [37] is as follows:

$${{\displaystyle A}}_i={\displaystyle \sum _{j=1}^n{{\displaystyle T}}_{ij}}/n$$

(1)

where i,j is the demand point in the region and the provincial emergency stockpile site; T_ij is the shortest passage time for i to reach the demand point j through the highway transportation network in the region; n is the number of demand points; and ${{\displaystyle A}}_i$ is the average passage time for the provincial emergency stockpile site i. The smaller the value, the better the accessibility of the provincial emergency stockpile site, i.e., the more convenient and quicker it is for the site to reach other demand points in the network, and vice versa, the worse it is. The smaller the value, the better the accessibility of the site of the provincial emergency stockpile. In other words, the more convenient and faster it is for the provincial emergency stockpile to reach other demand points in the network, the better the accessibility. Conversely, a larger value indicates worse accessibility.

3.3.2. Optimizing the Site Selection Model

The siting decision of emergency material reserve depot is directly related to the response speed of emergency material supply and the effectiveness of regional emergency material reserve system construction [38]. The classic siting models in the field of facility location include the P-center model, maximum coverage model, and P-median model. The central theory of these models focuses on the distance between the site of the provincial emergency stockpile and the demand point. The P-center model mainly emphasizes fairness: once the emergency stockpile is selected, no matter which demand point in the system is affected by a disaster, the demand for materials can be guaranteed to be met in the shortest possible time. The P-median model, on the other hand, aims to minimize the total transportation cost or time, given the locations of the demand points and the set of potential facility sites. The model aims to find suitable locations for P facilities and assign each demand point to a specific facility to achieve the lowest total weighted travel cost (distance, time, cost, etc.) between the demand points and the sites of the provincial emergency material reserve depots. This approach is applicable when the objective is to optimize spatial efficiency without considering other factors, given the known location of the demand points, the number of demands, and the set of the potential facility sites. The relationship between distance and time at the demand point is directly addressed through the method of weighted distance. By using highway traffic distance weighted with the time of material transportation, the model takes into account both the efficiency of emergency relief and coverage, which better meets the characteristics of emergency disaster relief work in the Qinghai Plateau, especially in the context of frequent natural disasters. The maximum coverage model, which aims to maximize coverage demand, achieves full coverage of all demand points when both the coverage range of the emergency material reserve and the number of demand points are known. However, it does not take into account the demand efficiency of each demand point and untimely demand mismatches may occur. Qinghai Plateau has a special geographical location and is a region where many ethnic groups gather. Therefore, it is necessary to ensure that, in the event of a major natural disaster, the emergency stockpile reflects the fairness of material supply by addressing the sporadic needs of the sparsely populated demand points in marginalized areas, ensuring that the supply of materials from the emergency stockpile covers all the demand points.

In summary, the dual-objective model combining the P-median model with the maximum coverage model, which has the shortest time for emergency supplies to reach the demand point, is more suitable for the optimization of the site selection of the emergency relief supplies depot in Qinghai Plateau.

3.3.3. Maximum Covering Location Problem (MCLP) and P-Median Model

(1) Maximum Covering Location Problem (MCLP)

The Maximum Covering Location Problem (MCLP) is an important type of siting problem that focuses on how to locate and allocate services to maximize the coverage area or demand for a service given a limited number of service facilities. Maximum coverage problems are commonly used for siting public facilities such as fire stations, emergency centers, etc., where these services need to respond quickly and cover as many demand points as possible [39]. The following is a detailed description of the maximum coverage model and the modeling and solution process:

Basic Assumptions:

The locations of demand points and potential service facilities are known.

Each service facility has a fixed service radius.

Service facilities can meet the demands of all demand points within their service range.

Mathematical Model Construction:

Parameters:

I: Set of demand points.

J: Set of potential facility locations.

p: Total number of service facilities.

d_ij: Distance from demand point i to potential facility location j.

R: Service radius of the facility.

w_i: Weight of demand point i, representing demand quantity or intensity.

Decision Variables:

x_j: If a facility is established at location j, then x_j = 1; otherwise, x_j = 0.

y_i: If demand point i is covered by a service facility, then y_i = 1; otherwise, y_i = 0.

Objective Function:

max∑_i∈I w_iy_i

(2)

Constraints:

Service facility quantity constraint:

∑_j∈J x_j = p

(3)

Service coverage constraint:

y_i ≤ ∑_j∈J x_j·δ(d_ij ≤ R) ∀ i ∈ I

(4)

Decision variable range:

x_j ∈ {0,1} ∀ i ∈ J y_i ∈ {0, 1} ∀ i ∈ I

(2) P-median Model

The P-median model is a classical model in the facility siting problem which is mainly applied to determine how to select the location of p facilities in a given set of demand points in terms of number and location and assign a facility to each demand point in order to minimize the total distance or the total cost from all the demand points to their assigned facilities. In the fields of urban planning, logistics and distribution, and emergency services, rational site selection is the key to improving efficiency and reducing costs [40]. The following is a detailed description of the P-neutral siting model and the modeling and solution process:

Basic assumptions: The locations of the demand points and alternative provincial emergency stockpile site locations are known. The distance or cost from the demand point to the facility is calculable, and each demand point must be assigned to a facility.

The mathematical model is constructed as follows:

Basic Assumptions:

The locations of demand points and potential emergency material reserve warehouse sites are known.

The distance or cost from demand points to facilities is calculable.

Each demand point must be assigned to a facility.

Mathematical Model Construction:

Parameters:

I: Set of demand points.

J: Set of potential facility locations.

w_i: Weight of demand point i (e.g., demand quantity).

d_ij: Distance or cost from demand point i to potential facility location j.

Decision Variables:

x_j: If a facility is established at location j, then x_j = 1; otherwise, x_j = 0.

y_ij: If demand point i is assigned to facility j, then y_ij = 1; otherwise, y_ij = 0.

Objective Function:

min∑_i∈I∑_j∈Jw_id_ijy_ij

(5)

Constraints:

Each demand point can only be assigned to one facility:

∑_j∈J y_ij = 1 ∀ i ∈ I

(6)

Only p emergency material reserve warehouse sites can be established:

∑_j∈Jx_j = p

(7)

Demand points can only be assigned to established facilities:

y_ij ≤ x_j ∀ i ∈ I, ∀ j ∈ J

Decision variable range:

x_j ∈ {0, 1} ∀ j ∈ J

y_ij ∈ {0, 1} ∀ i ∈ I, ∀ j ∈ J

4. Analysis of Results

4.1. Transportation Accessibility Analysis to Obtain Alternative Sites

Transportation accessibility is the basis for site selection study in this paper. To ensure the efficiency of shipping emergency supplies from the emergency stockpile, the administrative centers of the 46 counties and districts were used as both the potential sites for the provincial emergency stockpile and the demand points for transportation accessibility analysis.

Based on the natural breaks classification method in ArcGIS 10.8.1, the accessibility data were categorized into five levels: high, higher, medium, lower, and low. The spatial analysis is combined with the administrative divisions of Qinghai Plateau to derive the spatial distribution of transportation accessibility. The areas with medium or higher accessibility are primarily distributed in the southeast of the Qinghai Plateau. These regions have well-developed transportation infrastructures and a dense network of major roads in towns and cities, resulting in better transportation accessibility. Shamdo County, Qumalai County, etc. Because of the location of the Qinghai-South Plateau, the region’s complex terrain, poor natural environment, and the region in addition to the G214 other than the low-grade highway, road infrastructure is backward, and other counties and cities in the administrative seat of the distance, the accessibility is low. Xining city, characterized by its river valley topography and elongated east–west shape, is influenced by its terrain, leading to traffic congestion and poor traffic smoothness. The accessibility of highways is also affected by the overall traffic conditions within the city [41], resulting in relatively poor transportation accessibility.

Considering the principle of high efficiency in organizing the proximity and loading and unloading of personnel and vehicles at the site of the provincial emergency material reserve depot, 19 alternative sites (Figure 3) with medium or higher accessibility value and closer to the main urban area are selected, namely Dulan Depot, Hualong Depot, Minhe Depot, Ping’an Depot, Ledu Depot, Zeku Depot, Gand Depot, Tongren Depot, Guinan Depot, Gonghe Depot, Xinghai Depot, Maqin Depot, Dari Depot, Bamma Depot, Yushu Depot, Wulan Depot, Mardo Depot, Gelmu Depot and Tongde Depot.

4.2. Preliminary Site Selection Options (Option 1)

The preliminary site selection adopts the method of “brand-new site selection”, which means completely re-siting the layout without considering the existing reserve points. This method can also be used to assess and verify the reasonableness of the site selection of the built reserves. In this paper, we choose the “brand-new siting” method, assuming that there are no existing reserve point in the region. Using ArcGIS to conduct spatial analysis, we employed the new location-allocation tool in ArcGIS Network Analyst to add 19 alternative points as potential sites for the provincial emergency reserve, with the administrative centers of 46 counties, districts, and municipalities as the demand point data, selects the maximum coverage model, sets the time impedance as 8 h to solve the problem. The goal was to maximize coverage and minimize the number of reservoirs to be constructed. Initially, we prioritized maximizing coverage and minimizing the number of reservoirs to be constructed. We set the number of sites for the provincial emergency supplies reserve as 1 when combined with the maximum coverage model 46 demand points can not all be covered, continue to set the site for the provincial emergency supplies reserve as 2, 3, 4. Repeatedly carried out the calculation and solution, and finally concluded that when the minimum number of provincial emergency material warehouse locations is set to 4, it can achieve 100% coverage of the 46 demand points. Calculation results show that the 46 demand points can meet the eighth delivery requirement when choosing four depots, namely, the Republican Depot, Golmud Depot, Yushu Depot, and Grande Depot. At this time, there are 10 “basic relief zones” with eighth delivery and 36 “easy relief zones” with 4 h delivery, and the Republican Depot has to undertake the supply of materials for 31 demand points, while the Golmud Depot safeguards three demand points, the Gand Depot safeguards five demand points, and the Yushu Depot safeguards six demand points. The Republic Depot is responsible for supplying 31 demand points, the Golmud Depot for three demand points, the Gand Depot for five demand points, and the Yushu Depot for six demand points. However, in terms of the number of supply–demand points and geographic location, the Republican Depot needs to cover the densely populated area in the eastern part of the Qinghai Plateau and bears the burden of emergency supplies for most of the Qinghai Plateau, which is an area with a high concentration of disaster-carrying bodies in the study area, with the resident population accounting for more than 70% of the province’s population; it also features a great deal of pressure on the supply of emergency supplies, which is high-risk and needs to be further deconstructed.

4.3. Final Optimized Site Selection Option (Option 2)

The construction of the emergency material storage system should comprehensively consider the characteristics of regional disasters, natural geographic conditions, population distribution, productivity layout, and transportation infrastructure, and follow the principles of nearby storage, rapid mobilization, strong protection, scientific assessment, and adaption to local conditions [42]. Therefore, the final site selection plan is based on the preliminary site selection plan combined with the actual Qinghai Plateau to further optimize the site selection with full consideration of natural geography and socio-economic attributes.

4.3.1. Establishment of Additional Emergency Stockpiles to Share Stockpile Pressure

In line with the goal of minimizing the cost, the program siting results on the basis of only 1 increase in the number of stockpile points and attempt to share the preliminary siting program more effectively. However, the construction cost of the library allocation point is extremely high. To address this, we add 19 alternative points for the provincial emergency supplies reserve depot site data, as well as 46 counties, districts, and municipalities as the demand point data, select the maximum coverage model, set the time impedance for 8 h to carry out the solution, first set the number of provincial emergency supplies reserve depot site number of 5 for the calculation of the solution to obtain the full coverage of the 46 demand points, and continue to combine with the P-neutral model to set the 8 h for the time impedance, set the provincial emergency, continuing to combine the P-median model to set 8 h as the time impedance, set the number of site selection points of the provincial emergency reserve depot as 5 for solving, and obtain all the demand points to meet the requirement of 8 h transportation. Finally, we obtain the five site selection points of the provincial emergency reserve depot: Ping’an Depot, Golmud Depot, Yushu Depot, Gande Depot, and Tongde Depot.

4.3.2. Decomposition and Reconstruction of Reserve Banks in High-Population-Density Areas in Northeast Qinghai Plateau

In order to optimize site selection to obtain the peace of the library to undertake 23 demand points of the supply of materials, a region needs to have several provincial emergency supplies depot sites as a safeguard, and the size of the population in the region, the degree of regional influence, the degree of dependence on the facilities of a number of factors, such as, combined with the complexity of the natural geographic environment of the Qinghai Plateau, the frequency of disasters and the population and economy, such as the distribution of bearers, “small Clustering, large dispersion” of the reality, the northeastern region accounted for 70% of the province’s population (Figure 4), coupled with the distribution of the region’s disaster-bearing body is dense, the concentration of disaster–hazardous points, the safety of the library reserve pressure, the risk of damage to the high, in order to cover the densely populated eastern part of the Qinghai Plateau, combined with the practical considerations of setting up a library, the layout of the “double library”. “Dual library” to share the pressure of the safety pool reserve and the risk of damage to the emergency supplies reserve. The Ping’an Depot, identified through the calculation and solution process, is tasked with covering 23 demand points that have dense populations, including Xining, Minhe, and Hualong. This results in significant pressure on the supply and guarantee of emergency relief materials. To address this challenge, the 23 regions within the coverage area of the Ping’an Depot were designated as demand points. In order to ensure efficient rescue operations, the time impedance was set at 4 h (“easy to rescue the population” time requirements). With the Ping’an Depot designated as a mandatory facility point, the number of facility points was set to 2 for the calculation and solution process. Ultimately, this approach yielded two provincial emergency material reserve depot site selection points: the Ping’an Depot and the Chengxi Depot.

Based on the accessibility analysis, the final optimization of the site selection plan was further refined by integrating the natural geographic and socio-economic attributes of the Qinghai Plateau. Using the GIS maximum coverage model in conjunction with the P-median model, we optimized the selection of six sites for the Provincial Emergency Stockpile Depot: the Ping’an Depot, Chengxi Depot, Tongren Depot, Gande Depot, Yushu Depot, and Golmud Depot. At this point, the number of “basic relief zones” with 8 h transportation was reduced to 6 and the number of “easy relief zones” with 4 h transportation was increased to 40 (Figure 5).

5. Discussion

5.1. The Location of Emergency Relief Stockpiles Should Take Full Account of the Time Constraints for Their Delivery

In this paper, when selecting the site for the emergency relief materials reserve depot, combined with the Qinghai Plateau emergency materials, transferring mainly relies on the actual road traffic and transportation. We fully accounted for different highway speed limits. The entire site optimization process is based on the accessibility of traffic, with the “basic relief area” being limited to an 8 h delivery time and the “easy relief area” being limited to a 4 h delivery time to the demand points. Through optimized site selection, the timeliness of materials has been significantly improved. The whole process of site selection optimization is based on traffic accessibility. However, the maximum distribution time (458 min) has not changed, which is mainly due to the two programs in Tanggula Mountain Town. This demand point can only rely on the Golmud Depot as the supplier for the provincial emergency supplies reserve site; the distribution line in the two programs is the optimal choice, and material transfers can only rely on the G109 national highway. The national highway transportation speed is limited, which also reflects the direct impact of geographic factors on the time efficiency of a material transfer (Figure 6).

5.2. The Location of Emergency Relief Material Storage Depots Should Take Full Account of Natural Constraints

In this paper, the influencing factors of natural geography and the socio-economic base of Qinghai Plateau are fully considered in the site selection of the emergency relief materials reserve depot to ensure the practicability and effectiveness of site selection. The Qinghai Plateau is located on the Tibetan Plateau, characterized by complex topography and diverse climate types, which impose special requirements on the storage and transportation of disaster relief materials. Through the optimization of site selection, this study identifies six depots: Ping’an Depot, Chengxi Depot, Tongren Depot, Gande Depot, Yushu Depot, and Golmud Depot. By combining these optimized results with the analysis of the existing depots, it is evident that the locations of the existing Golmud Depot and Chengxi Depot remain unchanged. Therefore, only four new depots, namely Ping’an Depot, Tongren Depot, Gande Depoty, and Yushu Depot, are needed. This approach aligns with the study’s site selection goal of “minimum cost, shortest time, maximum coverage, and practicality”.

5.3. The Location of Emergency Relief Stockpiles Should Take Full Account of Carrier Density and Risk of Damage

In this paper, during the process of selecting the site of the provincial emergency supplies reserve, the population distribution of the Qinghai Plateau, the distribution of transportation infrastructure, and other disaster-bearing bodies have an important impact on the selection of the site for the disaster relief supplies reserve. Given the reality that disaster-bearing bodies in the Qinghai Plateau are “dense in the east and sparse in the west”,and considering the potential damage that large-scale disasters may cause to the provincial emergency supplies reserve, we conducted a risk impact analysis. Based on this analysis, we considered arranging a “double depot” in the eastern part of the Qinghai Plateau to share the risk of possible damage. The practical site selection results not only improved the efficiency of disaster relief but also provided affected people with more timely and effective rescue and assistance guarantees.

6. Conclusions

This study proposes an optimized site selection plan for emergency material storage depots in the Qinghai Plateau by integrating a traffic accessibility analysis, location-allocation models (L-A model), and the Maximum Covering Location Problem (MCLP). The main conclusions are as follows:

(1) Optimized Layout and Significant Improvement

This study proposes an optimized site selection plan for emergency material storage depots in the Qinghai Plateau by integrating a traffic accessibility analysis, location-allocation models, and maximum coverage location planning. By adding four new depots, a six-depot collaborative layout is formed. The 8 h coverage rate reaches 100% (an increase of 20%), the 4 h coverage rate reaches 87% (an increase of 28%), transportation efficiency is improved by 46%, and time costs are reduced by 139 min.

(2) Integration of Geographic and Socio-Economic Factors and Risk Balancing

The study adopts a “dual-depot” strategy to share the storage pressure in densely populated areas and adjusts traffic speed parameters based on the plateau’s geographical characteristics (such as topography and low-oxygen environments). It balances the timeliness requirements of the “basic rescue area” (8 h coverage) and the “easy rescue area” (4 h coverage) and achieves fairness and efficiency in resource allocation through a multi-objective model.

(3) Research Limitations

The study assumes that roads remain passable during disasters and does not consider the potential damage to road networks caused by extreme disasters, which may affect the actual accessibility.

(4) Future Research Directions

Future work will introduce real-time traffic data and disaster simulation technologies to build dynamic emergency material scheduling models, optimize the multifunctional layout of storage depots considering multiple disaster scenarios, and enhance the predictive and adaptive capabilities of the site selection model by integrating artificial intelligence and geographic information systems.

(5) Methodological Applicability and Promotion

The multi-model integration approach proposed in this study can be extended to emergency facility planning in other geographically complex or sparsely populated areas, such as optimizing the site selection of urban emergency services, designing logistics networks in remote areas, and allocating public health resources. It has broad application potential.