Multi-hazard direct economic loss risk assessment on the Tibet Plateau

Abstract

Natural disasters are becoming increasingly regional, frequent, and complex. Semi-quantitative risk assessments based on risk indices or relative risk levels lack quantitative parameters to characterize losses adequately. This study focuses on earthquakes and geological disasters, quantitatively assessing the risk of probable maximum loss (PML) from multi-hazard on the Tibet Plateau (TP) in terms of both absolute and relative values. The results indicate that total PML from multi-hazard on the TP is estimated at 465.5 billion CNY, with a relative value of 5.3%. The total loss value of towns with very high absolute PML is estimated at 254.8 billion CNY, while the relative loss value of towns with very high relative PML is 25.1%. Towns with high absolute value of PML for multi-hazard may have higher fixed asset values, while those with high relative value imply higher vulnerability. Adjusting building structures and reducing the vulnerability of buildings could effectively mitigate the risk of disaster losses caused by multiple hazards.

Keywords:

• Direct economic loss
• multi-hazard
• probable maximum loss
• Tibetan plateau (TP)
• vulnerability

1. Introduction

Due to its strong tectonic activity and complex topography, the Tibet Plateau (TP) is prone to multiple natural disasters, including earthquakes, landslides, debris flows, and flash floods (Cui and Jia 2015). Since 2000, earthquake-triggered landslide events on the TP have accounted for 42% of the total number of such events worldwide (Zhao et al. 2023a). Multi-hazard events with characteristics such as spatio-temporal clustering or chain triggering have broadened the scope of disaster impact and the extent of losses on the TP (Cui et al. 2017). For instance, on September 5, 2022, an Ms 6.8 (Mw 6.6) earthquake struck Luding County, Sichuan Province, China, triggering at least 5336 landslides, affecting 545000 people, and resulting in a direct economic loss (DEL) of 15.5 billion CNY (Dai et al. 2023). Similarly, on June 1, 2022, an Ms 6.1 earthquake occurred in Lushan County, Ya’an City, Sichuan Province, China, leading to approximately 2352 landslides and affecting 14427 people (Shao et al. 2022). Furthermore, on April 25, 2015, the Mw 7.8 Gorhka earthquake hit Nepal, triggering at least 4312 landslides and resulting in the loss of approximately 9000 lives (Kargel et al. 2016).

Around the world, the consecutive hazardous events caused by the 2023 Turkey earthquake and the 2024 Japan earthquake resulted in significant losses. On February 6, 2023, a series of earthquakes with magnitudes of Mw 7.8 and 7.6 struck Mala, Kahraman, Turkey, killing more than 52000 people (Dal Zilio and Ampuero 2023) and causing direct infrastructure losses exceeding 34 billion USD (Ozkula et al. 2023). This event triggered at least 3673 landslides (Görüm et al. 2023). On January 1, 2024, an earthquake with a magnitude of Mw 7.5 on the Nengden Peninsula in Japan killed at least 241 people (Ishikawa 2024), followed by a tsunami with a maximum rise of 6.2 meters (Heidarzadeh et al. 2024) and at least 930 landslides in mountainous areas (Gomez 2024).

The expansion of infrastructure and socio-economic development has heightened asset exposure risks amidst the prevalence of multiple hazards on the TP (Wu et al. 2019; Ran et al. 2022). Between 2000 and 2020, the annual urbanization growth rate on the TP stood at 12.6%, with a significant portion of this expansion occurring in regions highly susceptible to geological, hydrological, and other hazards (Yin et al. 2024). Moreover, infrastructure such as roads, cable channels, railways, and associated facilities on the TP are extensively distributed in areas prone to debris flows and floods (Chang et al. 2021; Cui et al. 2022). For instance, along the Ya’an to Nyingchi section of the Qinghai–Tibet Railway, a total of 1702 instances of rockfalls, landslides, and debris flows were recorded (Zou et al. 2020).

The United Nations’ Agenda for Sustainable Development for the twenty first Century first introduced the concept of ‘multi-hazard’ (UNEP 1992). Assessing the risks posed by multiple potential hazards is essential for comprehensive disaster risk management and the formulation of disaster reduction measures. Some scholars have developed multi-hazard susceptibility zoning maps using factors like hazard-influencing factors and land use data to assess the exposure to multi-hazard within susceptibility zoning maps (Nachappa et al. 2020; Yanar et al. 2020; Karakas et al. 2023).

The primary challenges in current multi-hazard risk assessment research include scientifically and reasonably depicting the relationships and impacts of multiple hazards, as well as quantitatively expressing the risks associated with them (De Angeli et al. 2022; López-Saavedra and Martí 2023). Gill and Malamud (2016) summarized the relationship between multiple hazards as triggering, increased probability, and catalysis/impedance. Tilloy et al. (2019) classified the relationship among multiple hazards into triggering, change condition, compound, independence, and mutually exclusive categories. Lee et al. (2024) utilized the Global Hazard Database to categorize global multi-hazard events into four categories: preconditioned/triggering, multivariate, temporally compounding, and spatially compounding multi-hazard events, and highlighted landslides as the primary secondary hazards. Karatzetzou et al. (2022) devised an earthquake-flood hazard scenario and evaluated the individual and combined (triggered) hazard risks confronting the road network. Liu et al. (2021) examined the triggering relationships among different hazards using a hazard matrix, employed radar maps to assess the multi-hazard intensity influenced by the hazard coupling effect in Beijing Huairou Science City, and integrated land use data to perform multi-hazard risk analysis.

Currently, research in the field of multi-hazard events on the TP primarily focuses on two main areas. First, combining the susceptibility of individual hazards to obtain the spatial distribution characteristics of multi-hazard susceptibility. Secondly, exploring multi-hazard chains within a small scale through historical disaster cases and numerical simulations, including scenario reconstruction of formula characteristics. Rusk et al. (2022) combined susceptibility maps of floods, landslides, and wildfires in the Hindu Kush Himalayas to identify areas with high susceptibility to multi-hazard events. Zhu et al. (2021) conducted a hypothetical scenario simulation of the landslide-flood disaster chain based on the heavy rainfall–debris flow–dam break–landslide disaster chain in Aniangzhai, Sichuan Province, China, on June 17, 2020. Based on field surveys and remote sensing interpretation, Zhao et al. (2021) identified the locations of landslides and barrier lakes triggered by the Linzhi earthquake on November 18, 2017, and then constructed a landslide susceptibility map.

Disaster risk is commonly defined as the likelihood of life or asset loss or destruction within a specified timeframe (UNISDR 2015). Several studies have examined disaster risks on the TP focusing on economic losses. Yang et al. (2023) integrated inundation range, flood depth, land use, and socioeconomic data to produce a flood loss risk map for the Yarkand River Basin in Xinjiang, China. Chen et al. (2024) employed probabilistic models and numerical simulations to estimate the DEL incurred by buildings from the Yangling Gully debris flow. Wei et al. (2022) assessed the risk of building loss in two villages in Jiuzhaigou by considering the intensity of debris flow and rockfall disasters, along with building vulnerability.

However, there remains a paucity of studies focusing on multiple hazard events and conducting quantitative analyses of economic risks on a large scale on the TP. Quantifying the DEL risks associated with various hazards on the TP, estimating their potential financial impact, and mitigating the accumulation of ‘bad assets’ will aid decision-makers in conducting cost-effective analyses of disaster impacts, avoiding capital inflow into high-risk areas, allocating resources judiciously, and enhancing the future development prospects of the social economy (UNISDR 2013, 2015; Wu et al. 2017; Inter-Agency Standing Committee and the European Commission 2020).

Therefore, our focus lies on earthquakes and geological hazards, which inflict severe damage and occur frequently on the TP. We quantitatively assessed the probable maximum loss (PML) risk from multi-hazard to economic assets. The primary objective of this study is to identify regions with high values of loss risk and loss rates, providing a quantitative depiction of the risk patterns associated with the multi-hazard on the TP in terms of loss magnitude. Furthermore, we explored the impact of changes in vulnerability indicators on mitigating PML from multi-hazard on the TP.

2. Materials and methods

2.1. Study area

The study area is situated on the TP, spanning from 24°16′24″N to 40°1′9″N and 73°29′56″E to 105°40′29″E (Figure 1). The interior of the TP is relatively flat, whereas the topography at the plateau’s edge is undulating and characterized by steep slopes (Fielding et al. 1994; Liu-Zeng et al. 2008). Notably, all peaks in the world over 7000 m are located on the TP (Yao et al. 2021). The altitude range in the study area is 86–8682 m, with a relative height difference of 8596 m. The annual mean temperature of the TP decreases from east to west, while the annual mean precipitation decreases from southeast to northwest (Kuang and Jiao 2016). In the future, the plateau is expected to experience a general warming trend, which will increase the risk of geological hazards (Cui and Jia 2015; You et al. 2021; Ya et al. 2023). In the study area, numerous rivers are predominantly distributed in the east and south, with many of their valleys characterized by deep terrain, rendering them more susceptible to landslides (Wang et al. 2021; Zhao et al. 2023b).

Figure 1. Location of the study area: the names of 6 provinces are shown.

Owing to crustal uplift and intense river downcutting, mountain disasters like landslides and debris flows occur frequently (Cui et al. 2019). It is estimated that there are over 2000 debris flows and more than 1500 landslides on the TP (Cui and Jia 2015). The increasing frequency of geological hazards in China corresponds to a rising trend in economic losses. As per the China Statistical Yearbook, the DEL attributed to geological disasters in China amounted to 2.3 billion CNY and 5.0 billion CNY in 2019 and 2020, respectively. Notably, Sichuan Province incurred the highest DEL due to geological disasters, totaling 2.3 billion CNY and securing the top rank. Gansu Province and Yunnan Province followed, with losses ranking second and third, respectively.

The TP is one of the most seismically active regions in the contemporary era, characterized by six major fault zones and six active blocks (Zhang et al. 2003). According to statistics, approximately 118 earthquakes of Mw 7 or higher have been recorded on the TP, with the majority of them concentrated on the southern and eastern margins (Deng et al. 2014). Notably, earthquakes with Mw 7 or above were predominantly located in the Qilian Mountains. Additionally, eight significant earthquakes of Mw 7 or above occurred along the four boundary zones of the Bayan Har Block. The giant arc-shaped Himalayan front thrust-nappe fault zone is known for frequently experiencing Mw 8 earthquakes (Li et al. 2021). Earthquakes occurring on the periphery of the TP are characterized by large magnitudes and result in severe disasters (Xu et al. 2023). In contrast, earthquakes in the central and eastern regions of the plateau also exhibit large magnitudes and lead to significant disasters (Xu et al. 2023). However, earthquakes in the central and western regions of the plateau tend to have moderate magnitudes and cause comparatively minor disasters (Xu et al. 2023).

2.2. Methods

In the context of the study, multi-hazard risk pertains to the risk of PML resulting from the clustering of earthquakes and geological disasters (such as landslides and collapses) in either space or time. PML, in this study, refers to the maximum loss expected to occur within the 475-year return period (UNISDR 2013), encompassing direct damage to buildings, infrastructure, equipment, etc. The risk outcome is expressed through the absolute or relative value of the expected annual loss that the disaster may induce. According to GAR2013 (UNISDR 2013), the relative value of multi-hazard PML denotes the absolute value of multi-hazard PML as a percentage of fixed assets. Initially, we estimated the PML caused by earthquakes and geological disasters separately using their respective PML models. Subsequently, we aggregated and combined the results to obtain a quantitative assessment of multi-hazard PML risks at the township level on the TP, alongside risk zoning maps (Figure 2).

Figure 2. Methodology flow chart of this study.

2.2.1. Earthquake risk assessment of PML

According to the disaster risk framework for establishing a disaster loss risk assessment model (Wu et al. 2019), the functional relationship between the regional earthquake PML value (L) and hazard factors (H), exposure factors (E), vulnerability factors (V), and physical environmental conditions (N) is as follows, (1) $$L=H^{\alpha }\times E^{\mathrm{\beta }}\times V^{\mathrm{\gamma }}\times N^{\mathrm{\delta }}\times e^{\mathrm{\eta }}$$(1)

Based on data from 192 seismic events in mainland China spanning from 1990 to 2015, seismic intensity, fixed asset value, non-steel-concrete residential building (NSCRB) proportion, and terrain undulation were ultimately chosen as input variables for the assessment model (He 2019). The analysis of the relationship between the assessed value and the observed value indicated that the model, with R² = 0.7 (p < 0.05), can explain 70% of the changes in PML. The results of the PML risk assessment were found to be satisfactory, with the calibrated parameters for $\alpha ,$ $\beta ,$ $\gamma ,$ $\mathrm{\delta },$ and $\mathrm{\eta }$ being 8.05, 0.64, 3.25, 0.11, and −16.26, respectively.

2.2.2. Geological disaster risk assessment of PML

PML model of geological hazards is given by Nadim et al. (2006), (2) $$R=G\times \mathit{\text{WKS}}\times \mathit{\text{Vul}}$$(2) (3) $$R=\mathit{\text{PHExp}}\times \mathit{\text{Vul}}$$(3) where $R$ is the risk of geological hazards, $G$ is the hazard of geological hazards, $\mathit{\text{WKS}}$ is the value of fixed assets exposed to the risk of geological disasters, $\mathit{\text{Vul}}$ is the DEL rate, $\mathit{\text{PHExp}}$ is the exposure.

The calculation of geological hazard model is given by Nadim et al. (2006), (4) $$G=(S_t\times S_l\times S_h)\times (T_s+T_p)$$(4) where $G$ is the relative geological hazard (Figure S1), $S_t$ is the slope factor, $S_l$ is the lithological condition factor, $S_h$ is the soil moisture condition, $T_p$ represents the precipitation factor, and $T_s$ represents the seismic condition.

The exposure is estimated by multiplying the occurrence frequency of geological disasters by the value of fixed assets in the exposed area, that is, the exposure degree of township i ($\mathit{\text{PHExp}}$), (5) $$\mathit{\text{PHExp}}=\sum G_i\times {\mathit{\text{WKS}}}_i$$(5) where $G_i$ represents the probability of occurrence of disaster events within a 1 km² spatial grid unit (Table 1), and ${\mathit{\text{WKS}}}_i$ is the value of fixed assets within a 1 km² spatial grid unit. Assuming that a typical severe geological disasters event impacts, on average, 10% of the assets in the exposed area.

Table 1. The probability of occurrence of geological hazard events within a 1 km² spatial grid unit.

Geological hazard level	Frequency of geological hazard events within a 1 km^2 grid
Very low	0.0025–0.01%
Low	0.0063–0.025%
Moderate	0.0125–0.05%
High	0.025–0.1%
Very high	0.05–0.2%

2.2.3. Multi-hazard risk assessment of PML

The PML from multi-hazard on the TP are obtained by accumulating the PML of earthquakes (Figure S2) and geological disasters (Figure S3). The classification method for PML risk of multi-hazard refers to the World Resources Institute (WRI) and adopts a quantile classification method to classify them. The PML for multi-hazard were divided into five risk levels: very low, low, moderate, high and very high.

The PML from earthquakes and geological disasters are divided into five grades using a grading method. Different risk zones of PML from multi-hazard were obtained by combining the risk grades of individual disasters (Table 2). The risk classification of multi-hazard on the TP was divided into five levels: low multi-hazard risk (LMR), moderate multi-hazard risk (MMR), high multi-hazard risk (HMR), high earthquake risk (HER), and high geological risk (HGR).

Table 2. Multi-hazard risk classification method based on the expected DEL risk level induced by a single hazard.

Risk level		Geological risk
		Very low	Low	Moderate	High	Very high
Earthquake risk	Very low	LMR	LMR	MMR	HGR	HGR
	Low	LMR	LMR	MMR	HGR	HGR
	Moderate	MMR	MMR	MMR	HGR	HGR
	High	HER	HER	HER	HMR	HMR
	Very high	HER	HER	HER	HMR	HMR

Note: Low multi-hazard risk (LMR), moderate multi-hazard risk (MMR), high multi-hazard risk (HMR), high earthquake risk (HER), and high geological risk (HGR).

2.3. Data

This study used four types of data (Figure S4). First, the peak ground acceleration (PGA) data and seismic intensity data for the 475-year return period were acquired from the Chinese Earthquake Administration (https://www.csi.ac.cn/).

Second, asset value data were obtained from Wu et al. (2014, 2018). They established Chinese prefecture-level asset values, then used fixed asset value spatialization technology combined with population density, night lighting, and road network density, created the grid distribution map of the fixed assets value of China in 2015. Subsequently, by drawing on the research findings of Wu et al. (2017), we updated and generated the spatial distribution map of fixed assets value in the study area for 2020.

Third, NSCRB proportion was derived from Chinese census data (https://data.cnki.net/), which included the number of houses based on load-bearing types at the county level across the entire study area in 2010 and at the county level in Tibet in 2020. The DEL value of geological disasters was sourced from the statistical yearbook of China (https://data.cnki.net/) from 2004 to 2022. The proportion of DEL for geological disasters to exposed fixed assets was calculated to obtain the DEL rate of geological disasters year by year. Then, the multi-year average was calculated to obtain the DEL rate of geological disasters.

Fourth, the terrain relief and slope factor were calculated based on SRTM digital elevation data provided by USGS (https://www.usgs.gov/). The lithological condition factor was derived from the global lithology map (Hartmann and Moosdorf 2012). Soil moisture conditions were assessed using terrestrial soil moisture data of China from 2002 to 2018 (Mao 2021), with the average soil moisture calculated over this period. The 100-year extreme monthly rainfall was determined based on monthly precipitation data from CN051, released by the Climate Change Research Center of the Chinese Academy of Sciences (Wu and Gao 2013).

3. Results

3.1. Absolute PML and risk classes from multi-hazard

The total PML for multi-hazard on the TP is estimated at 465.5 billion CNY. In terms of regional distribution, 30.3% of the total PML for multi-hazard on the TP is concentrated in Sichuan Province, 24.3% in Yunnan Province, 17.2% in Gansu Province, 16.8% in Tibet, 9.3% in Qinghai Province, and 2.2% in Xinjiang (Figure 3a). According to the classification of absolute loss, the absolute value in very high PML risk areas is estimated at 254.8 billion CNY. The absolute value in very high PML risk areas is estimated at 97.4 billion CNY (Figure 3b).

Figure 3. Township-level absolute value (a) and risk class (b) of multi-hazard PML on the TP.

Areas on the TP experiencing very high and high absolute PML for multi-hazard are likely to coincide with regions of significant asset valuation. For instance, the total asset value of towns classified as very high risk amounts to 4191.4 billion CNY, representing 48.0% of the TP's asset value. Meanwhile, the absolute value of PML incurred by towns within these very high-risk areas account for 54.7% of the TP's total PML.

3.2. Relative PML and risk classes from multi-hazard

The relative value of PML could better express the economic impact of disaster events (UNISDR 2013). The relative value of PML for multi-hazard on the TP is 5.3%. The relative value of PML in 64 towns was greater than 50%, accounting for 1.7% of all towns (Figure 4a). According to the classification of relative losses, the relative value in very high PML risk areas is 25.1%. The relative value in high PML risk areas is 14.2% (Figure 4b).

Figure 4. Township-level relative value (a) and risk class (b) of multi-hazard PML on the TP.

Although the absolute loss value and fixed asset value within townships with very high and high relative value risk levels may not be very high, the value of their exposed assets may be more vulnerable compared to other townships. For example, the absolute value of PML in towns with very high relative value risk level is 72.0 billion CNY, and their fixed asset value account for 3.3% of the TP's total asset value. However, the relative value of PML in these towns is 4.7 times higher than that of the TP.

3.3. Multi-hazard PML in different zoning combinations

According to the absolute loss combination risk classification (Figure 5a), 13.0% of towns are confronting HER, with the same percentage facing HGR, while an additional 27.0% are encountering HMR. The PML in the HER is estimated at 77.6 billion CNY, with HGR accounting for 33.4 billion CNY, and HMR totaling 274.6 billion CNY. Based on the relative loss combination risk classification (Figure 5b), 20.3% of towns are facing HER, 19.5% are facing HGR, and 19.6% are facing HMR. The relative value of PML in the HER is 17.0%, in HGR it is 4.8%, and for HMR, it is 17.1%.

Figure 5. Absolute risk class (a) and relative risk class (b) of multi-hazard PML at the township-level on the TP.

4. Discussion

4.1. Regional differences in risk of multi-hazard PML on the TP

Disasters are inevitable, but their impact can be mitigated through government-promulgated disaster mitigation measures (Li et al. 2017). Comprehending the PML in different regions is of great significance for the government to formulate disaster reduction measures and minimize economic losses (De Moel et al. 2009; Whitman et al. 1997).

We calculated the multi-hazard PML and the value of fixed assets in six provinces (Table 3). The absolute value of multi-hazard PML from high to low are Sichuan, Yunnan, Gansu, Tibet, Qinghai, and Xinjiang. The relative value from high to low is Xinjiang, Sichuan, Tibet, Yunnan, Gansu, and Qinghai. The multi-hazard PML in Sichuan Province reached 140.8 billion CNY, making it the largest on the TP. Sichuan Province is situated in the southeastern part of the TP. Earthquakes with a magnitude greater than 7.0 on the Richter scale are primarily distributed along the southern and eastern edges of the TP (Deng et al. 2014). These regions undergo intense tectonic activity, feature steep terrain, and have abundant fragmented materials. The likelihood of landslides triggered by earthquakes is high, presenting a significant risk and resulting in substantial losses (Liu et al. 2015; Kargel et al. 2016). On May 12, 2008, the Wenchuan earthquake occurred, with its epicenter in Wenchuan County, Aba, Sichuan Province. By April 25, 2009, the earthquake had triggered over 50000 landslides, causing 69225 deaths, 374640 injuries, and extensive property losses (Qi et al. 2010).

Table 3. PML for multi-hazard and fixed asset values in six provinces.

Province	Absolute value of PML (billion)	Asset value (billion)	Relative value of PML (%)
Gansu	80.02	1691.25	4.73
Qinghai	43.16	1793.02	2.41
Sichuan	140.82	1974.59	7.13
Tibet	78.23	1260.08	6.21
Xinjiang	10.01	69.26	14.45
Yunnan	113.28	1945.21	5.82

For relative losses, Xinjiang has the highest loss rate, reaching 14.5%. The loss rate is associated with the vulnerability of the affected structures, and the resilience of building structures is one of the key factors influencing the extent of losses (Yamin et al. 2017). Xinjiang had the highest average proportion of non-reinforced concrete structures. Structures like bricks and stones, which lack reinforcement, are more vulnerable to earthquakes and geological disasters (Işık et al. 2023; Adanur 2010).

4.2. The impact of changes in vulnerability indicators on PML

The extent of building damage is closely related to the type of building structure (Kang and Kim 2016). The NSCRB are commonly used to assess vulnerability to natural hazards (Luo et al. 2023). Mitigating disaster risks can be effectively achieved through reducing the vulnerability of buildings (de Ruiter et al. 2021). Figure 6 illustrates the spatial changes in the proportion of NSCRB in Tibet for 2010 and 2020. Compared to 2010, Tibet’s NSCRB ratio decreased by 0.2. Particularly notable is the region of Rikaze, which experienced the most significant decrease in NSCRB, with the average ratio dropping from 0.9 in 2010 to 0.6 in 2020.

Figure 6. Proportion of NSCRB in Tibet in 2010 (a) and 2020 (b).

With the decrease in the number of NSCRB, Tibet’s multi-hazard PML decreased by 39.2 billion CNY, representing a decrease of 50.1%. The multi-hazard PML in Rikaze experienced the most substantial decrease, reaching 63.1% (Figure 7). This outcome underscores the beneficial impact of improving housing building structures in Tibet on reducing the risk of multi-hazard PML.

Figure 7. The multi-hazard PML in Tibet for 2010 (a) and 2020 (b).

4.3. Limitation

This study has two main limitations. First, there are limitations in expressing risk. Given the complexity of earthquake and geological hazard mechanisms, the PML of multi-hazard events in this study was computed by aggregating and combining the PML of individual hazards. The explicit modeling of disaster mechanism processes was lacking, and the distinctive characteristics of multi-hazard factors could not be adequately captured, including interaction relationships and the extent of cascading risks.

Secondly, there is uncertainty in the risk assessment results. We used PML to quantify the DEL risk of multi-hazard. However, according to the concept of PML, we may have underestimated the PML. There are three primary reasons for this. First, the probability of occurrence and the DEL rate of geological disasters in the geological disaster PML model tend to reflect annual averages. This results in significant neglect of years with high historical disaster frequencies and substantial losses, leading to an underestimation of the final loss results. Secondly, with economic development, fixed assets continue to accumulate, and exposure continues to increase, suggesting that risks will expand in the future. Thirdly, we failed to take into account the impact and damage inflicted on individuals by multi-hazard events.

In future research on multiple hazards on the TP, we could focus on a more comprehensive consideration of the various types of hazards occurring on the TP. Additionally, we could aim to more accurately depict the relationships between the effects and impacts of different hazards, while also fully considering the dynamic changing trends of hazards on the TP within the context of climate change. For instance, Shugar et al. (2021) investigated the ice avalanche–mudslide disaster chain in Uttarakhand, India, which occurred in February 2021. Their analysis of the dynamic process, causes, and social impacts of the event contributes to a better understanding of the heightened hazard risks and increased impacts of disasters in the Himalayas amid climate warming. Peng et al. (2023) employed the HEC–RAS hydrodynamic model to analyze the process of ice avalanche–landslide–dam breach and its impact range on two glacial lakes (Ranzeria Co and Jiweng Co) in the Nyainqentanglha Mountains of the TP. Their research contributes to understanding the mechanisms of glacial lake dam breaches and to better managing glacial lake disasters.

5. Conclusion

While academic research on multiple hazards on the TP is growing, there remains a scarcity of studies that quantitatively evaluate risks from the standpoint of PML. Through the integration of models assessing PML stemming from earthquakes and geological disasters, we have estimated the PML for multi-hazard on the TP. Our study offers a more intuitive quantification of the severity of multi-hazard occurrences on the TP and the efficacy of measures aimed at mitigating disaster risks at the building level.

The total PML for multi-hazard on the TP is estimated at 465.5 billion CNY, with a relative value of 5.3%. Risk classification results based on the absolute value indicate that the absolute losses in very high PML risk areas amount to 254.8 billion CNY, while in high PML risk areas, the absolute losses are calculated at 97.4 billion CNY. Regarding risk classification results based on relative values, the relative losses in very high PML risk areas are 25.1%, and in high PML risk areas, they are 14.2%. Towns with very high and high absolute values of PML for multi-hazard often have higher fixed assets value. Conversely, towns with very high and high relative values imply a higher proportion of fixed assets value lost due to earthquakes and geological disasters, indicating greater vulnerability.

Reducing the vulnerability of residential buildings is paramount to mitigating the risk of PML from multi-hazard. In Tibet, the average proportion of NSCRB decreased by 0.2, leading to a 50.1% reduction in PML. Consequently, the implementation of more stringent seismic building codes for existing non-seismically resistant structures and the execution of relocation measures for high-vulnerability communities will diminish potential future property losses.

Disclosure statement

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability statement

The data that support the findings of this study are available from the corresponding author on reasonable request.

Additional information

Funding

This research was funded by the National Key R&D Program of China (Grant No.2022YFC3004404) and the Second Tibetan Plateau Scientific Expedition and Research Program (Grant No.2019QZKK0606) and the National Natural Science Foundation of China (Grant No.42077437).