Multidimensional Assessment of Meteorological Hazard Impacts: Spatiotemporal Evolution in China (2004–2021)

Abstract

Meteorological hazards threaten sustainable development by affecting human safety, economic stability, and food security. Climate change increases extreme weather frequency, underscoring the urgency for comprehensive evaluation frameworks. However, existing frameworks rarely integrate multiple impact dimensions, limiting their practical utility. To address this gap, our core objective is to develop two novel index series, a single-hazard composite impact index (SHCI) and a multi-hazard composite impact index (MHCI), employing entropy weighting to integrate demographic and economic factors, enabling a more holistic assessment of meteorological hazard impacts in China. Analysis of 2004–2021 data on drought, rainstorm and flood (RF), hail and lightning (HL), typhoon, and low-temperature freezing (LTF) revealed decreases in the national MHCI and SHCI. Key results include the following: (1) the relative MHCI decreased by 74.8%, exceeding 61.21% of absolute MHCI; (2) nationally, 2010, 2013, and 2016 had high MHCI values, and Sichuan has the most extreme hazard years (three) among all the provinces; and (3) provincially, Ningxia has the highest absolute and relative MHCI, while SHCIs varied spatially. These findings provide specific references for climate adaptation planning and the optimization of hazard risk reduction strategies. The methodology offers a versatile framework for multi-hazard risk assessment in nations experiencing climatic and demographic transitions.

1. Introduction

Meteorological hazards are the most prevalent and economically disruptive type of natural hazard, accounting for over 80% of global hazard impacts [1]. These hazards significantly threaten not only public safety but also economic development, social stability, and ecosystems. Intensifying compound meteorological hazards (IPCC AR6) necessitate multidimensional assessment frameworks because of their cascading impacts on coupled human–environment systems. China’s distinctive monsoon climate dynamics and topographic complexity render it especially vulnerable to meteorological hazards [2]. Statistical data indicate that from 2004 to 2021, meteorological hazards affected an average of 292.4 million people annually in China, with direct economic GDP losses averaging 0.68% per year [3]. The United States, China, and Italy were the three leading contributing countries in natural disaster-related research from January 1990 to June 2015, which also highlights China’s high level of attention to natural disasters [4]. Furthermore, anticipated increases in the frequency of extreme weather and climatic events in SSP scenarios significantly complicate climate change adaptation efforts [5], further underscoring the urgency of sophisticated impact assessment approaches.

Existing studies on meteorological hazard impact assessment have made significant progress in investigating spatiotemporal patterns and socioeconomic consequences across diverse scales. Globally, research has examined the spatial distribution of major meteorological disasters (e.g., storms, floods, and droughts) [6] and analyzed mortality patterns induced by such hazards in countries like Brazil [7], Australia [8], and South Korea [9]. In China, comprehensive assessments have been conducted at national, regional, and local scales, covering nationwide loss dynamics [10,11,12,13], broader regional analyses (e.g., East China) [14], and focused studies on specific provinces [15] or smaller locales [16,17].

Methodologically, diverse approaches have been employed to quantify multidimensional impacts: Huang et al. developed an SVM-based model to evaluate flood impacts using indicators such as affected population and economic losses [18]; Xie et al. implemented grey clustering to assess agricultural and socioeconomic impacts of multiple hazards (droughts, floods, etc.) [19]; and Shi et al. proposed the Total Risk Index (TRI) and Multi-Hazard Risk Index (MHRI) to quantify population and property risks across 197 countries at a 0.5° × 0.5° grid scale [20]. Regional case studies, such as those on Vietnam [21] and U.S. natural disasters [22], have further enriched the understanding of context-specific impact mechanisms.

However, significant research gaps persist in the current literature. Firstly, most studies rely on limited single indicators (e.g., fatalities [9] and economic losses [18]) to characterize impacts, and even those incorporating multiple indicators [11] lack comprehensive analysis of their disparate influences. Secondly, composite indices often depend on static assumptions or expert assessments, neglecting dynamic interactions between hazards and evolving socioeconomic factors—for instance, models like SVM [23] and TOPSIS [21] derive weights from historical correlations rather than real-time adaptive mechanisms. Thirdly, frameworks predominantly focus on single-hazard scenarios [21,23,24] or regional scales [20], limiting applicability to multi-hazard compound events and cross-regional comparisons. Finally, existing research emphasizes spatiotemporal characterization of impacts but overlooks in-depth exploration of impact structures and fails to account for dynamic changes in socioeconomic conditions (e.g., population and GDP) that correlate with hazard impacts [25,26].

To address the aforementioned limitations, we constructed indices by selecting distinct evaluation indicators for different meteorological hazards, thereby achieving a systematic assessment that integrates multiple hazards and indicators. We employed the entropy weight method to develop a novel dual-index system: the absolute impact index quantifies the direct impacts of meteorological hazards, while the relative impact index illustrates the relative impacts of hazards on society and the economy. By incorporating socioeconomic factors, this system resolves the scale dependence issue in cross-regional comparisons. This study provides a valuable reference for comprehensively evaluating the spatiotemporal distribution pattern of meteorological hazard impacts in China.

The specific objectives of the study are as follows: to construct a single-hazard composite impact index (SHCI) and a multi-hazard composite impact index (MHCI) to assess the comprehensive impacts of single meteorological hazards and multiple meteorological hazards, respectively; and to comprehensively evaluate the spatiotemporal evolution characteristics of meteorological hazard impacts at the national and provincial scales in China based on data from 2004 to 2021.

2. Data

The data utilized for assessing the impact of meteorological hazards from 2004 to 2021 are summarized in

Table 1. Summary of the data used and their sources.

Class		Data Source	Absolute Impact Index	Relative Impact Index
Meteorological Hazard Losses	Population Loss	Yearbook of Meteorological Disasters in China (Yearbook of Meteorological Disasters in China)	Affected Population	Affected Population/Total Population (A/T)
	Economic Loss		Direct Economic Loss	Direct Economic Loss/GDP (E/G)
	Crop Loss		Crop Damage Area	Crop Damage Area/Crop Total Sown Area (D/T)
	House Loss		Number of Damaged Houses (Typhoon-Replace: Number of Collapsed Houses)	Number of Damaged Houses (Typhoon-Replace: Number of Collapsed Houses)/Built-up Area (D/A) (C/A)
Socioeconomic Factors	China Consumer Price Index (CPI)	National Bureau of Statistics of China (National Bureau of Statistics of China)
	Total Population	China Statistical Yearbook (China Statistical Yearbook)
	GDP
	Crop Total Sown Area
	Built-up Area	China Urban Construction Statistical Yearbook (China Urban Construction Statistical Yearbook)

. The losses of meteorological hazards are obtained from the Yearbook of Meteorological Disasters in China. These data encompass the 31 provinces of China, excluding Hong Kong, Macao, and Taiwan. Meteorological hazards are classified into five categories: drought, rainstorm and flood (RF), hail and lightning (HL), typhoon, and low-temperature freezing (LTF). The annual direct economic losses owing to meteorological hazards are uniformly converted into 2021 prices, ensuring consistency by using the China Consumer Price Index (CPI) [27]. The CPI data are obtained from the National Bureau of Statistics of China. The original data used in this paper are presented in Appendix A, Figure A1, Figure A2, Figure A3, Figure A4 and Figure A5. The socioeconomic data used in this study include the total sown area of crops, the national and provincial populations, and the national and provincial gross domestic products (GDPs), which are sourced from the China Statistical Yearbook. Provincial built-up area data are obtained from the Urban Construction Statistical Yearbook. Meteorological hazard losses data and socioeconomic data are collected at the provincial level. The data on extreme climate events are sourced from the Climate Change Center of the China Meteorological Administration and peer-reviewed journal articles. The infrastructure data are derived from local policy documents.

The relative impact on the population is defined as the ratio of the affected population to the total population, which captures the impact density of meteorological hazards. For the economy, the relative impact is measured by the proportion of direct economic losses to GDP, reflecting the impact depth of meteorological hazards. Relative crop impacts are characterized by the ratio of damaged crop area to total sown area, which serves as an indicator of impact breadth. Finally, the relative impact on housing is operationalized as the ratio of damaged houses to built-up areas, which indicates the impact intensity.

3. Methods

3.1. Research Framework

After collecting meteorological hazard losses data and socioeconomic data, the entropy weight method is employed to construct comprehensive impact indices at both national and provincial scales. Each spatial scale includes indices for single-hazard and multi-hazard impacts. Ultimately, the temporal variation characteristics of the national comprehensive impact index and the impact factor structure distribution of the provincial comprehensive impact index are derived, which collectively characterize meteorological hazard impacts across China. The specific research framework is illustrated in Figure 1.

3.2. Entropy Weight Method

The entropy weight method is a technique that calculates the entropy of indicators according to the total influence of the numerical variations in each indicator to establish their respective weights. A higher entropy value of an indicator indicates an increase in disorder, greater uncertainty, diminished information quantity, and reduced weight [28].

The entropy weight method is an objective weighting approach, with weights determined solely by the data, thus avoiding subjective interference from human subjective factors. The analytic hierarchy process relies on experts’ judgments and experience to determine weights, which may lead to subjective biases. Principal component analysis is mainly used for dimensionality reduction, but the practical significance of these principal components may be unclear. This study selects the entropy weight method because it can fully utilize the inherent information of the data and objectively reflect the importance of each indicator in the assessment.

Before the entropy weight method is used for the weight calculation, the data must be normalized. The statistics are divided into two categories: positive indicators and negative indicators. For positive indicators, a higher value represents a superior evaluation; for negative indicators, a lower value denotes a poorer evaluation. Let the initial data matrix X consist of i samples and j indicators, represented as $X=\left(x_{ij}\right)$. The normalization calculation for positive indicators is shown in Equation (1), whereas the normalization calculation for negative indicators is shown in Equation (2).

$$Y_{ij}={\displaystyle \frac{X_{ij}-min{X}_j}{max{X}_j-min{X}_j}}$$

(1)

$$Y_{ij}={\displaystyle \frac{max{X}_j-X_{ij}}{max{X}_j-min{X}_j}}$$

(2)

The ratio of the j-th indicator for the i-th sample is subsequently calculated by the formula presented in Equation (3). $P_{ij}$ represents the proportion of the normalized value of the i-th sample in the j-th indicator to the sum of normalized values of all samples in that indicator, which intuitively reflects the relative importance of the sample in the indicator.

$$P_{ij}={\displaystyle \frac{Y_{ij}}{\sum _{i=1}^i{Y}_{ij}}}$$

(3)

The information entropy of the j-th indicator is then calculated by Equation (4).

$$E_j=-ln{\left(i\right)}^{-1}{\sum }_{i=1}^i{P}_{ij}ln\left(P_{ij}\right)$$

(4)

According to the information entropy, the weight of the j-th indicator is determined, as delineated in Equation (5).

$$w_j={\displaystyle \frac{1-E_j}{j-\sum E_j}}$$

(5)

Finally, $Z_i$, which is the comprehensive score of the i-th sample, is calculated using the formula presented in Equation (6).

$$Z_i={\sum }_{j=1}^j{w}_j\ast Y_{ij}$$

(6)

3.3. Construction of the Comprehensive Impact Index for Meteorological Hazards

In constructing the comprehensive impact index for meteorological hazards, the initial step involves the selection of indicators, where diverse impact indicators are chosen according to different hazard types. The construction encompasses both a single-hazard composite impact index (SHCI) and a multi-hazard composite impact index (MHCI). Both SHCI and MHCI cover two scales: national and provincial. The three matrices shown in Figure 2 are the initial matrices in the construction process of different indices, which are then processed using Equations (1)–(6) to calculate the index values. Among them, the initial matrices for the national and provincial scale MHCI share the same structural form but differ in the underlying data they incorporate.

The SHCI and MHCI can be further categorized into absolute and relative indices. These two variants follow an identical calculation process, with differences only in indicator selection. Therefore, only the absolute is presented in the figure, and the differences in indicator selection are elaborated in subsequent sections. The specific method framework is shown in Figure 2: arrows indicate the sequence of data processing and calculation, while colors are employed solely for emphasis and differentiation between various types of meteorological hazards or distinct samples.

3.3.1. Construction of the SHCI

In the Sendai Framework, core data directly related to disaster impact assessment include the number of deaths, missing persons, and affected people caused by disasters, direct economic losses relative to GDP, as well as damage to critical infrastructure and disruption of basic services [29]. Accordingly, this study selects the affected population, direct economic losses, crop-affected area, and housing losses as assessment indicators. Corresponding socioeconomic variables are also chosen to calculate the relative impacts. A sensitivity analysis is conducted on the selected variables, and the results are shown in Appendix C.

The indicators used for evaluating the impacts of various meteorological hazards are as follows. The indicators used in the calculation of the SHCI are as follows. When calculating the absolute SHCI of drought events, $j=3$, which includes the affected population, direct economic loss, and crop damage area. When calculating the absolute SHCI of RF, HL, and LTF events, $j=4$, which includes the affected population, direct economic loss, crop damage area, and number of damaged houses. When calculating the absolute SHCI of typhoon events, $j=4$, which includes the affected population, direct economic loss, crop damage area, and number of collapsed houses.

When calculating the relative SHCI of drought events, $j=3$, which includes A/T, E/G, and D/T. When the relative SHCIs of RF, HL, and LTF events are calculated, $j=4$, which includes A/T, E/G, D/T, and D/A. When calculating the relative SHCI of typhoons, $j=4$, which includes A/T, E/G, D/T, and C/A.

Construction of the National SHCI

The absolute national SHCI is calculated as follows. The matrix $X_a=\left(x_{ij}\right)$ represents the j-th absolute impact evaluation indicator for the i-th province in the a-th year on a national scale.

$$X_a=\left(\begin{array}{ccc}{x}_{11}& \dots & x_{1j}\\ ⋮& \ddots & ⋮\\ x_{i1}& \dots & x_{ij}\end{array}\right)$$

(7)

where $x_{ij}$ represents the j-th absolute impact value of the i-th province and $i=31$. $X_a$ indicates the absolute national impact evaluation matrix of the a-th year.

The relative national SHCI is calculated as follows. The matrix $Y_a=\left(y_{ij}\right)$ represents the j-th relative impact evaluation indicators for the i-th province in the a-th year on a national scale.

$$Y_a=\left(\begin{array}{ccc}{y}_{11}& \dots & y_{1j}\\ ⋮& \ddots & ⋮\\ y_{i1}& \dots & y_{ij}\end{array}\right)$$

(8)

where $y_{ij}$ represents the j-th relative impact value of the i-th province and $Y_a$ indicates the relative national impact evaluation matrix of the a-th year.

Given that all the impact evaluation indicators are positive indicators, Equation (1) was employed to normalize matrices $X_a$ and $Y_a$.

$$X_a^{\ast }=\left(\begin{array}{ccc}{x}_{11}^{\ast }& \dots & x_{1j}^{\ast }\\ ⋮& \ddots & ⋮\\ x_{i1}^{\ast }& \dots & x_{ij}^{\ast }\end{array}\right)$$

(9)

$$Y_a^{\ast }=\left(\begin{array}{ccc}{y}_{11}^{\ast }& \dots & y_{1j}^{\ast }\\ ⋮& \ddots & ⋮\\ y_{i1}^{\ast }& \dots & y_{ij}^{\ast }\end{array}\right)$$

(10)

where $X_a^{\ast }$ and $Y_a^{\ast }$ represent the normalized matrices of $X_a$ and $Y_a$, respectively, and $x_{ij}^{\ast }$ and $y_{ij}^{\ast }$ indicate normalized $x_{ij}$ and $y_{ij}$, respectively.

Equations (5) and (6) were subsequently applied to Formulas (9) and (10) to calculate the weight coefficient $w_j$ of the j-th impact indicator and derive the absolute and relative national SHCIs of the a-th year, respectively.

Construction of the Provincial SHCI

The absolute provincial SHCI is calculated as follows. The matrix $Z_b=\left(z_{kj}\right)$ represents the j-th absolute impact indicator for the b-th province in the k-th year.

$$Z_b=\left(\begin{array}{ccc}{z}_{11}& \dots & z_{1j}\\ ⋮& \ddots & ⋮\\ z_{k1}& \dots & z_{kj}\end{array}\right)$$

(11)

where $z_{kj}$ represents the j-th absolute impact value in the k-th year, with $k=18$, covering the years from 2004 to 2021, and $Z_b$ denotes the absolute SHCI of the b-th province.

The relative provincial SHCI is calculated as follows. The matrix $R_b=\left(r_{kj}\right)$ of the j-th relative impact indicators for the b-th province in the k-th year is constructed.

$$R_b=\left(\begin{array}{ccc}{r}_{11}& \dots & r_{1j}\\ ⋮& \ddots & ⋮\\ r_{k1}& \dots & r_{kj}\end{array}\right)$$

(12)

where $r_{kj}$ represents the j-th relative impact value in the k-th year and $R_b$ denotes the relative SHCI of the b-th province.

Since all the impact evaluation indicators are positive indicators, Equation (1) was used to standardize matrices $Z_b$ and $R_b$.

$$Z_b^{\ast }=\left(\begin{array}{ccc}{z}_{11}^{\ast }& \dots & z_{1j}^{\ast }\\ ⋮& \ddots & ⋮\\ z_{k1}^{\ast }& \dots & z_{kj}^{\ast }\end{array}\right)$$

(13)

$$R_b^{\ast }=\left(\begin{array}{ccc}{r}_{11}^{\ast }& \dots & r_{1j}^{\ast }\\ ⋮& \ddots & ⋮\\ r_{k1}^{\ast }& \dots & r_{kj}^{\ast }\end{array}\right)$$

(14)

where $Z_b^{\ast }$ and $R_b^{\ast }$ represent the standardized matrices of $Z_b$ and $R_b$, and $z_{kj}^{\ast }$ and $r_{kj}^{\ast }$ indicate standardized $z_{kj}$ and $r_{kj}$, respectively.

Then, Equations (5) and (6) are applied to Formulas (13) and (14) to calculate the weight coefficient $w_j$ of the j-th impact evaluation indicator and derive the absolute and relative provincial SHCIs of the b-th province, respectively. According to the natural breakpoint method, the indices were divided into five impact levels. An example of calculation using this method is provided in Appendix B.

3.3.2. Construction of the MHCI

When calculating the MHCI, $m=5$, which includes drought, RF, HL, typhoon, and LTF events.

Construction of the National MHCI

The absolute national MHCI is calculated as follows. The matrix $U_a=\left(u_{mj}\right)$ of j absolute impact evaluation indicators for m meteorological hazards in the a-th year nationwide is constructed.

$$U_a=\left(\begin{array}{ccc}{u}_{11}& \dots & u_{1j}\\ ⋮& \ddots & ⋮\\ u_{m1}& \dots & u_{mj}\end{array}\right)$$

(15)

where $u_{mj}$ represents the j-th absolute impact value of the m-th meteorological hazard. $U_a$ indicates the absolute national MHCI of the a-th year.

The relative national MHCI is calculated as follows. The matrix $V_a=\left(v_{mj}\right)$ of j relative impact evaluation indicators for m meteorological hazards in the a-th year nationwide is constructed.

$$V_a=\left(\begin{array}{ccc}{v}_{11}& \dots & v_{1j}\\ ⋮& \ddots & ⋮\\ v_{m1}& \dots & v_{mj}\end{array}\right)$$

(16)

where $v_{mj}$ represents the j-th relative impact value of the m-th meteorological hazard. $V_a$ indicates the relative national MHCI of the a-th year.

Since all the impact evaluation indicators are positive, Equation (1) was used to normalize matrices $U_a$ and $V_a$.

$$U_a^{\ast }=\left(\begin{array}{ccc}{u}_{11}^{\ast }& \dots & u_{1j}^{\ast }\\ ⋮& \ddots & ⋮\\ u_{m1}^{\ast }& \dots & u_{mj}^{\ast }\end{array}\right)$$

(17)

$$V_a^{\ast }=\left(\begin{array}{ccc}{v}_{11}^{\ast }& \dots & v_{1j}^{\ast }\\ ⋮& \ddots & ⋮\\ v_{m1}^{\ast }& \dots & v_{mj}^{\ast }\end{array}\right)$$

(18)

where $U_a^{\ast }$ and $V_a^{\ast }$ represent the normalized matrices of $U_a$ and $V_a$, respectively, and $u_{mj}^{\ast }$ and $v_{mj}^{\ast }$ indicate normalized $u_{mj}$ and $v_{mj}$, respectively.

Then, Equations (5) and (6) are applied to Formulas (17) and (18) to calculate the weight coefficient $w_j$ of the j-th impact evaluation indicator, and the absolute and relative national MHCIs of the a-th year are derived.

Construction of the Provincial MHCI

The absolute provincial MHCI is calculated as follows. The matrix $H_{ad}=\left(h_{mj}\right)$ of j absolute impact evaluation indicators for m meteorological hazards in the a-th year for the d-th province is constructed.

$$H_{ad}=\left(\begin{array}{ccc}{h}_{11}& \dots & h_{1j}\\ ⋮& \ddots & ⋮\\ h_{m1}& \dots & h_{mj}\end{array}\right)$$

(19)

where $h_{mj}$ represents the j-th absolute impact value of the m-th meteorological hazard. $H_{ad}$ indicates the absolute MHCI of the d-th province in the a-th year.

The relative provincial MHCI is calculated as follows. The matrix $I_{ad}=\left(i_{mj}\right)$ of j relative impact evaluation indicators for m meteorological hazards in the e-th year for the d-th province is constructed.

$$I_{ad}=\left(\begin{array}{ccc}{i}_{11}& \dots & i_{1j}\\ ⋮& \ddots & ⋮\\ i_{m1}& \dots & i_{mj}\end{array}\right)$$

(20)

where $i_{mj}$ represents the j-th relative impact value of the m-th meteorological hazard. $I_{ad}$ indicates the relative MHCI of the d-th province in the a-th year.

Since all the impact evaluation indicators are positive, Equation (1) is used to normalize matrices $H_{ad}$ and $I_{ad}$.

$$H_{ad}^{\ast }=\left(\begin{array}{ccc}{h}_{11}^{\ast }& \dots & h_{1j}^{\ast }\\ ⋮& \ddots & ⋮\\ h_{m1}^{\ast }& \dots & h_{mj}^{\ast }\end{array}\right)$$

(21)

$$I_{ad}^{\ast }=\left(\begin{array}{ccc}{i}_{11}^{\ast }& \dots & i_{1j}^{\ast }\\ ⋮& \ddots & ⋮\\ i_{m1}^{\ast }& \dots & i_{mj}^{\ast }\end{array}\right)$$

(22)

where $H_{ad}^{\ast }$ and $I_{ad}^{\ast }$ represent the normalized matrices of $H_{ad}$ and $I_{ad}$, respectively, and $h_{mj}^{\ast }$ and $i_{mj}^{\ast }$ indicate normalized $h_{mj}$ and $i_{mj}$, respectively.

Then, Equations (5) and (6) are applied to Formulas (21) and (22) to calculate the weight coefficient $w_j$ of the j-th impact evaluation indicator, and the absolute and relative provincial MHCIs of the d-th province in the a-th year are derived. Finally, the average value during the period from 2004 to 2021 is calculated as the MHCI for the d-th province, which is divided into five impact levels according to the natural breakpoint method.

3.4. Identifying Extreme Years of Meteorological Hazards

The leverage statistic $h_i={\displaystyle \frac{1}{n}}+{\displaystyle \frac{{\left(x_i-\overline{x}\right)}^2}{{\sum }_{i^{’}=1}^n{\left(x_{i^{’}}-\overline{x}\right)}^2}}$ is calculated. When the leverage statistic of a given observation is greater than three times the average leverage value of all observations, it is considered a high leverage point [30], and the corresponding year is considered an extreme year owing to the comprehensive impacts of multiple meteorological hazards.

4. Results

4.1. Characteristics of Meteorological Hazard Impact in China from 2004 to 2021

4.1.1. Meteorological Hazard Impact in China from 2004 to 2021

Figure 3 shows temporal trends in national absolute (solid lines) and relative (dashed lines) MHCI and SHCI from 2004 to 2021. The absolute index reflects the scale of physical impacts caused by hazards, while the relative index reflects the intensity of impacts relative to the regional socioeconomic base. Both indices are calculated using the entropy weight method, with their values ranging from 0 to 1 (where 0 indicates no impact and 1 indicates the maximum impact within the study period), facilitating comparisons across different disasters and years. Different colors are solely used to distinguish between different disaster types.

From 2004 to 2021, the overall changes in both the absolute value and relative value of China’s national MHCI were downward (Figure 3a). The average decreasing rate of the relative national MHCI obtained by linear fitting is −0.046 year⁻¹, which exceeds that of the absolute national MHCI (−0.0216 year⁻¹). It should be noted that the "decline " mentioned here all refers to the decrease in index values, rather than the reduction in actual physical losses. From 2004 to 2021, the linear fitting results of the affected population (-2.08 × 10⁷ person year⁻¹), crop damage area (−7.16 × 10⁵ hectares year⁻¹), and the number of damaged houses (−2.44 × 10⁵ units year⁻¹) all showed an overall downward, while the direct economic losses (1.13 × 10¹⁰ yuan year⁻¹) showed an upward. Most of the actual physical losses also showed a downward, which indicates that the changes in the MHCI are consistent with the actual impacts of meteorological hazard.

The national MHCIs are notably high in 2010, 2013, and 2016. Upon eliminating these years, the average absolute national MHCI decreased by 13.22%, and the average relative national MHCI declined by 8.90%. The year 2010, an El Niño decay year, saw severe droughts in the southwest region and strong typhoons hitting the southeast coastal areas; in 2013, there was excessive autumn rainfall in West China, with severe rainstorm and flood in Sichuan and Gansu, coupled with high-intensity typhoons; 2016 was a super El Niño year, witnessing floods in the Yangtze River basin and extreme high temperatures in northern China. In years with significant impacts of meteorological hazards, severe meteorological hazard events often occur, which indicates that while strengthening the overall prevention of meteorological hazards, efforts should also be made to enhance the capacity to respond to extreme meteorological events.

From 2004 to 2021, the overall changes in both the absolute and relative national SHCIs of drought and HL were pronouncedly downward (Figure 3b,d). The absolute average rates of decrease are 2.41% and 2.27%, respectively, exceeding that in MHCI of 2.16%. The absolute and relative national SHCIs of RF, typhoon, and LTF events also slightly decreased overall, with absolute average rates of decline of 0.38%, 1.30%, and 1.49%, respectively (Figure 3c–f). The average rates of decrease in each relative national SHCI are all lower than those in the relative MHCI (4.60%). The decline in MHCI and SHCI is the combined result of climate change and improvements in national risk management. From 2004 to 2021, China implemented policies such as the National Comprehensive Disaster Prevention and Mitigation Plan [31] and the Regulations on Meteorological Disaster Prevention [32], which enhanced the country’s capacity to respond to meteorological disasters. During the same period, the national annual average number of days with extreme precipitation (≥50 mm) showed an upward trend [33].

National SHCIs for the five hazards showed significant fluctuations in specific years. The drought, RF, and HL values dramatically increased, suggesting an increased likelihood of causing severe societal impacts nationwide. The national SHCI of typhoons shows abnormal increases alone in 2005, 2006, and 2013, but the SHCIs of LFT events were exceptionally high in 2008.

4.1.2. Relationship Between the National Absolute and Relative MHCIs and SHCIs from 2004 to 2021

Figure 4 illustrates scatter plots and fitted lines of absolute and relative indices for different hazards, where the coefficient of determination (R²) quantifies the goodness of fit, reflecting the strength of the correlation between absolute and relative indices. A strong association exists between the absolute and relative impact indices of meteorological hazards in China (Figure 4). Their correlation is particularly robust for LTF, indicating that the relative influence of LTF depends mainly on the magnitude of the absolute impact. These findings underscore that for strongly associated hazards, proactive measures should be implemented in advance in areas where such hazards occur frequently to mitigate their absolute impacts.

In contrast, the correlation for RF events is comparatively weak, implying that socioeconomic factors exert a greater impact on their relative impact. These phenomena demonstrate that for hazards with a weak correlation (R² < 0.8), other aspects must be evaluated to allocate resources reasonably. In regions susceptible to RF, the drainage infrastructure is essential to increase urban flood control capacity. In drought-prone regions, the establishment of water conservancy facilities and the promotion of water-efficient irrigation technologies are critical measures to bolster drought resilience.

4.2. Characteristics of the Impact of Meteorological Hazards in China by Province from 2004 to 2021

4.2.1. Spatial Distribution Pattern of the Impact of Multiple Meteorological Hazards in China by Province

According to the provincial absolute MHCI grades (Figure 5a), Ningxia is the most severely impacted by multiple meteorological hazards, with the absolute MHCI reaching 0.38. In contrast, the Beijing–Tianjin–Hebei region, Henan, and Shanghai are among the least-impacted areas, with absolute MHCIs less than 0.19, meaning their impact degree is less than half of that in Ningxia. The general spatial distribution of the MHCIs reveals that northern regions experienced greater impacts than southern regions, and western regions than eastern regions. Through cluster analysis, it is found that Sichuan presents high–high clustering, while the Beijing–Tianjin–Hebei region belongs to a low–low clustering area.

In terms of the absolute impact structure within each province (Figure 5b), the influence of a single factor or the synergistic effect of two factors is notably prominent in most provinces. The impact in Ningxia is predominantly characterized by population and economic losses, with proportions exceeding 30%, although the impact on crops is comparatively minor, accounting for 8.46%. The losses of homes in the western and central regions are significant, with Shaanxi, Shanxi, Hubei, and Hunan accounting for nearly 50% of all the homes lost. The population losses in the northeast region are slightly greater, whereas the economic losses in the eastern and southeastern coastal regions are greater, with Shanghai accounting for 46.63% of the total losses. The crop losses in each province are relatively low, except those in Beijing, with the proportion reaching 30.32%.

From 2004 to 2021, the distribution of provincial relative MHCIs closely mirrored that of the absolute MHCIs. Overall, the impacts in the north are greater than those in the south, and the impacts in the west are higher than those in the east (Figure 5c). Ningxia and Qinghai had the most pronounced relative impacts, with relative MHCIs of 0.42 and 0.34, respectively. Shanghai has the lowest relative MHCI of 0.11. The disparity in the relative impact among provinces exceeds that in the absolute impact. Shandong has the highest relative impact index among the eastern coastal regions, whereas Tibet has comparatively lower relative impact indices in relation to its neighboring areas. Cluster analysis reveals that Gansu and Qinghai are high–high cluster areas, whereas Jiangsu and Zhejiang are low–low cluster areas.

On the basis of the relative impact structure in each province (Figure 5d), home losses are substantial in most provinces. Shanxi, Shaanxi, and Shandong account for nearly 50% of all home losses nationwide. The population losses are pronounced in the northeast and eastern coastal regions, whereas the economic losses are notably high in southern provinces.

When the absolute and relative MHCIs are integrated, the impacts on provinces are classified into five tiers (Figure 5e). The impact grades progress from the green zone to the red zone, with corresponding increases in the severity of the impact of meteorological hazards. Provinces in the red and orange zones should bolster infrastructure development to increase their capacity to mitigate meteorological hazards.

Comparing the absolute and relative provincial MHCIs reveals that many provinces are near the y = x line, suggesting that the relative and absolute impacts in these regions are comparable. Provinces with relatively low indices, including Beijing, Hebei, Tibet, and Guangxi, exhibit modest absolute and relative impact indices for meteorological hazards. Provinces such as Gansu, Shandong, and Guizhou present significant absolute and relative impacts, indicating a lower tolerance for meteorological hazards. Sichuan, Liaoning, Shanxi, Shaanxi, Zhejiang, and Shanghai Provinces suffer higher absolute MHCIs than relative MHCIs, indicating robust resilience to meteorological hazards, and more efforts are needed to reduce the absolute impact. Regions such as Ningxia, Qinghai, and Tianjin present relatively high MHCIs, which indicates that their diminished capacity to mitigate meteorological hazards and underscores the need for enhanced socioeconomic development.

4.2.2. Characteristics of Extreme Years of Multiple Meteorological Hazards in China by Province

From 2004 to 2021, both the provincial absolute and relative MHCIs decreased. Only the absolute MHCI of Henan exhibits a marginal increase. Severe rainstorm disasters in Henan in 2021 resulted in substantial losses. Nevertheless, the overall changes from 2004 to 2020 for Henan were also downward when the peak is excluded. Similar patterns are observed in other provinces, with extremely prominent impacts in specific years. Given the decline in the impact of meteorological hazards across provinces, focusing on extreme years is crucial to mitigate the impacts of extreme meteorological hazards.

Data points were plotted with the provincial absolute MHCI as the abscissa and the relative MHCI as the ordinate. Extreme years of each province from 2004 to 2021 are identified according to the method mentioned in Section 3.4. No years with extreme absolute or relative MHCIs are recorded in Liaoning, Anhui, or Inner Mongolia (Figure 6), suggesting that the impacts of meteorological hazards remain relatively stable in these provinces during the study period. In 15 provinces, including Beijing, there is one extreme year, indicating that the frequency of extreme meteorological events in these regions is very low but can lead to a significant societal impact. Two extreme years in 12 provinces, including Gansu, indicate that these regions experience a heightened impact of extreme meteorological hazards and require enhanced response measures and post-hazard recovery capabilities. Sichuan experienced three years of extreme meteorological impacts, representing the most severe conditions during this period. Thus, improving the capacity to manage extreme meteorological hazards is essential and urgent for Sichuan.

4.3. Spatial Distribution Pattern of the Impact of Single Meteorological Hazards in China by Province

4.3.1. Spatial Distribution Pattern of the Impact of Drought by Province

According to the provincial absolute SHCI grades of drought (Figure 7a), northern China experienced greater impacts than southern China from 2004 to 2021, with a spatially contiguous distribution pattern. The regions most adversely affected are the majority of provinces within the Yellow River Basin and Hebei Province. Ningxia stands out, with an absolute SHCI of 0.36. The region that is least impacted is Shanghai, where the absolute SHCI is zero. The impacts in Yunnan and Guangxi are relatively greater than those in other southern provinces, with absolute SHCIs of 0.19 and 0.18, respectively.

Provinces exhibit distinct characteristics in their absolute impact structures (Figure 7b). In certain provinces, the three impact components of drought are relatively similar. In Yunnan, Shaanxi, and Inner Mongolia, each impact accounts for approximately 33% of the total impact. However, in certain provinces, the individual component accounts for a dominant proportion. The economic losses in Tianjin, Shandong, and Guangdong are significantly high, with their proportions exceeding 40%. The population loss in Guangxi is substantial, accounting for nearly 50% of all population loss.

From 2004 to 2021, the distribution of provincial relative SHCI grades for drought closely resembled that of the absolute SHCI. However, notable discrepancies exist in the ratings among provinces. Ningxia experiences the most severe relative impacts, with a relative SHCI of 0.35, the sole province with a relative SHCI exceeding 0.3 (Figure 7c). It is followed by the provinces in the Yellow River Basin and Hebei Province. Shanghai has the smallest relative SHCI, with a value of 0. The relative impact in Yunnan Province is also relatively high compared with that in its neighboring regions.

Figure 7d illustrates the provincial relative impact structure. Most provinces, including Hunan, Sichuan, and Guangdong, exhibit three equivalent impact components of drought. However, an individual component accounts for a dominant proportion in some provinces. The population loss in Hebei, Xinjiang, and Guangxi is relatively high, with that of Guangxi accounting for 48.63% of total population loss. The proportion of economic loss in Tianjin, Chongqing, and Yunnan is substantial, accounting for more than 40% of all economic loss.

The impacts on provinces are classified into five tiers, and the impact grades of drought gradually increase, ranging from green zones to red zones. Provinces in the red and orange zones should optimize water resource allocation and increase their capacity to respond to drought to reduce the impact of drought.

Comparison of the absolute and relative provincial SHCIs of drought (Figure 7e) reveals that most provinces are near the y = x line, implying that the relative and absolute impacts in these regions are analogous. Among them, provinces with relatively low indices, such as Shanghai, Fujian, Zhejiang, Tianjin, and Hainan, should preserve their existing conditions. For provinces exhibiting both significant absolute and relative impacts, such as Ningxia, Hebei, and Inner Mongolia, mitigating the absolute impacts caused by drought is imperative. Moreover, enhancing socioeconomic development is critical for improving drought resilience.

Qinghai, Shandong, Shaanxi, Shanxi, and Guangdong have relatively high absolute drought SHCIs but low relative SHCIs, indicating strong drought resilience. In the future, more efforts should be undertaken to mitigate the absolute impacts of drought. Conversely, the relative SHCIs of drought in Gansu and Yunnan Provinces are greater than their absolute SHCIs of drought, indicating that these regions possess weaker capabilities to withstand drought. Therefore, enhancing water resource storage and scheduling capacities is imperative to better cope with drought.

4.3.2. Spatial Distribution Pattern of the Impact of Rainstorms and Floods by Province

According to the provincial absolute SHCI grades for RF events (Figure 8a), southeastern China experienced greater impacts than northeastern China from 2004 to 2021, while southwestern China was less affected than northwestern China. The regions that were most severely impacted include the middle and lower reaches of the Yangtze River, as well as Sichuan and Shaanxi Provinces. Hunan has the highest absolute SHCI of 0.27, whereas the Beijing–Tianjin–Hebei region has a value of approximately 0.1.

Figure 8b shows the absolute impact structure in each province. In western China, crop losses are the main influencing factor. The proportions of crop loss in Guizhou and Yunnan both exceed 30%. The proportion of household loss in the southeastern region is relatively large and peaks at 35.73% in Anhui. Shanghai is unique in that its housing loss is zero. The economic loss of Northeast China is significant, accounting for nearly 30% of the total economic loss. Nationwide, the proportion of population loss caused by RF events is lower than that caused by other hazards. The average proportion is 21.20%, but the average proportions of impact of the other three hazards all exceed 26%.

From 2004 to 2021, the distribution of the provincial relative SHCI grades of RF events (Figure 8c) is characterized by higher values in the southeast-northwest direction and lower values in the southwest-northeast direction. The provinces that were most severely impacted are Guizhou and Hunan, with relative SHCIs of 0.25. The relative SHCI of Tianjin is 0.08, which is the lowest in the country.

Figure 8d shows the relative impact structure across provinces. An individual impact component constitutes a dominant proportion in most provinces. Numerous provinces exhibit a relatively high proportion of economic loss. Over 10 provinces have percentages greater than 30%, while the mean percentage of all provinces is 28.66%. The proportion of crop loss in western China is greater than that in eastern China. The proportion of population loss in the southeastern region is slightly greater. Notably, housing loss due to RF events in Shanghai is 0.

By integrating absolute and relative impacts, provincial impacts can be categorized into five tiers via a green-to-red gradient, with RF severity increasing sequentially. Provinces in red and orange tiers are advised to upgrade urban drainage infrastructure and increase flood drainage capacities to mitigate RF risk.

Through a comparative analysis of the absolute and relative provincial SHCIs of RF events (Figure 8e), most provinces that are less impacted, including Tianjin, Hainan, Beijing, Henan, Shanghai, and Hebei, are near the y = x line, indicating that the relative and absolute impacts in these regions are commensurate. These regions should sustain their current performance while endeavoring to further mitigate the impacts. Most provinces with greater impacts, such as Hunan, Zhejiang, Sichuan, Qinghai, Fujian, and Jiangxi, lie on the right side of the y = x line, which means that the absolute provincial SHCIs of RF events are greater than the relative values. These findings indicate that these regions exhibit substantial resilience to RF events. Therefore, a focused approach aimed at decreasing the absolute impact is warranted in future endeavors. In Guizhou, Tibet, and Liaoning, the relative provincial SHCIs of RF events are significantly higher than the absolute values, demonstrating their relatively weak capacity to resist RF-induced damage and the stronger relative pressure exerted by RF events on these regions.

4.3.3. Spatial Distribution Pattern of the Impact of Hail and Lightning by Province

According to the provincial absolute SHCI grades for HL (Figure 9a), northern China was more severely impacted than southern China from 2004 to 2021. The regions that were the most severely impacted are the northwestern region, with significant impacts being observed in Beijing and Hubei. Impacts in the southwest, southeast, and northeast regions are moderately severe and evenly distributed nationwide. Shanghai and Hainan have the lowest absolute SHCIs, which are less than 0.1. Gansu has the highest absolute SHCI of 0.26.

Figure 9b shows that the absolute impact structure in each province varies among regions. The proportions of crop loss and housing loss in northeastern China are relatively high. Their sum in Inner Mongolia exceeds 71%. The proportions of economic loss and housing loss in western China are relatively high. Xinjiang and Tibet are particularly representative, with a sum of nearly 60%. The proportion of crop loss in the southwestern region is relatively prominent, accounting for as much as 35.65% of all crop loss in Yunnan. The proportion of economic loss is generally low, ranging from 12.63% to 29.86% nationwide. In contrast, the proportion of household loss predominantly ranges from 21.11% to 43.56%.

From 2004 to 2021, notable disparities existed between the provincial relative and absolute SHCI grades of HL events (Figure 9c). Overall, relative impacts are markedly severe in northern, northwestern, and southeastern regions but relatively minor in central and southwestern regions. The provinces that are the most severely impacted are Xinjiang, Gansu, Ningxia, and Beijing, with values of 0.24, 0.28, 0.23, and 0.24, respectively.

Figure 9d shows the relative impact structure in each province. Most regions have a dominant impact contributor. For example, the proportion of housing loss in the northwestern region is the highest, and the proportion of crop loss in the northeastern and southeastern regions is greater. For individual impacts, Tianjin has the highest proportion of population loss. Xinjiang has the highest proportion of economic loss. Jiangsu has the highest proportion of crop loss. Ningxia has the highest proportion of houses lost across the country. All of the above values exceed 34%.

By combining absolute and relative impacts, provincial impacts are categorized into five hierarchical levels, with the severity of HL events increasing sequentially from green to red. Provinces in the red and orange categories should prioritize strengthening meteorological monitoring and early warning capabilities while enhancing infrastructure and protective measures to alleviate HL-related impacts.

The comparative assessment of the absolute and relative provincial SHCIs reveals that most provinces cluster around the y = x line (Figure 9e), indicating that their relative and absolute impacts are nearly identical. Provinces with relatively low values, such as Shanghai and Hainan, should sustain their current trajectories to preserve the existing low-impact equilibrium. In contrast, for provinces with high absolute and relative impacts, including Gansu, Beijing, and Xinjiang, targeted strategies are needed. They should reduce absolute impacts while simultaneously advancing socioeconomic development to increase tolerance to HL events.

Some provinces, such as Hubei, Inner Mongolia, Shaanxi, Liaoning, and Tianjin, are near the right side of the y = x line. This suggests that within these regions, the absolute impacts of HL events are significantly greater than the relative impacts. These areas demonstrate a relatively high level of resilience to HL events; consequently, more targeted strategies should be formulated to mitigate the absolute impact of HL events. In regions such as Ningxia, Jilin, and Chongqing, the relatively high relative impacts of HL events suggest that these areas have a relatively low capacity to withstand HL events. This highlights the need to prioritize and strengthen social construction in these areas to increase the pressure absorption capacity of HL events.

4.3.4. Spatial Distribution Pattern of the Impact of Typhoons by Province

The distributions of the provincial absolute SHCI grades for typhoons are distinctly aggregated. From 2004 to 2021, a decrease was observed from the southeastern coastal regions to the northwestern inland regions (Figure 10a). The six provinces with the most severe impacts include eastern coastal provinces and inland Jiangxi, with absolute SHCIs ranging from 0.14 to 0.16. The impact in Shanghai is moderate, with an absolute SHCI of 0.10. The impacts in Beijing and Tianjin are lower than those in the surrounding areas. Five provinces in the northwestern inland region have an absolute SHCI of zero, indicating that they have not been affected by typhoons.

Figure 10b illustrates the absolute impact structure in each province. With the exception of Shanghai, the proportion of economic loss in the eastern coastal regions is relatively minimal, whereas the proportions of crop and household losses are relatively pronounced. The proportion of economic loss is the largest in Jiangxi, Guangdong, Guizhou, and Tibet. Some provinces are affected by only certain types of hazards. Inland provinces, including Inner Mongolia, Hebei, Guizhou, and Tibet, experienced no loss of houses. Only the population and economy of Beijing were affected, accounting for approximately 50% of the total population and economic losses.

From 2004 to 2021, the distribution of the provincial relative SHCI grades of typhoons was similar to that of the absolute SHCI (Figure 10c). The provinces that were the most severely impacted are Hainan, Guangdong, Zhejiang, and Anhui. Hainan has the highest relative and absolute SHCI values in the country, with values of 0.15 and 0.16, respectively.

Figure 10d illustrates the relative impact structure across provinces. The four impact components of typhoons are relatively equivalent in most provinces, with the average proportion of each component approaching 25% nationwide. However, some provinces, including Beijing, Hebei, Inner Mongolia, Guizhou, and Tibet, experience only population, economic, and crop losses.

By incorporating both absolute and relative impacts, provincial impacts are categorized into five tiers on a gradient from green to red, with each successive level indicating a progressively greater magnitude of the typhoon-related effect. Provinces classified within the red and orange zones require targeted upgrades to wind-resistant infrastructure to mitigate typhoon-induced impacts.

By contrasting the absolute and relative provincial SHCIs of typhoons (Figure 10e), most provinces are near the y = x line, indicating comparable relative and absolute impacts. Ningxia, Xinjiang, Qinghai, Shaanxi, and Gansu Provinces are spared from the impact of typhoons. In contrast, Hainan and Guangdong Provinces simultaneously presented high absolute and relative provincial SHCIs for typhoons, a pattern strongly linked to their southeastern coastal location.

Some provinces, including Zhejiang, Jiangxi, Fujian, and Shanghai, are near the right side of the y = x line, which means that their absolute impacts are greater than their relative impacts. This suggests that these regions possess relatively strong typhoon tolerance, likely due to robust infrastructure systems or ecological buffering capacities. However, more pronounced absolute impacts highlight the urgent need to prioritize strategies for reducing hazard intensity. Some provinces, such as Anhui and Henan, exhibited marginally higher relative provincial SHCIs of typhoons, indicating a pattern indicative of relatively weaker typhoon resistance. This is likely linked to their inland geographic positioning and socioeconomic structures, where limited typhoon exposure may result in less developed adaptive infrastructure systems. Consequently, targeted investments in resilience-building infrastructure are imperative.

4.3.5. Spatial Distribution Pattern of the Impact of Low-Temperature Freezing by Province

From the perspective of the provincial absolute SHCI grades for LTF (Figure 11a), from 2004 to 2021, the impacts in northern China surpassed those in southern China, and the impacts in western China exceeded those in eastern China. The most severely impacted regions are Xinjiang, Tibet, Yunnan, and Heilongjiang. Tibet is the region most severely affected by LTF, with an absolute SHCI of 0.18, which is three times the lowest value of 0.06 for Shanghai. The impacts in Shaanxi and Henan are also significant, with absolute SHCIs of 0.16 and 0.15, respectively.

Figure 11b shows the absolute impact structure for LTF for each province. The proportions of housing and crop losses in the southwestern and northeastern regions are relatively large. In Henan, the proportions of housing and economic losses are relatively high, at 29.05% and 26.91%, respectively. In the northern region, only Beijing, Tianjin, Shanghai, and Hainan are unaffected by house damage.

From 2004 to 2021, the distribution of the provincial relative SHCI grades of LTF was similar to that of the absolute SHCI (Figure 11c). Overall, relative impacts in Northeast China and Southwest China are the most severe, followed by Northwest China and North China. Among all the provinces, Heilongjiang, Yunnan, and Tibet have the greatest relative SHCIs of LTF, with values of 0.14, 0.17, and 0.20, respectively. The southeastern regions show lower relative impacts, particularly Hainan Province, where the relative SHCI of LTF is only 0.07. Notably, Liaoning Province has a lower relative impact, possibly due to effective protective measures.

Figure 11d illustrates the relative impact structure of LTF in each province. Individual impact components dominate in most regions. The northeastern region (such as Inner Mongolia and Heilongjiang) and southwestern region (including Yunnan and Tibet) accounted for a significant proportion of the total housing losses, all above 30%. Tianjin has substantial population loss, at 47.24%. The crop losses in Hainan and Beijing are notably prominent, reaching 37.05% and 39.45%, respectively. In provinces such as Hebei and Liaoning, the proportions of the four impact components of LTF are generally equivalent, each comprising approximately 25% of the total impact.

By integrating absolute and relative impacts, the provincial impacts of LTF are categorized into five tiers via a gradient from green to red, with each successive level denoting progressively intensified LTF effects. Provinces classified in the red and orange tiers require targeted interventions to mitigate LTF risks, particularly for agricultural protection, livestock cold prevention, and infrastructure resilience.

In the evaluation of the absolute and relative provincial SHCIs of LTF (Figure 11e), many provinces are near the y = x line, indicating that the absolute and relative impacts of LTF are nearly equal within these regions. For provinces with low absolute and relative SHCIs, including Shanghai, Hainan, Liaoning, and Guangdong, maintaining the current status is advisable. In contrast, provinces with high absolute and relative SHCIs, such as Yunnan, Beijing, and Shandong, require a dual-pronged approach. Strategies should be implemented to reduce the absolute impacts of LTF while simultaneously fostering socioeconomic development to increase their capacity to withstand the pressures of LTF.

Provinces such as Heilongjiang, Shaanxi, Henan, and Xinjiang, which are on the right side of the y = x line, exhibit greater absolute impacts relative to the relative impacts. This indicates that although these regions possess a certain degree of tolerance to LTF, targeted efforts are essential to mitigate the substantial absolute impacts associated with challenges from LTF. Conversely, Tibet, Ningxia, Fujian, and Zhejiang, where the relative impacts exceed the absolute impacts, face the pressing issue of weak LTF resistance. Therefore, prioritizing social development initiatives is crucial.

5. Discussion

5.1. Characteristics of the Impacts of Meteorological Hazards in China

From 2004 to 2021, both the national absolute and relative MHCI showed a downward trend overall. The national relative MHCI decreases at a higher rate than the absolute MHCI. The correlation coefficient between them is also greater, which is attributed to the improvement in the fundamental conditions related to meteorological hazards resulting from socioeconomic development. This trend aligns with results from Yin et al. [34], who employed an extended hazard-affected body model to evaluate the annual situation regarding the impact of meteorological hazards in China and concluded that the total impact from 2004 to 2018 showed a downward trend overall. The years of extreme impacts from individual types of meteorological hazards coincided with those in this paper.

However, investigations have reported divergent conclusions. The German nonprofit Germanwatch published the Climate Risk Index (CRI) dataset in 2005 [35]. The CRI index shows an increasing trend in terms of the impacts of meteorological hazards in China. This index is calculated on the basis of six indicators: absolute fatalities, relative fatalities (deaths per 100,000 inhabitants), absolute affectedness, relative affectedness (population affected per 100,000 inhabitants), absolute losses, and relative losses (economic losses as a percentage of GDP). It uses fixed manual weights, ignoring national differences. In contrast, this study adopts the entropy weight method, which derives varying weights for different years and regions based on the data itself, making it more realistic.

The China Climate Change Blue Book shows that China’s climate risk index is increasing [36]. Unlike the CRI introduced by Germanwatch, this index is calculated by weighting the indices of five hazards: droughts, rainstorms, high-temperature conditions, low-temperature freezing, and typhoons. China’s climate risk index findings indicate that there is an overall increase in China’s meteorological hazard risk. Nevertheless, the national MHCIs and SHCIs in this study show a decrease, reflecting enhanced national resilience to meteorological hazards.

Compared with other countries and regions, China has achieved more remarkable outcomes in mitigating the impacts of meteorological hazards. Jiang et al. [37] demonstrated that from 1980 to 2019, the frequency of meteorological hazards, economic losses, and mortalities in the “Belt and Road” region all showed increasing trends. Shen et al. [38] revealed that the rates of occurrence of global natural hazards and the impacts of hazards (such as casualties and property impacts) have all shown increasing trends on the basis of the EM-DAT database.

5.2. Regional Characteristics of the Impacts of Meteorological Hazards

There are substantial disparities in the impacts of meteorological hazards across various regions. The impacts of drought in the semiarid and semi-humid regions are particularly severe, followed by the southwestern region. The impacts resulting from RF events are severe in the Yangtze River Basin and the upper reaches of the Yellow River Basin. This finding parallels the results of the analysis by Hu et al. [12] regarding data from a general survey of meteorological hazard situations in China.

The impact grades of typhoons generally decrease from southeastern coastal areas to northwestern inland regions, which is a pattern closely correlated with geographical and climatic conditions. The regions with relatively severe impacts caused by LTF include several provinces in northwestern, southwestern, northeastern, and central China. Nevertheless, recent studies have shown that the regions most affected by LTF are located mainly in Hubei and Hunan [39], likely because previous studies have evaluated only the impacted area, whereas this work incorporates additional variables such as population and housing.

The ability to withstand meteorological hazards varies across provinces, shaped by factors such as infrastructure and local policies. Provinces with high resilience, like Shanghai, boast well-developed infrastructure: the city has established 1200 automatic weather stations (with a density of 4.6 per 100 square kilometers, five times the national average [40]). Zhejiang has implemented risk zoning management across the province: coastal counties are equipped with “hazard shelters” (1.2 per 10,000 people, compared to the national average of 0.5), and mountainous counties have promoted “automatic early warning devices for flash floods” (with a coverage rate of 95%) [41].

These cases offer references for policymaking and risk management: provinces with low resilience should prioritize addressing shortcomings in infrastructure and policies; those with high resilience can optimize investment structures and promote their experiences. Such differentiated strategies will enhance the overall national resilience to meteorological hazards.

5.3. Regional Characteristics of the Comprehensive Impact Structure of Meteorological Hazards

From the perspective of comprehensive impact assessment, housing losses account for a relatively large proportion in most provinces. In numerous provinces within central and western China, the proportion of housing losses approaches fifty percent. Furthermore, the relative impact of the proportion of housing loss exceeds that of the absolute impact. This indicates that the building structural integrity and hazard resilience capabilities in central and western China are inadequate. In the future, enhancing the capacity of houses to resist damage from multiple meteorological hazards is essential.

The proportion of crop losses in each province is relatively small, suggesting that the impacts caused by damage to agriculture are relatively small. However, the proportions of crop loss in Beijing and Shanghai are relatively large. In terms of the absolute MHCI, the proportion of economic losses in the eastern coastal areas is greater. In the relative MHCI, population loss proportions are relatively large. This shows that the eastern coastal areas are economically developed, with high total social and economic capital, which leads to a decrease in the proportion of relative economic impacts.

5.4. Extreme Meteorological Hazards

At the national scale, extreme meteorological hazards have had prominent socioeconomic impacts in China. In years such as 2010, 2013, and 2016, the absolute MHCI of meteorological hazards in China was notably high, indicating the occurrence of relatively severe meteorological hazard events in China during these years.

In 2010, the majority of southwestern China suffered severe drought [42]. Numerous rivers were inundated, and typhoons ravaged the southeastern coast [43]. In 2013, southern China experienced infrequent high-temperature heatwaves [44]. Northeastern China and Sichuan faced concentrated floods [45]. Southeastern China experienced numerous intense typhoons [46]. Southwestern China experienced recurrent droughts in winter and spring [47]. In 2016, concurrent floods occurred in northern and southern China [48]. Consecutive summer and autumn droughts occurred in the Jianghuai region [49]. Although the absolute MHCI in 2016 was relatively high, the relative MHCI was insignificant, suggesting that, owing to socioeconomic development, the impact remained within an acceptable range and that 2016 did not constitute a year of extreme meteorological hazard.

At the provincial scale, the impacts of multiple meteorological hazards in each province tend to decrease, but severe impacts occurred in specific years. An analysis of the extreme years of meteorological hazard impacts revealed that between 2004 and 2021, central, southwestern, and southeastern China had relatively high frequencies of extreme impacts. Tianjin, Shandong, and Hainan exhibit a considerable incidence of extreme meteorological events. Sichuan has the highest frequency of extreme hazards, with extreme hazards recorded in 2010, 2011, and 2013.

Extreme meteorological hazard events related to climate change have increased the uncertainty of their impacts. Although socioeconomic development can alleviate the impact of extreme events and reduce the overall impacts of meteorological hazards, increasing attention to and the ability to mitigate extreme meteorological hazard events to face potential future climate challenges is imperative.

5.5. Limitations

Several limitations of this study should be noted. First, the research relies primarily on the Yearbook of Meteorological Disasters in China, which might not fully represent the broader geographical area. Potential biases can skew the results. Additionally, the temporal coverage of the data is limited to 2004–2021, which restricts the ability to comprehensively capture longer-term trends. Second, methodological limitations exist within the study design. The indicators selected for constructing the comprehensive index are designated manually, which introduces a certain degree of subjectivity. Additionally, the entropy weight method is employed to determine the indicator weights, adding a certain level of uncertainty to the research outcomes. Future research can be further refined in terms of data acquisition and method selection to yield more reliable conclusions. Despite these limitations, the findings of this study can inform climate adaptation planning and hazard risk reduction strategies.

6. Conclusions

This study advances the quantification of the multidimensional impacts of meteorological hazards in China, including drought, rainstorm and flood, hail and lightning, typhoons, and low-temperature freezing. By introducing a single-hazard composite impact index (SHCI) and a multi-hazard composite impact index (MHCI), this research provides a comprehensive and systematic approach to hazard assessment.

On the basis of data from 2004 to 2021, our findings reveal a discernible downward in the national MHCI. Specifically, the relative MHCI decreased by 74.8% overall, exceeding the 61.21% decrease in the absolute MHCI. Similarly, both the absolute and relative national SHCIs exhibit a decline. When individual hazards were examined, the absolute SHCIs of drought and HL declined at a faster pace than did the national MHCI, whereas those of RF, typhoons, and LTF declined more slowly. In contrast, the rates of decline in all the relative SHCIs are lower than those in the national relative MHCIs.

At both national and provincial scales, there are years with exceptionally prominent impacts of meteorological hazards. Nationally, 2010, 2013, and 2016 exhibited relatively high MHCI values. Excluding these extreme years reduces the average absolute national MHCI by 13.22% and the average relative national MHCI by 8.90%. At the provincial scale, Sichuan experienced three extreme years during this period, the highest number nationwide. Provincially, Sichuan recorded the highest frequency of extreme hazard years, with three extreme hazard years during the study period, surpassing all other provinces.

At the provincial level, both the absolute and relative MHCIs in Ningxia rank the highest. Significant spatial differences exist in the SHCIs of various hazard types. The SHCIs of drought in northern China surpass those in southern China. The SHCIs of RF in the middle and lower reaches of the Yangtze River are relatively high. The SHCIs of HL are relatively high in North and Northwest China. The SHCIs of typhoons exhibit a coastal–inland gradient, decreasing systematically from the southeastern coastal provinces toward the northwestern inland areas. The SHCIs of LTF are the most severe in Northeast China and Southwest China.

From the perspective of the structure of the impacts of meteorological hazards, the influence of infrastructure damage is relatively large in most provinces, but the impact of agricultural losses remains relatively low across each province.

These composite indices provide a replicable model for different hazard assessments in other regions. From a policy perspective, these indices serve as actionable tools for evidence-based decision-making. They enable policymakers to prioritize resource allocation for hazard prevention and mitigation, target investments in climate-resilient infrastructure, and develop region-specific adaptation strategies. Future studies could validate this framework in other regions to test its applicability under diverse climatic and socioeconomic conditions. Additionally, new environmental or institutional variables, such as ecological environment quality and government governance capacity, can be incorporated to make the assessment more comprehensive.