Multi-Dimensional Impacts of Climate Change on China’s Food Security during 2002–2021

Abstract

Climate change poses a significant threat to food security, necessitating a thorough examination across multiple dimensions. Establishing appropriate food security evaluation indicators that align with the evolving concept of food security is imperative. This study enhances food security evaluation by designing a multi-dimensional framework and analyzing the impact of climate variations across various regions from 2002 to 2021. By constructing a food security evaluation system that encompasses the production quantity and quality, sustainability, affordability, and resources, and utilizing the entropy method for accurate weighting, the impacts of climate variations on food security are accessed using a climate–economic model. The food security structure in China largely mirrors the regional division of grain, with the production quantity being the primary contributor. Overall, China’s food security has generally demonstrated improvements across various dimensions, with the exception of production quality. Regarding climate change, which encompasses variations in mean states and climate extremes, the panel regression analysis uncovers a negative linear relationship between food security and temperature. Conversely, the impact of precipitation on food security is non-linear, manifesting as inverse U-shaped patterns. In regions with balanced production and consumption, both accumulated temperatures and extreme high temperatures have a negative linear effect on food security. On the other hand, both accumulated and extreme precipitation exhibit inverse U-shaped non-linear impacts on food security in the main production and main consumption regions. These findings highlight the intricate interplay between climate change, regional disparities, and food security in China, emphasizing the need to consider multi-dimensional factors and regional variations in addressing food security challenges. These insights are invaluable for policymaking and planning aimed at enhancing food security in China.

1. Introduction

Climate change poses a great challenge to global food security and the current changes in global food insecurity can be plausibly attributed to its impact [1,2]. Numerous studies have discussed the impact of climate change on various aspects of food security, including the quality of agricultural products [3] and the impact on prices [4], as well as the impact on agricultural production environments such as farmland [5,6], agricultural water use [7], and agricultural ecosystems [8,9]. In fact, the global food security situation is currently facing unprecedented challenges, with nations experiencing an increase in systemic food security risks [10,11]. According to the 2022 State of Food Security and Nutrition in the World report released by the United Nations Food and Agriculture Organization [12], food security issues still exist widely, especially against the backdrop of the changing climate. Climate change and extreme climate events have become one of the main contributors to the worsening of global hunger, seriously affecting all aspects of food security, including food supplies, access, utilization, and stability.

As the world’s largest producer and consumer of food, China supports nearly 20% of the global population with only 9% of the world’s cultivated land and 6% of its freshwater resources [13,14]. Food security has always been seen as China’s most important issue related to people’s livelihoods, as well as a crucial foundation for economic development, social stability, and national security. However, with rising temperatures, frequent extreme weather events, and changes in precipitation patterns, China’s food production and supply chain are facing increasingly severe threats due to climate change, bringing more uncertainty [15,16,17,18]. The government has implemented comprehensive strategies to ensure the stability and continuous improvement of food security through technological advancement, land protection, subsidies, food reserves, market regulation, agricultural insurance, and international cooperation. Climate in China typically varies significantly across regions due to its vast land area and diverse climate. Therefore, understanding the multi-dimensional impacts of climate change on China’s food security is crucial.

The concept of food security has undergone three significant evolutions internationally [19]. In 1974, the Food and Agriculture Organization of the United Nations (FAO) defined food security as “the ability of all people, at all times, to access enough food for an active and healthy life”. This definition emphasized quantitative security. In 1983, the definition evolved further to include affordability, emphasizing economic security, particularly for low-income groups. At the 1996 World Food Summit, it was stated that “food security exists when all people, at all times, have physical and economic access to sufficient safe and nutritious food that meets their dietary needs and food preferences for an active and healthy life”, highlighting nutrition and health. Across the world, there are mainly two authoritative international organizations for food security evaluation, which are the FAO and the Economist Intelligence Unit (EIU). The FAO’s food security evaluation system mainly focuses on the individual nutritional status [12], constructed from four dimensions: the food supply level, food availability, the food utilization level, and stability. The EIU’s food security index (the Global Food Security Index, GFSI) focuses the categories of affordability, availability, and quality and safety [20], and it holds global significance as a demonstrative tool.

The food security level is impacted by various natural and human factors, and there are many relevant studies. A study on the spatio-temporal evolution and scale differences of food security risk patterns in China [21] has provided a comprehensive analysis of the evolving risk patterns and their implications for policymaking and risk management. Zhang et al. (2022) [22] explored the spatio-temporal evolution and coordination of agricultural green efficiency and food security in China. A quantitative evaluation of the future food security risk considering water scarcity as a key factor was also conducted [23], which highlighted the importance of water resources in ensuring food security and the need for effective water management strategies. The relationship between gross primary production, solar-induced chlorophyll fluorescence, terrestrial water storage, crop grain production, and agricultural investment and policy have been examined [24], providing insights into the complex interactions between these factors and their impacts on food security. Focusing on the interaction between food production security and agricultural ecological protection, Liu et al. (2023) [25] emphasized the need for a balanced approach that considers both food production and ecological sustainability. Work on the impact of low-intensity pollution on China’s sustainable development and food security has highlighted the role of pollution in compromising food security and the need for effective pollution control measures [26]. Yang and Cui (2023) [27] discussed the balance between feed grain security and meat security, examining the trade-offs and synergies between the two. Zhang et al. (2023) [28] assessed the impacts of global climate change on water and food security in the black soil region of Northeast China using an improved SWAT-CO2 model. Their findings provided insights into the potential impacts of climate change on water resources and food production in this region, highlighting the need for adaptive measures to mitigate these impacts.

Based on this, it is crucial to assess global and regional food security to establish appropriate food security evaluation indicators. Heady and Ecker (2013) [29] suggested focusing on four key indicators: calories, poverty, dietary diversity, and subjective indicators. In contrast, Coates (2013) [30] advocated for the utilization of five dimensions: food sufficiency, nutrient adequacy, cultural acceptability, safety, and certainty and stability. Domestic scholars have constructed food security evaluation systems based on China’s national conditions. These systems have evolved from a single focus on the supply–demand balance to emphasizing food availability, especially for low-income groups, and then to focusing on ecological and resource security, as well as dietary nutrition and health. Representative scholars and their research include the following. Zhang et al. (2015) [31] constructed an indicator system covering eight aspects: supply, distribution, consumption, utilization efficiency, security outcomes, stability, sustainability, and regulatory capacity. Wang et al. (2022) [32] established a food security evaluation system from the perspective of food security guarantee capabilities, covering five dimensions: basic guarantee capabilities, market regulation capabilities, production supply capabilities, the utilization of international market resources, and agricultural modernization development capabilities. Gao and Zhao (2023) [33] proposed strengthening the foundation of food security from multiple aspects, such as total quantity security, quality improvement, structural optimization, ecological sustainability, and supply diversification. These studies have provided a useful theoretical foundation for an understanding of the multi-dimensional impacts of climate change on China’s food security.

In the past, most studies on food security systems primarily focused on grain production [27,34,35,36,37], supply and demand [11,38,39,40], and distribution [41], with a strong emphasis on quantitative food security. However, as economic and social development progressed, food security was given new connotations and objectives. In the contemporary era, the concept of food security has broadened to encompass not only ensuring the production capacity but also promoting nutritional, green, diverse, and open food production. This implies safeguarding both quantitative and qualitative food security while maintaining environmental friendliness, economic efficiency in food production, and high resource utilization efficiency. A recent work by Cui and Nie (2019) [42] proposed a comprehensive food security evaluation system encompassing five dimensions: quantitative security, qualitative security, ecological environmental security, economic security, and resource security. Building upon this framework, the present study refines and improves the food security evaluation indicator system, aiming to provide a more holistic and comprehensive assessment of food security in the modern context.

Therefore, based on inter-provincial panel data, this study constructs multi-dimensional food security evaluation indicators, including the production quantity and quality, sustainability, affordability, and resources, to explore the mechanisms and actual impacts of climate variations on food security. It also analyzes the regional differences in the impact of climate variations on food security. The entropy method is employed to determine the weights of the indicators and calculate the $FSI$ for each province in China. A climate–economy model, namely the C-D-C model [43,44], is introduced to assess the multi-dimensional impacts of climate variations on China’s food security. This work will complement the existing evidence of the factors influencing food security, help the Chinese government to formulate more comprehensive and targeted policies and measures, provide useful policy insights to address the complex challenges of climate change regarding China’s food security, and ensure the sustainability and safety of China’s food supply.

2. Materials and Methods

2.1. Data Sources

The climate data used in this study come from the CN05.1 gridded observation dataset developed and released by China’s National Climate Center [45] at the website https://ccrc.iap.ac.cn/resource/detail?id=228 (accessed on 26 April 2023). This dataset is based on daily observation data from more than 2400 national stations (basic, benchmark, and general stations) of the National Meteorological Information Center, and the precise locations of these stations are shown in the work by Wu and Gao (2013) [45]. The variables included are the daily mean temperature, precipitation, maximum temperature, minimum temperature, mean wind speed, relative humidity, and evaporation. In this study, the annual average temperature, annual precipitation, accumulated temperature above 10 $°$C ($Tm10$), extreme high temperature index ($TX90p$, the percentage of days when the daily maximum temperature > 90th percentile), and extreme precipitation index ($R95TOT$, the accumulation of summer precipitation when the daily precipitation ≥ 95th percentile) are calculated using these variables. These variables are listed in

Table 1. Detailed data sources involved in this study.

Category	Variable	Abbreviation	Source
Climate data	Annual mean temperature	$tm$	Calculated using CN05.1 dataset
	Annual mean precipitation	$pr$	Calculated using CN05.1 dataset
	Accumulated temperature above 10 $°$C	$Tm10$	Calculated using CN05.1 dataset
	Annual total precipitation	$Pre$	Calculated using CN05.1 dataset
	Extreme high temperature index	$TX90p$	Calculated using CN05.1 dataset
	Extreme precipitation index	$R95TOT$	Calculated using CN05.1 dataset
Agricultural data	Grain Production Fluctuation Rate	j1	China (Rural) Statistical Yearbook
	Grain Yield per Unit Area	j2	China (Rural) Statistical Yearbook
	Per Capita Grain Production	j3	China (Rural) Statistical Yearbook
	Pesticide Usage per Unit of Cultivated Area	j4	China (Rural) Statistical Yearbook
	Fertilizer Usage per Unit of Cultivated Area	j5	China (Rural) Statistical Yearbook
	Grain Disaster Rate	j6	China (Rural) Statistical Yearbook
	Consumer Price Index for Grain Products	j7	China (Rural) Statistical Yearbook
	Rural Residents’ Engel Coefficient	j8	China (Rural) Statistical Yearbook
	Cultivated Area Used per Unit of Grain Production	j9	China (Rural) Statistical Yearbook
	Water Resources Used per Unit of Grain Production	j10	China (Rural) Statistical Yearbook
Control variables	Cultivated land area	$CLA$	China (Rural) Statistical Yearbook
	Effective irrigation area	$EIA$	China (Rural) Statistical Yearbook
	Total agricultural machinery power	$TAMP$	China (Rural) Statistical Yearbook

.

The provincial agricultural data utilized in this study, also shown in Table 1, consist of two main components. Firstly, the construction of the food security index data encompasses various metrics, in the category of agricultural data in Table 1, including the grain production volatility, grain yield per unit area, per capita grain production, pesticide and fertilizer usage per unit of cultivated land area, proportion of grain affected by disasters, grain consumer price index, Engel’s coefficient for rural residents, cultivated land area per unit of grain production, and water resources used per unit of grain production. Secondly, the agricultural input data, used as control variables in the panel regression analysis, comprise the cultivated land area, effective irrigation area, and total agricultural machinery power. These datasets are sourced from multiple authorities, such as the National Bureau of Statistics, provincial statistics bureaus, the China Economic Data website, the National Intellectual Property Administration, the “China Statistical Yearbook”, the “China Rural Statistical Yearbook”, and provincial statistical yearbooks, primarily accessed through the publicly available website https://data.stats.gov.cn/ (accessed on 25 June 2023). Owing to data unavailability and to ensure the research’s feasibility, data from the Hong Kong, Macau, and Taiwan regions were excluded from this study.

According to China’s grain regional division, 31 provinces can be divided into three regions, as shown in Figure 1. These are the main grain production area (MPA), the main grain consumption area (MCA), and the production–consumption balance area (BA). The MPA refers to the key grain production areas with geographical, soil, climate, technological, and other advantages, which are suitable for the planting of certain grain crops and have certain resource advantages, technological advantages, and economic benefits. They include 13 provinces (Heilongjiang, Jilin, Liaoning, Neimenggu, Hebei, Henan, Shandong, Jiangsu, Anhui, Jiangxi, Hubei, Hunan, and Sichuan). The MCA refers to the grain consumption area, with a relatively developed economy but with a large population and limited land resources, a low self-sufficiency rate of grain, and a large gap between grain production and demand. It is mainly concentrated in the southeastern coastal areas and large cities, including 7 provinces (Beijing, Tianjin, Shanghai, Zhejiang, Fujian, Guangdong, and Hainan). The BA refers to the provinces that have a limited contribution to the national grain production but can basically maintain self-sufficiency, including 11 provinces (Shanxi, Ningxia, Qinghai, Gansu, Xizang, Yunnan, Guizhou, Chongqing, Guangxi, Shaanxi, and Xinjiang).

2.2. Establishment of Food Security Indicator System

When measuring the food security level of each province in China, the entropy method is used to determine the weights of the three-level indicators, as is shown in

Table 2. Food security indicator system involved in this study.

Tier 1 Indicator	Tier 2 Indicator	Tier 3 Indicator	Indicator Direction *
Food Security Index ($FSI$)	Quantitative Security (s1)	j1: Grain Production Fluctuation Rate (%)	Positive
		j2: Grain Yield per Unit Area (kg/hectare)	Positive
		j3: Per Capita Grain Production (kg)	Positive
	Qualitative Security (s2)	j4: Pesticide Usage per Unit of Cultivated Area (kg/hectare)	Negative
		j5: Fertilizer Usage per Unit of Cultivated Area (Pure Quantity, kg/hectare)	Negative
	Ecological Environment Security (s3)	j6: Grain Disaster Rate (%)	Negative
	Economic Security (s4)	j7: Consumer Price Index for Grain Products (Previous Year = 100)	Negative
		j8: Rural Residents’ Engel Coefficient (%)	Negative
	Resource Security (s5)	j9: Cultivated Area Used per Unit of Grain Production (hectare/ton)	Negative
		j10: Water Resources Used per Unit of Grain Production (cubic meters/ton)	Negative

* “Positive” indicates that a higher numerical value of the indicator corresponds to a higher level of food security. “Negative” indicates that a lower numerical value of the indicator corresponds to a higher level of food security.

, and the food security index of each province is calculated accordingly.

2.2.1. Detailed Explanation of the Indicators

Firstly, quantitative security (s1) solves the problem of whether people can “eat enough”, directly reflecting the level of grain supply. It is the foundation and most important indicator of food security. In this work, three indicators are selected as follows, both of which are positive indicators.

• Grain Production Fluctuation Rate (%). It refers to the measure of changes in grain production over a specific period of time, indicating the stability or variability in grain production from year to year.
• Grain Yield per Unit Area (kg/hectare). It measures the grain production efficiency of a region.
• Per Capita Grain Production (kg). It refers to the average amount of grain produced per person in a given region or country.

Secondly, qualitative security (s2) measures the issue of whether people can “eat safely and healthily”. This study selects two indicators, both negative, as follows.

• Pesticide Usage per Unit of Cultivated Area (kg/hectare). It measures the amount of pesticide applied to a specific area of land, reflecting the intensity of pesticide application.
• Fertilizer Usage per Unit of Cultivated Area (pure quantity, kg/hectare). It refers to the quantity of fertilizer applied to a specific land area. Fertilizer is known as the “food” of grain, mainly used to improve the soil fertility and land yield per unit area. However, the excessive application of fertilizer leads to a decrease in soil organic matter, soil hardening, a decrease in land quality, and the pollution of irrigation water.

Thirdly, ecological environment security (s3) examines the sustainability of food production. Food production evaluation should not only consider whether the current food production and quality can meet the needs of contemporary citizens but also investigate the sustainability of the resources. The grain disaster rate is selected as follows.

• Grain Disaster Rate (%). It measures the impact and response capabilities of each province regarding diseases, pests, and climate disasters through the proportion of food crops affected by disasters.

Then, economic security (s4) measures whether people can “afford” to eat. This work selects two negative indicators as follows.

• Consumer Price Index for Grain Products (previous year = 100). It is a measure that tracks changes in the prices of grain-related products consumed by the general population.
• Rural Residents’ Engel Coefficient (%). It refers to the proportion of food expenditure in the total consumption expenditure of rural households. This coefficient provides insights into the living standards and consumption patterns of rural residents.

Lastly, resource security (s5) examines the resource occupation of food production. This study selects two negative indicators as follows.

• Cultivated Area Used per Unit of Grain Production (hectare/ton). It refers to the amount of land required to produce a certain quantity of grain. It is a measure that indicates the efficiency of land use in grain production.
• Water Resources Used per Unit of Grain Production (cubic meters/ton). It refers to the amount of water consumed in the production of a specific quantity of grain. This metric is crucial in assessing the water efficiency of agricultural systems and the sustainability of grain production practices.

2.2.2. Entropy Weight Method

The entropy weight method is a commonly used approach for the evaluation of the effectiveness of multiple attributes or criteria in decision-making problems. It is based on the concept of entropy from information theory, which measures the uncertainty or the amount of information contained in a random variable or a set of data. In the entropy weight method, the entropy value of each evaluation criterion is calculated. This entropy value reflects the degree of dispersion or uncertainty associated with the data for this criterion. The weights of the criteria are then determined based on their entropy values. Criteria with higher entropy values (indicating greater uncertainty or dispersion) are assigned lower weights, while criteria with lower entropy values (indicating less uncertainty or dispersion) are assigned higher weights. The method provides a quantitative measure of the contribution of each criterion to the overall decision result. It is often used in multi-attribute decision-making problems where there are multiple criteria or indicators that need to be considered simultaneously.

In this study, the entropy weight method is performed to determine the weight of each indicator in the food security indicator system in Table 2. For a comprehensive understanding of the implementation process, readers are referred to Appendix A.1, which outlines the detailed steps involved in this methodology. By adhering to these steps, we are able to derive provincial food security index ($FSI$) series that span a range of 0 to 1. A higher $FSI$ score is indicative of a superior level of food security within a given province.

2.3. Panel Data Analysis Method

Panel data analysis is a statistical technique used to analyze data that combine both cross-sectional and time-series dimensions. This type of data, often referred to as “panel data” or “longitudinal data”, involves observations of multiple entities across multiple time periods. They can be used to estimate the effects of time-invariant and time-varying explanatory variables on a dependent variable, taking into account both the cross-sectional and time-series dimensions of the data. The analysis typically involves fitting a regression model to the panel data, where the dependent variable is regressed on a set of explanatory variables. Panel regression analysis has a wide range of applications in various fields, including economics, finance, sociology, and political science. It is particularly useful when studying processes that involve both individual and time-varying factors, such as economic growth, firm performance, or social change.

In this work, a climate–economic model is used to assess the multi-dimensional impacts of climate change on China’s food security. The establishment of this model can be seen in Appendix A.2.

3. Results

3.1. Climate Mean State and Variations

Agriculture is deeply intertwined with the climate. Climate conditions, such as temperature, rainfall, and humidity, significantly impact crop growth, yields, and diversity. Adapting agricultural practices to specific climates is crucial for sustainable food production. Thus, the basic conditions of climate and its changes are presented. The surface air temperature and precipitation over China for the period of 2002–2021 are derived from the CN05.1 observational dataset, as is shown in Figure 2.

For the 20-year mean annual averaged temperature in Figure 2a, it is shown that the temperature in China typically varies significantly across regions due to its vast land area and diverse climatic zones. However, in general, the average annual temperature in China ranges between −5 $°$C and 25 $°$C, with a mean of around 10 $°$C to 15 $°$C. The northern regions, including the northeast, north, and parts of the northwest, generally have cooler annual temperatures, averaging around 0 $°$C to 10 $°$C. These areas experience colder winters and relatively warmer summers. The central and southern regions, including the east, center, south, and southwest, have warmer annual temperatures, averaging between 10 $°$C and 20 $°$C. The mountainous regions, such as the Himalayas in the west and the Qinghai–Tibet Plateau in the northwest, have significantly lower annual temperatures, often below 0 °C, due to their high altitudes. In Figure 2b, it is shown that the most pronounced year-to-year variations occur in the north and northeast regions. Along with global warming, China has seen a general trend of warming over the past 20 years, with the average annual temperatures increasing slightly, as is shown in Figure 2c. However, this trend is not uniform across the country. Some regions, especially in the north, central, and southwest parts, have experienced more pronounced warming than other regions, with linear trends larger than 0.4 $°$C per 10 years.

The climatological mean precipitation is presented in Figure 2d, generally showing that the southern regions of China receive higher levels of average annual precipitation than the northern regions. The southern areas, including the south, east, and parts of the central and southwest regions, experience abundant rainfall, with more than 1000 mm per year. These regions are influenced by the East Asian monsoon, which brings moist air from the oceans, resulting in frequent rainy seasons and abundant precipitation. By contrast, the northern regions, including the north and northwest and parts of the northeast, receive lower average levels of annual precipitation, ranging from 200 to 600 mm per year. These areas have a more continental climate, with dry winters and a limited moisture supply, leading to lower precipitation levels. Figure 2e shows that the year-to-year variations have a similar pattern to the mean precipitation, and the most pronounced variations occur over the southeast. The linear trends of precipitation in Figure 2f show that the regions of Sichuan–Chongqing–Guizhou, Jiangsu–Zhejiang–Shanghai–Anhui, and parts of the northeast have experienced significant increases in annual precipitation, while the regions of Eastern Guangdong and Western Guangdong, Henan, and Yunnan have experienced a decrease in annual precipitation, resulting in higher risks of drought in these regions.

3.2. Food Security Evaluation

To more accurately assess the impact of climate change on China’s food security, we should establish an indicator system that can reflect the level of food security. Based on the indicator system shown in Table 2, the entropy weight method is performed to determine the weight of each indicator in the subsystem. Details of how to implement this method can be seen in Appendix A.1.

Using provincial panel data collected from a public dataset, the food security index ($FSI$) is derived, together with the score series of the five tier 2 indicators of quantitative security (s1), qualitative security (s2), ecological environment security (s3), economic security (s4), and resource security (s5). The respective weights assigned to each individual indicator within the food security indicator system are presented in

Table 3. Weights of indicators in food security indicator system involved in this study.

Tier 2 Indicator	Tier 3 Indicator	Weight
Quantitative Security (s1)	j1: Grain Production Fluctuation Rate	2.54%
	j2: Grain Yield per Unit Area	19.15%
	j3: Per Capita Grain Production	41.25%
Qualitative Security (s2)	j4: Pesticide Usage per Unit of Cultivated Area	5.64%
	j5: Fertilizer Usage per Unit of Cultivated Area	7.51%
Ecological Environment Security (s3)	j6: Grain Disaster Rate	3.78%
Economic Security (s4)	j7: Consumer Price Index for Grain Products	3.08%
	j8: Rural Residents’ Engel Coefficient	7.67%
Resource Security (s5)	j9: Cultivated Area Used per Unit of Grain Production	6.09%
	j10: Water Resources Used per Unit of Grain Production	3.30%

.

Utilizing the available provincial panel data and the food security indicator system outlined in Table 2, it becomes apparent that quantitative security (s1) has a substantial weight of approximately 63% in determining the overall $FSI$. This significant influence primarily originates from the key indicator of j3: per capita grain production (kg). Following closely is qualitative security (s2) with a weight of around 13%, while economic security (s4) and resource security (s5) contribute with weights of approximately 11% and 9%, respectively. Lastly, ecological environment security (s3) accounts for the least, weighing in at around 4%. This distribution of weights highlights the significant influence of various factors on food security and their relative importance within the indicator system.

The fluctuations in the national $FSI$ and its associated indicators are depicted in Figure 3. From a holistic perspective, China’s food security exhibited a fluctuating upward trend over this period as per the $FSI$. To delve deeper, the tier 2 indicators in Table 2, which compose the $FSI$, provide further insights into these variations. Specifically, quantitative security (s1), ecological environment security (s3), economic security (s4), and resource security (s5)—contributing 62.9%, 3.8%, 10.8%, and 9.4% to the $FSI$, respectively—all exhibit strong positive correlations with the $FSI$ (correlation coefficients of 0.98, 0.94, 0.94, and 0.93, respectively, all exceeding the 99% confidence level), reflecting a similarly fluctuating growth pattern during the specified period. In contrast, qualitative security (s2) demonstrates a negative correlation (−0.62, surpassing the 99% confidence threshold), suggesting a declining trend throughout the same period.

Similar work is conducted by combining panel data on the food security indicators in 31 provincial administrative regions of China. Figure 4 introduces a boxplot to show the diversity of $FSI$ in the spatial and temporal dimensions. In the spatial dimension, as Figure 4a illustrates, when the provinces are sorted based on their median food security index values over the past 20 years, five provinces (Heilongjiang, Jilin, Neimenggu, Liaoning, Xinjiang) emerge as having a higher level of food security, with median values exceeding 0.50. Generally, provinces categorized as the MPA tend to obtain higher scores, whereas provinces categorized as the MCA have relatively lower scores. The lowest level of food security occurs for Hainan province in the MCA, with a median value of less than 0.30. In terms of mean scores, the top 13 provinces are all in the MPA, which is consistent with the national grain production’s regional division. The 14th to 25th places include 11 provinces, in which six provinces (Xinjiang, Chongqing, Yunnan, Guangxi, Shanxi, Ningxia) are BA areas, and five provinces (Shanghai, Zhejiang, Tianjin, Beijing, Guangdong) are MCA areas. The remaining seven provinces with the lowest average grain scores are Fujian, Shaanxi, Guizhou, Xizang, Gansu, Hainan, and Qinghai. Among them, five provinces (Shaanxi, Guizhou, Tibet, Gansu, Qinghai) are BA areas, and two provinces (Fujian, Hainan) are MCA areas. It can be seen that the food security of the BA areas has already received a warning regarding food security. From a temporal perspective, as is shown in Figure 4b, the provincial food security level generally shows a continuous upward trend. The median value of the 31 provinces has risen from below 0.4 to nearly 0.5.

To quantitatively evaluate the provincial food security level, a Taylor diagram is introduced in Figure 5, which can provide a visual framework for the comparison of the provincial $FSI$ to the national mean $FSI$. The radial distance from the origin represents the fraction of provincial $FSI$ variations and national mean $FSI$ variation. The temporal correlation coefficient between the provincial $FSI$ and national mean $FSI$ is denoted by the angular distance from the x-axis.

As is evident from this figure, there are notable regional disparities in the food security index among provinces across the country. Provinces categorized as the MPA typically demonstrate stronger correlations with the national mean food security level, often exceeding 0.90. In contrast, provinces in the MCA exhibit weaker correlations, suggesting a more limited contribution to the national mean food security level. Provinces in the BA occupy a median position in terms of their food security indices. From the perspective of variance, compared to the national mean $FSI$, the standard deviation of $FSI$ for most provinces falls within the range of 0.5 to 1.5. There are four provinces (Heilongjiang, Neimenggu, Jilin, Xinjiang) with standard deviations above 1.5, indicating larger interannual fluctuations. On the other hand, three provinces (Fujian, Guangdong, Zhejiang) have standard deviations below 0.5, indicating smaller interannual variations in their $FSI$. From the perspective of correlations, the interannual variations in the majority of the provinces are relatively consistent with that of the national mean $FSI$. There are 28 provinces with correlation coefficients exceeding 0.7, with the highest exceeding 0.99 (Neimenggu). However, there are three provinces (Guangdong, Hainan, Beijing, all in the MCA) with correlation coefficients below 0.7, indicating that their variations differ significantly from those of other provinces.

3.3. Empirical Analysis of the Impact of Climate Variations on China’s Food Security

On the basis of the understanding of China’s climate conditions and changes over the past 20 years, combined with the newly constructed provincial-level food security index in China, this section employs panel regression analysis methods to investigate the multi-dimensional impacts of climate variations on China’s food security during the period of 2002–2021.

Climate change is a significant and growing threat to food security, already affecting vulnerable populations across the world [46]. In the context of climate change, both climate extremes and the climate mean are important considerations. Climate change is causing shifts in both the mean and extreme values of weather variables. For example, global warming is leading to higher average temperatures and more frequent and intense extreme heat events. Similarly, changes in precipitation patterns are resulting in both wetter and dryer regions, with more intense rainfall events in some areas and decreased rainfall in others. It is important to consider both climate extremes and the climate mean when assessing the impacts of climate variations and developing adaptation strategies. Understanding how these two aspects of the climate are changing can help us to prepare for and mitigate the potential negative impacts of climate change on human society and the natural environment. Thus, two groups of climate variables are analyzed.

3.3.1. Impact of Climate Mean State Variations on $FSI$

Climate change refers to long-term shifts in the Earth’s climate system, including temperature, precipitation, wind patterns, and other weather variables. In this work, the variables of the accumulated temperature above 10 $°$C ($Tm10$) and yearly accumulated precipitation ($Pre$) are used to represent the climate mean state change for the period of 2002–2021 and are designated as the explanatory variables. The provincial food security index ($FSI$) is designated as the dependent variable. Moreover, the cultivated land area ($CLA$), the effective irrigation area ($EIA$), and the total agricultural machinery power ($TAMP$) are selected as the control variables. Both entity-specific fixed effects and time fixed effects are incorporated into the model.

Firstly, employing a panel regression analysis on the provincial dataset, we confined our attention to the linear effects by considering only the first-order terms of the explanatory variables. This was achieved using the panel data analysis model (Equation (A11)) detailed in Appendix A.2. The outcomes of this specific analysis are displayed in

Table 4. Linear benchmark regression model results quantifying the impact of climate mean state variations on food security across China.

Variables	s1	s2	s3	s4	s5	$\mathit{F}\mathit{S}\mathit{I}$
$Tm10$	−0.553 ***	0.137	−0.272	0.418 ***	−0.175 *	−0.188 ***
$Pre$	0.063 *	0.024	0.061	−0.022	0.010	0.024
$CLA$	0.242 ***	0.478 ***	0.114	0.002	−0.066 *	0.176 ***
$EIA$	0.261 ***	−0.267 ***	−0.020	−0.016	0.187 ***	0.087 ***
$TAMP$	−0.134 ***	−0.079 ***	−0.004	0.206 ***	−0.001	−0.026 *
Constant	−0.811	−4.957 ***	−2.620	−7.471 ***	−2.159 ***	−1.399 ***
Entity FE	Yes	Yes	Yes	Yes	Yes	Yes
Time FE	Yes	Yes	Yes	Yes	Yes	Yes
R2	0.652	0.381	0.366	0.859	0.298	0.758
Observation	620	620	620	620	620	620

All variables take the logarithm; *** p < 0.01, * p < 0.1.

.

From the presented table, one can discern that, for the food security index ($FSI$), the coefficient of $Tm10$ stands at −0.188, a value that is statistically significant at the p < 0.01 level. This implies a negative linear relationship between $Tm10$ and $FSI$, suggesting that as $Tm10$ rises, there is a corresponding decrease in the $FSI$. On the contrary, no statistically significant association exists between $Pre$ and $FSI$. Regarding the control variables, both $CLA$ and $EIA$ exert significant positive influences on food security. The model’s R-squared value amounts to 0.758, signifying that our model effectively accounts for 75.8% of the variability in the dependent variable $FSI$. This is considered a relatively high R-squared value, indicating that our model offers a good fit to the data, with substantial explanatory power regarding fluctuations in China’s provincial food security levels. Regarding the sub-indicators, it is evident that both quantitative security (s1) and economic security (s4) demonstrate strong correlations with variations in $Tm10$, where the significance level surpasses p < 0.01. This signifies a detrimental effect on quantitative security but a favorable impact on economic security due to changes in $Tm10$. Additionally, resource security (s5) exhibits a negative relationship with $Tm10$ at a significance level of p < 0.1, suggesting a decrease as $Tm10$ increases. However, no statistically significant effects are observed between $Tm10$ and qualitative security (s2) or ecological environment security (s3).

Subsequently, to account for the non-linear effects of precipitation and temperature on food security, quadratic terms for the explanatory variables $Pre$ and $Tm10$ are incorporated into the model. A refined panel regression analysis is performed using a modified version based on Equation (A13), detailed in Appendix A.2. Post-regression, individual U-tests were carried out to examine each variable’s non-linearity. The results of this advanced analysis are showcased in

Table 5. Non-linear benchmark regression model results quantifying the impact of climate mean state variations on food security across China.

Variables	s1	s2	s3	s4	s5	$\mathit{F}\mathit{S}\mathit{I}$
$Tm10$	−0.512 ***	0.138	−0.201	0.414 ***	−0.152 *	−0.166 ***
$Pre$	2.397 ***	0.088	4.076 ***	−0.224	1.342 ***	1.321 ***
${\left(Pre\right)}^2$	−0.173 ***	−0.005	−0.297 ***	0.015	−0.099 ***	−0.096 ***
$CLA$	0.208 ***	0.478 ***	0.056	0.005	−0.085 **	0.157 ***
$EIA$	0.253 ***	−0.267 ***	−0.035	−0.016	0.182 ***	0.082 ***
$TAMP$	−0.116 ***	−0.078 ***	0.027	0.204 ***	0.009	−0.016
Constant	−8.755 ***	−5.174 ***	−16.288 ***	−6.781 ***	−6.693 ***	−5.814 ***
Entity FE	Yes	Yes	Yes	Yes	Yes	Yes
Time FE	Yes	Yes	Yes	Yes	Yes	Yes
R2	0.671	0.381	0.408	0.859	0.319	0.777
Observation	620	620	620	620	620	620

All variables take the logarithm; *** p < 0.01, ** p < 0.05, * p < 0.1.

. The findings reveal that while there is no statistically significant U-shaped relationship between $Tm10$ and food security, an inverse U-shaped relationship between $Pre$ and food security has been established as significant. This leads to the conclusion that the effect of temperature ($Tm10$) on food security is linear, whereas the influence of precipitation ($Pre$) exhibits a non-linear pattern. Consequently, only the first-order term of temperature is retained in the model, while both the first-order and second-order terms for precipitation are maintained to reflect its non-linear impact. This indicates that when the annual precipitation falls within a certain optimal range, its increase can indeed contribute positively to ensuring food security. However, beyond a particular threshold, any further rise in precipitation will inversely affect food security levels, leading to a decline. This inverse U-shaped relationship between precipitation and food security is consistently mirrored in the quantitative security (s1), ecological environment security (s3), and resource security (s5) sub-indicators, suggesting that the effects of precipitation on these dimensions follow a similar pattern of initially improving and then potentially deteriorating food security conditions under excessive levels.

3.3.2. Impact of Climate Extremes on $FSI$

Climate extremes are the occurrence of unusually high or low values of weather variables, such as temperature or precipitation, that fall outside the range of normal variations expected in a given region. These extreme events can include extreme heat waves, droughts, floods, and other severe weather conditions. Climate extremes can have significant impacts on human health, agriculture, infrastructure, and ecosystems.

In a similar vein to the examination of mean climate state changes ($Pre$ for precipitation and $Tm10$ for temperature), we use two variables that embody climate extremes: $TX90p$, which represents the percentage of days when the daily maximum temperature exceeds the 90th percentile, and $R95TOT$, denoting the accumulated precipitation on days when the daily precipitation is equal to or above the 95th percentile. These variables encapsulate the climate extreme data for the period between 2002 and 2021 and are thus designated as the explanatory variables in our subsequent analysis. The fixed effect configuration, as well as the inclusion of other variables—including the control variables and the constant term—remain consistent across the analyses.

In the context of solely examining the linear effects of the explanatory variables, we utilized the regression model constructed in accordance with Equation (A12), detailed in Appendix A.2. The outcomes of this regression analysis are showcased in

Table 6. Linear benchmark regression model results quantifying the impact of climate extremes variations on food security across China.

Variables	s1	s2	s3	s4	s5	$\mathit{F}\mathit{S}\mathit{I}$
$TX90p$	−0.069 ***	−0.028	−0.155 ***	0.020	−0.054 ***	−0.041 ***
$R95TOT$	0.017	0.002	−0.068 *	−0.024	−0.015	−0.005
$CLA$	0.221 ***	0.484 ***	0.102	0.017	−0.072 **	0.169 ***
$EIA$	0.250 ***	−0.265 ***	−0.020	−0.007	0.183 ***	0.083 ***
$TAMP$	−0.137 ***	−0.079 ***	−0.015	0.205 ***	−0.004	−0.028 *
Constant	−4.674 ***	−3.660 ***	−3.566 ***	−4.231 ***	−3.246 ***	−2.595 ***
Entity FE	Yes	Yes	Yes	Yes	Yes	Yes
Time FE	Yes	Yes	Yes	Yes	Yes	Yes
R2	0.642	0.381	0.379	0.853	0.304	0.758
Observation	620	620	620	620	620	620

All variables take the logarithm; *** p < 0.01, ** p < 0.05, * p < 0.1.

. As evidenced by the table, there is a highly statistically significant (p < 0.01) negative association between extreme high temperatures ($TX90p$) and food security, indicating that an increase in intense heat events results in a reduction in food security levels. However, no discernible linear correlation exists between extreme precipitation ($R95TOT$) and food security. Upon closer inspection of the tier 2 indicators, it becomes apparent that $TX90p$ consistently exerts detrimental impacts on quantitative security (s1), ecological environment security (s3), and resource security (s5). This uniform manifestation of an adverse influence suggests that the deleterious effect of extreme heat on food security is channeled through its impact on these specific sub-indicators.

In the context of examining the non-linear effects encapsulated by the quadratic terms of the explanatory variables, we utilized a regression model derived from Equation (A14), detailed in Appendix A.2. The results of this analysis are showcased in

Table 7. Non-linear benchmark regression model results quantifying the impact of climate extreme variations on food security across China.

Variables	s1	s2	s3	s4	s5	$\mathit{F}\mathit{S}\mathit{I}$
$TX90p$	0.482 **	0.022	−0.140	−0.421 **	0.016	0.045
${\left(TX90p\right)}^2$	−0.126 **	−0.011	−0.005	0.101 ***	−0.016	−0.020
$R95TOT$	1.054 ***	0.062	2.319 ***	−0.160	0.730 ***	0.631 ***
${\left(R95TOT\right)}^2$	−0.086 ***	−0.005	−0.197 ***	0.012	−0.062 ***	−0.053 ***
$CLA$	0.197 ***	0.482 ***	0.073	0.028	−0.083 **	0.160 ***
$EIA$	0.247 ***	−0.265 ***	−0.042	−0.012	0.177 ***	0.078 ***
$TAMP$	−0.137 ***	−0.080 ***	−0.003	0.209 ***	−0.000	−0.026 *
Constant	−8.135 ***	−3.877 ***	−10.416 ***	−3.448 ***	−5.442 ***	−4.493 ***
Entity FE	Yes	Yes	Yes	Yes	Yes	Yes
Time FE	Yes	Yes	Yes	Yes	Yes	Yes
R2	0.658	0.382	0.428	0.855	0.325	0.773
Observation	620	620	620	620	620	620

All variables take the logarithm; *** p < 0.01, ** p < 0.05, * p < 0.1.

. Our research findings reveal that there is no statistically significant non-linear correlation between $TX90p$ and overall food security. However, it is noteworthy that the sub-indicators quantitative security (s1) and economic security (s4) exhibit opposing U-shaped trends. On the other hand, an inverted U-shaped relationship with statistical significance was observed between $R95TOT$ and food security, which successfully underwent validation via the U-test. This particular relationship is primarily underpinned by the inverse U-shape associations existing between $R95TOT$ and the following tier 2 indicators: quantitative security (s1), ecological environment security (s3), and resource security (s5).

3.4. Regional Differences in the Impact of Climate Variations on $FSI$

In order to delve deeper into the regional disparities in the effects of climate variations on food security, this study introduces three regional dummy variables: $D_1$ is set to 1 for provinces within the MPA region and 0 for those outside it; similarly, $D_2$ assumes a value of 1 for provinces in the BA region and 0 for others; and $D_3$ takes the value of 1 for provinces in the MCA region and 0 otherwise. We proceed with a panel regression analysis using provincial panel data based on the panel data analysis model presented in Equations (A15) and (A16) within Appendix A.2. Throughout all analyses, the fixed effect settings are maintained consistently, as is the inclusion of other variables, such as control variables and the constant term. This approach allows us to discern how the impacts of climate variations, including the climate mean and climate extremes, vary across different regions with respect to food security.

3.4.1. Impacts of Climate Mean State Variations

We initially explore the impacts of climate mean state variations, as represented by $Tm10$ (accumulated temperature above 10 $°$C) and $Pre$ (yearly total precipitation). The outcomes of this investigation are presented in

Table 8. Linear benchmark regression model results for the heterogeneity analysis of the impact of variations in climate mean states on food security, stratified into three groups of provinces based on the regional grain division.

Variables	FSI in MPA	FSI in BA	FSI in MCA
$Tm10$	−0.155	−0.114 **	0.024
$Pre$	0.042 *	0.111 ***	−0.024
$CLA$	0.169 ***	0.125 ***	0.121 **
$EIA$	0.285 ***	0.004	0.010
$TAMP$	−0.029	−0.063 ***	−0.071 *
Constant	−3.348 ***	−1.443 **	−1.503
Entity FE	Yes	Yes	Yes
Time FE	Yes	Yes	Yes
R2	0.889	0.811	0.615
Observations	260	220	140

All variables take the logarithm; *** p < 0.01, ** p < 0.05, * p < 0.1.

. These results reveal that a significant negative linear relationship between $Tm10$ and the food security index ($FSI$) prevails, predominantly in the BA region, with a p-value less than 0.05. Moreover, the analysis shows a positive linear correlation between $Pre$ and $FSI$, which is statistically significant at the p < 0.01 level in the BA region and marginally significant at the p < 0.1 level in the MPA region.

In contrast to a linear analysis, the investigation incorporates non-linear effects by including quadratic terms of the explanatory variables.

Table 9. Non-linear benchmark regression model results for the heterogeneity analysis of the impact of variations in climate mean states on food security, stratified into three groups of provinces based on the regional grain division.

Variables	FSI in MPA	FSI in BA	FSI in MCA
$Tm10$	3.321	−1.233	5.616
($Tm10$)2	−0.208	0.078	−0.330
$Pre$	1.866 ***	0.242	1.424 ***
($Pre$)2	−0.135 ***	−0.010	−0.104 ***
$CLA$	0.123 ***	0.128 ***	0.129 **
$EIA$	0.283 ***	−0.000	−0.004
$TAMP$	−0.025	−0.059 ***	−0.038
Constant	−23.615 **	2.056	−30.319
Entity FE	Yes	Yes	Yes
Time FE	Yes	Yes	Yes
Observations	260	220	140
R2	0.905	0.813	0.646

All variables take the logarithm; *** p < 0.01, ** p < 0.05.

showcases the outcomes of this non-linear examination. It is noteworthy that no significant non-linear relationship between $Tm10$ and the food security index ($FSI$) can be observed across all three designated regions. However, regarding precipitation ($Pre$), a distinct non-linear pattern emerges: there is a statistically significant inverse U-shaped association between $Pre$ and food security in both the MPA and MCA regions. This implies that, beyond certain thresholds, an increase in precipitation does not proportionally enhance food security in these areas; instead, it may lead to diminishing returns or even negative impacts on food security.

3.4.2. Impact of Climate Extremes

Similarly, we performed regression analyses on the climate extreme changes, represented by $TX90p$ (extreme high temperatures) and $R95TOT$ (extreme precipitation), examining both linear and non-linear effects. The outcomes of these analyses are presented in

Table 10. Linear benchmark regression model results for the heterogeneity analysis of the impact of variations in climate extremes on food security, stratified into three groups of provinces based on the regional grain division.

Variables	FSI in MPA	FSI in BA	FSI in MCA
$TX90p$	−0.017	−0.062 ***	−0.019
$R95TOT$	0.019	0.028	−0.031
$CLA$	0.172 ***	0.096 **	0.123 **
$EIA$	0.281 ***	−0.003	0.012
$TAMP$	−0.027	−0.073 ***	−0.073 *
Constant	−4.447 ***	−1.334 ***	−1.251 ***
Entity FE	Yes	Yes	Yes
Time FE	Yes	Yes	Yes
R2	0.888	0.815	0.620
Observations	260	220	140

All variables take the logarithm; *** p < 0.01, ** p < 0.05, * p < 0.1.

for linear effects and

Table 11. Non-linear benchmark regression model results for the heterogeneity analysis of the impact of variations in climate extremes on food security, stratified into three groups of provinces based on the regional grain division.

Variables	FSI in MPA	FSI in BA	FSI in MCA
$TX90p$	−0.002	−0.066 ***	−0.029
$R95TOT$	1.359 ***	0.233 *	0.981 ***
($R95TOT$)2	−0.108 ***	−0.018	−0.080 ***
$CLA$	0.118 ***	0.097 **	0.135 ***
$EIA$	0.274 ***	−0.009	0.022
$TAMP$	−0.024	−0.072 ***	−0.067 *
Constant	−8.098 ***	−1.860 ***	−4.571 ***
Entity FE	Yes	Yes	Yes
Time FE	Yes	Yes	Yes
Observations	260	220	140
R2	0.905	0.817	0.655

All variables take the logarithm; *** p < 0.01, ** p < 0.05, * p < 0.1.

for non-linear effects. In terms of extreme temperatures, a consistent negative linear relationship with food security is observed across all three regions, although this correlation reaches statistical significance at the p < 0.01 level only within the BA region. On the other hand, the influence of extreme precipitation exhibits an inverse U-shaped relationship with food security in all three regions. Notably, this non-linear relationship surpasses the p < 0.01 significance threshold in both the MPA and MCA regions, indicating that there is a threshold beyond which increased extreme precipitation negatively impacts food security in these areas.

4. Discussion

4.1. Strengths and Limitations in the Food Security System

The assessment of food security is a multifaceted and intricate process, greatly influenced by regional disparities and diverse policy frameworks [47]. Given the complexity of the subject, distinct indicator systems have emerged, reflecting individual research priorities [48]. Consequently, when evaluating the food security status of a nation like China, these varying indicator systems may yield differing conclusions [17,31,42,49]. This underscores the importance of adopting a holistic and comprehensive approach that incorporates various perspectives and metrics to arrive at a more accurate and comprehensive understanding of food security.

This study builds upon the work of Cui and Nie (2019) [42], updating its indicators and introducing a comprehensive framework for food security analysis. Encompassing five critical dimensions, namely quantitative security, qualitative security, ecological environmental security, economic security, and resource security, this framework aims to provide a more thorough and holistic evaluation of China’s food security status. By building upon and refining existing indicators, it strives to offer a more integrated understanding of food security in the context of contemporary challenges and dynamics, serving as a scientific basis for the formulation of effective strategies and policies.

Over time, there has been a general strengthening of China’s food security, exhibiting a similar trend to previous works [31,49]. Positive correlations are evident in the quantitative, ecological environmental, economic, and resource security aspects. However, the decline in qualitative security, indicated by its negative correlation, underscores the urgent need for continued improvements in grain production efficiency and the strict enforcement of cultivated land protection policies. Core strategies should focus on advancing agricultural technologies, prudent land use management, and rigorous measures against overexploitation to ensure sustainable food production and resilience against challenges.

Nevertheless, the food security system still has a few limitations. Due to the limited availability of data, certain aspects of food security may not be fully captured or accurately assessed. Therefore, it is crucial to continue exploring and refining the indicator system to provide a more comprehensive understanding of China’s food security status.

4.2. Risks of the Food Security Structure in China

The food security structure in China generally aligns with the regional grain division depicted in Figure 1. Provinces categorized as the MPA typically exhibit a higher level of food security, along with a stronger correlation with the national mean level, indicating a significant contribution to the overall national food security. Conversely, provinces in the MCA tend to have a lower food security level, with a weaker correlation and reduced variance. Provinces in the BA occupy a median position in terms of food security indices.

However, a closer analysis reveals that among the 31 provinces examined, Qinghai, Yunnan, Guizhou, Guangxi, Gansu, and Shaanxi, categorized as balanced production–consumption regions, exhibit relatively low food security levels, as is shown in Figure 4 as well as Figure 5. This signifies heightened food insecurity and poses significant challenges to national food security. Characterized by less fertile land and underdeveloped infrastructure, these regions are prime targets for agricultural improvement initiatives.

In response to the escalating food security risks in these balanced regions, urgent and comprehensive action is imperative. Measures such as enhancing agricultural productivity, protecting arable land, augmenting grain reserves, optimizing logistics, establishing robust emergency plans, promoting climate-smart agriculture, encouraging moderate-scale farming, and implementing targeted support measures are crucial. These strategic steps are indispensable in maintaining stable grain supplies and strengthening the national food security amidst the growing threats posed by climate change, dwindling land resources, and volatile market conditions.

4.3. Increasing Climate Risks to Food Security

Climate change, encompassing both long-term climate trends and yearly variations, is marked by rising temperatures and a surge in extreme heat events, posing considerable threats to China’s food security. This aligns with the conclusions drawn by Su et al. (2022) [49]. In contrast, the effect of precipitation on food production exhibits a U-shaped trend, where, initially, increased precipitation benefits agricultural production but it eventually becomes detrimental with excessive levels, consistent with the observations made by Su et al. (2022) [49] and Lee et al. (2024) [17]. Given the increasing climate risks to food security, it is imperative to adopt relevant adaptation strategies.

Firstly, rising temperatures degrade farmland’s production environment, slowing grain growth, intensifying agricultural pests and diseases, and increasing the severity of meteorological disasters. The imbalance in the irrigation water supply and demand also poses a challenge. To address these issues, we must enhance the climate monitoring and early warning systems, strengthen the infrastructure, promote technological innovation, optimize the crop structure and varieties, cultivate stress-resistant varieties, and leverage digital agriculture for intelligent analysis, monitoring, management, and precision farming.

Secondly, the agricultural precipitation demand varies greatly by region and season in China. The north, a major grain-producing area, faces limited precipitation and severe droughts. Conversely, the south experiences concentrated seasonal rains, often leading to flooding. This creates a serious imbalance in the irrigation water supply and demand. To tackle this, we must strengthen the agricultural water conservancy facilities and the ecological environment, optimize the irrigation management systems, and bolster the agricultural sector’s resilience to climate risks. Initiatives like high-standard farmland construction, reservoir reinforcement, small and medium-sized river management, and the promotion of low-carbon water and fertilizer management models are crucial to enhancing the agricultural water use efficiency.

Thirdly, grain-producing regions vary in their safeguarding needs. Regions with superior technological and resource conditions are less impacted by climate change. Hence, stabilizing their sowing areas, enhancing the yields per unit area, and safeguarding food security are priorities. Conversely, balanced production and consumption regions face land scarcity, poorer agricultural conditions, and greater climate vulnerability. Therefore, policy and financial support are crucial. Investments in agricultural infrastructure, irrigation facilities, and production equipment upgrades are vital to ensure grain self-sufficiency in these regions.

Lastly, the adaptation of local food security strategies to climate change requires comprehensive actions. Assessing the regional climate risks, diversifying the agricultural outputs, optimizing water allocation, upgrading agricultural facilities, promoting resilient farming practices, strengthening community resilience, and fostering stakeholder collaboration are all essential. Preserving the red line for cultivated land, improving the soil quality, and supporting moderate-scale farming are fundamental to enhancing the agricultural stability and resilience in a changing climate.

5. Conclusions

To conduct a comprehensive quantitative assessment of China’s food security, this study devised a provincial food security index ($FSI$) system consisting of the production quantity and quality, sustainability, affordability, and resources. The entropy method was employed to assign weights and compute the $FSI$ for each province. Leveraging a climate–economic model, the investigation scrutinized the multifaceted impacts of climate variations on China’s food security landscape over the 2002–2021 period. In this evaluation system, the production quantity holds a significant share (63%) in China’s overall food security, which generally shows improvements across various dimensions. However, production quality exhibited a decreasing trend. Although the provincial food security levels have collectively improved, several provinces designated as balanced production–consumption areas are notably threatened. It is crucial to address these concerns to ensure sustainable food security in China.

China’s diverse climates are experiencing increasing temperatures and variable precipitation patterns. Notably, the northern, central, and southwestern regions are observing significant warming trends. While the southern provinces receive greater annual rainfall due to monsoons, the southeastern regions experience interannual fluctuations that increase the risk of drought. Statistical regression analyses revealed a robust negative linear relationship between higher temperatures and food security. In contrast, precipitation influences food security in a non-linear manner, exhibiting an inverse U-shaped curve that becomes significant when exceeding certain thresholds, especially in major production and consumption areas.

Extreme high temperatures consistently reduce food security across multiple indicators, whereas no such linear correlation is observed for extreme precipitation. However, there is a statistically significant inverse U-shaped association between extreme precipitation and food security. Specifically, in the production–consumption balanced areas, a higher temperature ($Tm10$) has a significantly negative impact on food security, contrasting sharply with the positive correlation found between the yearly total precipitation ($Pre$) and food security. This relationship appears more pronounced in balanced areas compared to production areas. Furthermore, the non-linear analyses reveal that, above certain thresholds, increasing precipitation can lead to diminishing returns or even adverse effects on food security.