An Evaluation of Food Security and Grain Production Trends in the Arid Region of Northwest China (2000–2035)

Abstract

Food security is crucial for social stability and economic development. Ensuring food security in the arid region of Northwest China presents unique challenges due to limited water and soil resources. This study addresses these challenges by integrating a comprehensive water and soil resource matching assessment with grain production forecasting. Based on data from 2000 to 2020, this research projects the food security status to 2035 using the GM(1,1) model, incorporating a comprehensive index of soil and water resource matching and regression analysis to inform production forecasts. Key assumptions include continued historical trends in population growth, urbanization, and dietary shifts towards an increased animal protein consumption. The findings revealed a consistent upward trend in grain production from 2000 to 2020, with an average annual growth rate of 3.5%. Corn and wheat emerged as the dominant grain crops. Certain provinces demonstrated comparative advantages for specific crops like rice and wheat. The most significant finding is that despite the projected growth in the total grain output by 2035 compared to 2020, the regional grain self-sufficiency rate is projected to range from 79.6% to 84.1%, falling below critical food security benchmarks set by the FAO and China. This projected shortfall carries significant implications, underscoring a serious challenge to regional food security and highlighting the region’s increasing vulnerability to external food supply fluctuations. The findings strongly signal that current trends are insufficient and necessitate urgent and proactive policy interventions. To address this, practical policy recommendations include promoting water-saving technologies, enhancing regional cooperation, and strategically utilizing the international grain trade to ensure regional food security.

1. Introduction

Food security, ensuring stable access to sufficient, safe, and nutritious food, remains a cornerstone of global stability and development [1,2]. As the world’s most populous nation and a major player in global agriculture, China’s ability to feed itself carries profound domestic and international weight. Fluctuations in China’s food supply and demand balance can influence global agricultural markets and food price volatility [3], highlighting the interconnectedness of national food systems in a globalized world. Domestically, China faces a prolonged challenge in balancing grain production and consumption, driven by industrialization and urbanization, a population projected to peak around 2030, rising incomes, and a significant dietary transition [4]. This transition involves an increased demand for meat, eggs, and milk, shifting the grain demand structure towards feed grains like corn and soybeans, adding complexity to agricultural planning and food security management [5,6], a pattern observed in many rapidly developing economies globally as incomes rise [7].

These national trends place particular pressure on regions like Northwest China. Despite its vast land area, this region is characterized by an arid to semi-arid climate, with an annual precipitation often below 200 mm and high evaporation rates, making water resources extremely scarce—possessing only a fraction of the national average per unit area [8]. Such water scarcity is a defining challenge for agriculture in many arid and semi-arid zones across the globe, from the Middle East and North Africa (MENA) to Central Asia and parts of Australia and the Americas [9,10]. Nevertheless, Northwest China serves as a critical grain production base, historically contributing significantly to the national total from about 15% of the nation’s grain cultivation area. This highlights its strategic importance for national food security and the immense pressure on its limited water resources. This reliance on a water-scarce region is increasingly precarious.

The inherent vulnerability of agriculture in Northwest China is significantly amplified by climate change. Rising temperatures increase evapotranspiration, further straining water supplies, while changing precipitation patterns can lead to more frequent and severe extreme events such as droughts and heatwaves [11,12]. For this arid region, this translates into a heightened risk of prolonged droughts, which can devastate crop yields, alongside potential disruptions from heatwaves or even intense rainfall events causing localized flooding and soil erosion. These climate change impacts on water availability and extreme weather events significantly threaten agricultural systems in vulnerable arid regions worldwide, demanding urgent adaptation strategies [13]. Ensuring future food security here necessitates navigating limited water and land resources, evolving consumption demands, and the growing threat posed by these climate change-induced pressures, which directly challenge the sustainability of current agricultural practices [14,15]. This study, therefore, investigates the interplay of these factors, evaluating historical trends and projecting the future food security under these complex conditions.

While previous research has provided valuable insights into food security in Northwest China, a significant gap exists in studies that holistically integrate the crucial interplay between water and soil resource availability and agricultural demand, particularly under the dual pressures of climate change and evolving consumption patterns. Much of the existing work focuses on individual factors, such as water scarcity or land use change, often lacking a comprehensive metric to assess resource matching in the context of future food security projections. This study addresses these gaps by first developing and applying a comprehensive index that evaluates the dynamic matching between water and soil resources (considering both availability and agricultural demand, including environmental water requirements implicitly via gray water), offering a more integrated assessment of resource pressure. Second, unlike studies potentially relying on a single demand estimation method, we employ both quota and consumption statistics methods to provide a more robust and nuanced analysis of the supply–demand balance. Furthermore, a lack of integrated long-term projections represents another critical gap. This research fills this void by forecasting the food security status specifically to 2035, constrained by projected resource matching, offering crucial insights for proactive, long-term policy planning and sustainable agricultural strategies in this vulnerable region, thus contributing a forward-looking perspective beyond previous analyses.

Using the northwest arid region as a case study, this paper aims to (1) examine the changing characteristics of the grain output and grain crop planting structure from 2000 to 2020, (2) define the balanced relationship between the grain supply and consumption capacity using multiple methods, (3) evaluate the matching degree between water and soil resources, and (4) project the level of food security (via self-sufficiency rate) by 2035 based on resource-constrained production forecasts and demand projections. Given Northwest China’s strategic importance for the national food supply and the parallels between its environmental challenges and those faced by arid regions globally, the insights generated by this research are pertinent to national policy formulation in China and broader international discussions on sustainable agriculture and food security under resource constraints.

2. Materials and Methods

This section details the methodology used to evaluate food security in the arid regions of Northwest China, focusing on historical trends from 2000 to 2020 and projecting outcomes to 2035. The specific empirical goals are to (1) quantify historical (2000–2020) grain production and consumption trends, (2) calculate historical self-sufficiency rates (SSRs) using two methods to assess the supply–demand balance, (3) compute a water–soil resource matching index to evaluate the water resource pressure relative to the agricultural demand, and (4) forecast these indicators to 2035 using GM(1,1) modeling for index and demand components and regression to link the index to the production potential, ultimately projecting the future SSR (Figure 1).

2.1. Study Area

The arid region of Northwest China covers 3.45 million km², representing 35.9% of China’s total land area [16]. It is abundant in light, heat, and soil resources and is a crucial reserve base for China’s grain production and cultivated land resources [17]. Situated inland and on the periphery of the monsoons, the annual precipitation in this region is less than 200 mm (only 47% of the national average), with the annual evaporation exceeding 1000 mm, resulting in water resources per unit area accounting for only 1/6 of the national average [18]. Furthermore, the northwest arid region features complex terrain and landforms, such as mountains, plateaus, basins, plains, and diverse ecosystems, including grassland, farmland, deserts, and forests [19]. Due to challenging natural environmental conditions and slow economic development, among other adverse factors leading to a severe mismatch between soil and water resources, agricultural development has been significantly impeded in this region [20]. This region holds a prominent position in national agricultural policy, as it is essential for ensuring food security in China. Additionally, it confronts severe water scarcity and climate change, which necessitate effective strategies for sustainable grain production despite limited resources. This study aims to develop strategies that align with regional agricultural policies by understanding the agricultural characteristics and issues within this context.

2.2. Calculation of Key Indices and Forecasting Methods

To comprehensively assess food security and its underlying resource constraints, several key indices were calculated following a logical sequence: first quantifying the grain consumption demand (the denominator for self-sufficiency), then estimating the associated land and water resource requirements for that consumption (for context, though not directly used in the SSR forecast), next evaluating the degree of matching between available water resources and the agricultural water demand (the core resource constraint metric), and finally using these elements for forecasting.

(1)

Calculation of Grain Consumption Demand. The calculation of the grain consumption demand usually includes quota statistics and consumption statistics. Calculating the grain consumption demand is a necessary step for determining the food self-sufficiency rate (SSR), which was chosen as the key metric for the food security assessment in this study because it provides a widely recognized benchmark of a region’s ability to meet its needs from domestic production, directly reflecting the supply–demand balance.

The quota statistics method is widely used for estimating consumption demands, as it offers a straightforward calculation based on the product of per capita annual consumption quotas and the total population. This method provides a standardized benchmark for demand based on policy targets or planning norms.

$$D=P\times d$$

(1)

where D is the total grain demand (t), P is the total population, and d is the per capita grain demand. For the calculation of the food self-sufficiency rate, d is based on the data provided by the Outline of the National Medium- and Long-Term Program for Food Security (2008–2020), 400 kg/(person ·year); for the calculation of the self-sufficiency rate of cereals, 360 kg/(person ·year) is used according to the proportion of cereals in the grain output.

The assumptions behind the replanting index (CI_i) and national average yields (FY_j) should be explicitly addressed. The CI_i values reflect static historical averages and do not account for potential fluctuations due to advancements in agricultural practices or climate variability. As such, while these figures provide a necessary reference, their static nature may limit their applicability in the face of ongoing agricultural changes. Future projections would be strengthened by incorporating a dynamic approach considering these evolving factors.

The method of consumption statistics offers a more detailed approach to estimating the grain demand, considering all relevant factors comprehensively despite the intricate calculation process. This method breaks down demand into specific end-uses. Grain consumption is classified into five categories: direct consumption, feed consumption, industrial consumption, seed use (replacing “loss during breeding” for clarity), and losses during post-harvest processes (e.g., transportation and storage; replacing “transportation storage processes”). Direct consumption refers to grain consumed directly by residents; feed consumption refers to grain consumed indirectly in meat, poultry, eggs, and milk, which residents subsequently consume. Calculating feed consumption relies on specific assumptions regarding the feed composition and conversion efficiency: By referencing feeding standards for cattle, sheep, pigs, chickens, and other livestock [21,22,23,24], we derived the structure of the grain feed (

Table 1. Proportion of raw materials for livestock and poultry feed.

Livestock and Poultry Type	Wheat (%)	Maize (%)	Bean Cake (%)	Millet (%)	Potato (%)
Cattle	8	69	8	8	8
Sheep	13	60	13	7	7
Pig	10	65	10	5	10
Cow	8	62	15	8	8
Chicken	10	60	10	10	10

) and combined it with the feed conversion rate to calculate the indirect consumption of various grains.

The grain conversion rates (representing the ratio of grain input required per unit of livestock product output) were determined based on Xu et al. [25]. These conversion rates were vetted against the region’s current agricultural practices and standards to ensure their relevance and accuracy. Specifically, the conversion ratio used for beef and mutton was 4.1:1, indicating that producing 1 unit of beef or mutton requires an average feed grain input of 4.1 units. Similarly, the ratios applied were pork 2.9:1, poultry 2.0:1, and eggs 1.8:1. Industrial grain is utilized in food production as raw materials or auxiliary materials for food processing and starch production. Trends in industrial grain consumption were projected using a combination of historical data analysis and growth rate estimation, assuming past growth trends and distributions would largely continue. Current agricultural policies and market demand considerations were considered to derive these estimates. By analyzing recent trends and historical growth patterns, we used statistical techniques such as moving averages and compound annual growth rates (CAGR) to accurately project the future industrial grain consumption. This transparent methodology ensures that the present trends reflect past behaviors and anticipated future conditions of grain consumption in the industrial sector.

According to China’s Food Administration statistics, industrial grain consumption has increased yearly over the past decade. In 2006, 2008, 2010, and 2012, the industrial consumption of grain in China reached 68 million tons, 73.5 million tons, 93 million tons, and 101.3 million tons, respectively. The average annual growth rate of China’s industrial grain consumption is estimated to be 6.8%. The distribution of rice, wheat, corn, potato, and soybean in industrial grains is approximately 14.04, 15.2, 45.79, 7.40, and 14.40 percent, respectively. The annual industrial consumption for each type of grain was calculated accordingly.

Furthermore, seed use accounts for the grain needed for sowing the next crop, and a certain proportion of ineffective loss occurs during post-harvest stages (replacing “sowing, harvesting, transportation, storage, processing, eating, and other stages”) to ensure food reproduction and reserve a certain amount for seed grain purposes. According to the relevant literature, the total amount lost or retained accounted for approximately 5 percent of the year’s output [21]. The demand for grain consumption was calculated by summing up the five types of grains mentioned above.

$$G_d=P_u·d_u+P_r·d_r$$

(2)

$$G_c=P_u{\sum }_{i=1}^9{U}_i{a}_i+P_r{\sum }_{i=1}^9{R}_i{a}_i$$

(3)

$$D=G_d+G_i+G_{in}+G_{seed}+G_{loss}$$

(4)

where G_d is the direct consumption (10⁴ t), G_c is the food used in feed (10⁴ t); G_in is the amount of grain used in industry; G_seed is the amount of pre-stored seeds; G_loss is the loss (10⁴ t); P_u and P_r are the total populations of urban and rural areas; d_u and d_r are the per capita direct annual consumption of urban and rural residents, respectively; and U_i and R_i are the amounts of the i type of food consumed per capita by urban and rural residents. i = 1 (pork), 2 (beef), 3 (mutton), 4 (animal oil), 5 (aquatic products), 6 (poultry meat), 7 (poultry eggs), 8 (fresh milk), and 9 (wine); and a represents the grain conversion ratio.

(2)

Estimation of Water and Land Resource Requirements for Consumption. Having estimated the grain demand, the next step is to quantify the land and water resources hypothetically needed to meet this demand within the region, providing insight into the resource footprint of consumption.

Employing the ecological footprint methodology, each province’s total arable land area required to produce consumed agricultural products is calculated based on the current per capita consumption of agricultural products and the national average yield of various agricultural products. The ecological footprint calculation assumes that all agricultural products consumed within a province are produced using the average national yield. This simplification allows for a standardized comparison but does not account for regional variations in actual agricultural practices, technology, or environmental conditions, which may affect actual land requirements. The specific calculation method is as follows:

$${TAD}_{arable i}=\left(\left({\sum }_{j=1}^n{\displaystyle \frac{{TFD}_{ij}}{{FY}_j}}\right)/{CI}_i\right)$$

(5)

where TAD_{arable i} is the total demand for arable land area in province i (10⁴ ha); CI_i is the multiple cropping index (also referred to as the replanting index) of province i, representing the average number of cropping cycles per year on a given unit of arable land (dimensionless). This index reflects the land use intensity, accounting for planting multiple crops sequentially on the same land within a single agricultural year. The multiple cropping index of each province was sourced from Xie Hualin [26]; FY_j is the national average yield of j agricultural products (kg/ha); n is the number of agricultural products, including 11 types of crops and 6 types of livestock products. The arable land equivalent for livestock products was calculated following Xie Honghong [27], which only calculates the amount of arable land used. The use of static values and national average yields are key assumptions. These national averages and historical indices do not capture regional yield variations or dynamic changes in cropping intensity, limiting the precision of the footprint estimate.

From the perspective of natural resources and the environment, sustainable agricultural water consumption encompasses water utilized for agricultural production and water employed to purify agricultural pollutants, ensuring a consistent agricultural output and a pristine environmental setting. The former is quantified by virtual water in agriculture, while the gray water footprint measures the latter. The specific calculation method for the virtual water consumption related to grain and other agricultural product consumption is as follows:

$${TWD}_i={\sum }_{j=1}^n{AW}_i+{PW}_i$$

(6)

where TWD_i is the total water demand for sustainable agricultural production (10⁸ m³) in province i; AW_ij is the virtual water consumption (10⁸ m³) of the j type of agricultural products produced within province i, obtained by multiplying the total demand of each type of agricultural product by its virtual water consumption coefficient; and PW_i refers to the water used for purifying agricultural pollutants in province i’s agricultural production (10⁸ m³).

A key assumption in the virtual water calculation is using a uniform virtual water content for each agricultural product, often derived from national or broader regional averages. This neglects variations in water use efficiency and irrigation practices across different local production areas within Northwest China. Furthermore, accurately estimating the gray water footprint (PW_i) associated with consumption is complex and relies on assumptions about production methods and pollution assimilation capacities.

(3)

The Calculation of the Water and Soil Resource Matching Index. To evaluate the balance between the actual water resource availability and current agricultural water demand within the region, a matching index (I) was calculated. This index helps identify areas where the current agricultural water use is potentially straining available resources. This index was chosen explicitly for analysis because it quantitatively links the agricultural water demand (driven by land use and cropping structures) to the availability of water resources, which is recognized as the most critical limiting factor for sustainable agriculture and food security in the severely arid environment of Northwest China.

A matching index greater than 1 indicates relatively abundant available water resources compared to the current agricultural water demand, while a value less than 1 suggests that the current agricultural water demand exceeds the readily available water resources.

$$I={\displaystyle \frac{AWA}{AWR}}$$

(7)

where I is the matching index between agricultural water and soil resources in the region; AWA is the available agricultural water resources in the region, measured in cubic meters; and AWR is the agricultural water demand in the region, also measured in cubic meters.

The available agricultural water resources encompass both regional blue and green water resources.

$$AWA=AW{A}_{blue}+{AWA}_{green}$$

(8)

where AWA_blue is the available blue water resources for agriculture in the region, in cubic meters (m³), and AWA_green is the green water resources for agriculture in the region, in cubic meters (m³).

The region’s available agricultural blue water resources are determined through a combination of local blue water resources, transboundary water flow, and the proportion of agricultural water use to the total water consumption. Locally available water resources encompass the non-environmental flow in the regional surface runoff and groundwater. After accounting for environmental flow requirements, the calculation assumes a fixed coefficient (α) represents the usable portion of surface water.

$$AW{A}_{blue}=\left((R\times \alpha +U_p)+W_{transit}\right)\times {\displaystyle \frac{AWU}{WU}}$$

(9)

where AWA_blue is the available blue water resources for regional agriculture, in cubic meters (m³); R is the river flow, in cubic meters (m³); α is the coefficient of available surface water resources, which was assumed to be 40% based on common regional practices, implying 60% is reserved for environmental flows; U_p is the precipitation infiltration recharge volume, in cubic meters (m³); W_transit is the transit water volume, in cubic meters (m³); AWU represents the total water consumption in the region, in cubic meters (m³); and WU is the agricultural water consumption in the region, in cubic meters (m³).

The region’s green water resources reflect the utilization of soil water formed by precipitation within the current crop planting structure (i.e., effective rainfall consumed by crops). They constitute an integral component of the available agricultural water resources. The calculation of regional green water resources is based on effective precipitation and crop evapotranspiration, as expressed by regional AWA_green.

$${AWA}_{green}=10\times {\sum }_{i=1}^n\mathit{min}(E{T}_{c,i},P_{e,i})\times A_i$$

(10)

where AWA_green is the green water resources for regional agriculture (m³); ET_c,i represents the evapotranspiration during the growing period of crop i in millimeters (mm); P_i represents the effective precipitation during the growing period of crop i (mm); and A_i represents the sown area of crop i (ha).

Crops evapotranspiration is calculated according to the single crop coefficient method.

$$E{T}_c=K_c\times E{T}_0$$

(11)

where ET is the crop evapotranspiration (mm); K_c is the crop coefficient (dimensionless), representing the ratio of the specific crop’s evapotranspiration to the reference evapotranspiration under non-stressed conditions; and ET₀ is the reference evapotranspiration of the reference crop (mm). The reference crop evapotranspiration is calculated according to the Penman formula. The Penman formula is widely recognized for its applicability in estimating the reference evapotranspiration (ET₀) across diverse climatic conditions, including arid regions. Its ability to account for both the energy balance and aerodynamic factors makes it particularly suitable for the northwest arid region of China, where high evaporation rates and limited water resources are prevalent. Studies have shown that the Penman formula provides reliable estimates of ET₀ in arid and semi-arid climates, making it a robust choice for this study.

The overall water requirement for agriculture in the area, encompassing both irrigated and rain-fed farming, is ascertained by calculating the crop evapotranspiration and the corresponding cultivated acreage of crops.

$$AWR=10\times {\sum }_{i=1}^{n}E{T}_{c,i}\times A_i$$

(12)

where AWR is the water demand for regional agriculture (m³).

(4)

GM(1,1) Gray Prediction Model for Forecasting. To project future trends (specifically the soil–water matching index and per capita grain consumption needed for the 2035 food security assessment) based on the limited historical data available (2000–2020), the GM(1,1) model was employed. The GM(1,1) model was chosen for its ability to handle short time series and uncertainty, which are common characteristics of agricultural data in arid regions. Unlike traditional statistical models that often require large datasets and specific data distributions, the GM(1,1) model can generate reliable predictions with limited data points by identifying underlying exponential trends.

The application of the GM(1,1) model involved the following specific steps: (a) Data Preparation: selecting the historical time-series data (e.g., matching index 2000–2016 for forecasting); (b) Accumulation: generating the first-order accumulated sequence x⁽¹⁾; (c) Parameter Estimation: estimating parameters ‘a’ and ‘u’ via least squares using the background value sequence; (d) Model Construction: establishing the prediction equation for x⁽¹⁾; (e) Forecasting: calculating future x⁽¹⁾ values (to 2035); (f) Inverse Accumulation: obtaining forecast values x⁽⁰⁾ for the original sequence; and (g) Validation: assessing accuracy by comparing forecasts against actual data for a hold-out period (e.g., 2017–2020). The GM(1,1) model within the gray system represents a first-order differential function of a variable over time, with its corresponding differential equation being:

$${\displaystyle \frac{{dx}^{\left(1\right)}}{dt}}+{ax}^{\left(1\right)}=u$$

(13)

where x⁽¹⁾ is a sequence generated through one-time accumulation, t denotes time, and a and u are the parameters to be estimated, known as the development gray number and endogenous control gray number, respectively. Acknowledging the model’s primary limitation is crucial: it excels at capturing monotonous trends. However, it may be less accurate if future conditions involve significant non-linear changes, cyclical patterns, or abrupt shifts caused by policy interventions or external shocks not reflected in the historical data series used for model calibration.

(5)

Statistical Trend Analysis. To statistically evaluate the significance and direction of trends in time-series data, such as the grain consumption over the 2000–2020 period, the Mann–Kendall (M-K) non-parametric test was employed. This test is widely used for detecting monotonic trends in environmental and hydrological time series without requiring assumptions about data distribution. The M-K test assesses whether the data have a consistent upward or downward trend. The test statistic S is calculated as follows:

$$S={\sum }_{i=1}^{n-1}{\sum }_{j=i+1}^n\left(\mathrm{s}\mathrm{g}\mathrm{n}(x_j-x_i)\right)$$

(14)

$$\mathrm{s}\mathrm{g}\mathrm{n}\left(x\right)=\left\{\begin{array}{c}1\\ 0\\ -1\end{array} \right. \begin{array}{c}if x>0\\ if x=0\\ if x<0\end{array}$$

(15)

$$Z=\left\{\begin{array}{c}{\displaystyle \frac{S-1}{\sqrt{Var\left(S\right)}}}\\ 0\\ {\displaystyle \frac{S+1}{\sqrt{Var\left(S\right)}}}\end{array} \right. \begin{array}{c}if S>0\\ if S=0\\ if S<0\end{array}$$

(16)

where n is the number of data points in the time series (e.g., years from 2000 to 2020); x_i,x_j are the data values (e.g., grain consumption) at times i and j, respectively, where j > i; and Z is the standardized statistic, where a positive value indicates an increasing trend and a negative value indicates a decreasing trend. The significance is tested by comparing Z to critical values from the standard normal distribution.

(6)

Consumption Structure Analysis. The changes in the grain consumption structure over the study period (discussed in Section 3.3.2) were analyzed by calculating the percentage contribution of each component (direct food, feed grain via different animal products, industrial, seed, and loss) to the total consumption derived from the consumption statistics method. The illustrative water footprint values mentioned in the discussion of dietary shifts are based on the established literature values and are used to highlight the resource implications of changing diets.

2.3. Data Source

The data on the sowing area, total output, national average yield, per capita consumption (for demand calculation and footprint), nitrogen fertilizer application, arable land resources area, and total water resources (surface and groundwater) of various crops in each province (autonomous region) within the study area were primarily sourced from the “China Statistical Yearbook” (relevant editions covering data for the years 2000–2020, https://www.stats.gov.cn/sj/ndsj/, accessed on 5 October 2024). Crop-specific unit sowing amount data are obtained from the “Compilation of National Agricultural Cost-benefit Data”. Data on industrial water use, domestic water use, and ecological environment supplementary water use in each province (autonomous region) are extracted from the “China Water Resources Statistical Yearbook”.

3. Results

3.1. Grain Production

3.1.1. Characteristics of Regional Production Change

The grain production in the arid northwest region exhibited a fluctuating but strong increasing trend from 2000 to 2020, as depicted in Figure 2a. Throughout this timeframe, the region achieved substantial growth, with the grain production’s average annual growth rate at 1.93 × 10⁶ t, with an average compound annual growth rate of 3.5%. The years of increased grain production accounted for 81.0% of the study period, indicating consistent progress despite fluctuations. The grain production escalated from 4.16 × 10⁷ t in 2000 to 8.21 × 10⁷ t in 2020, representing a remarkable overall growth rate of 97.4%. The average annual production stood at 6.21 × 10⁷ t, with the peak occurring in 2020 at approximately one and a half times higher than the yearly average output. Conversely, the lowest production occurred in 2001 at around seventy percent of the mean value. The year with the maximum grain production during the study period was twice that of the year with the minimum output, highlighting a significant inter-annual variability.

(1)

Low growth stage from 2000 to 2007. During this period, the overall grain production in the arid northwest region was relatively low, averaging 42.4 million tons with significant fluctuations and an average annual growth rate of only 220,000 tons, which is significantly lower than the overall average growth rate of 1.93 million tons for the entire study period (2000–2020). The main issues that affected the grain production during this stage were as follows: China reduced the planting area of grain crops during the agricultural restructuring process to prioritize the cultivation of higher-value cash crops, such as cotton, fruits, and vegetables. The advancement of urbanization, along with the implementation of the Western Development Strategy, which prioritized infrastructure development and industrial growth in the region, led to the conversion of agricultural land for non-agricultural uses. The “South-to-North Water Diversion” water resources protection project, aimed at transferring water from the Yangtze River basin to the drier north, also placed restrictions on water use for agriculture in some areas of the northwest; as well, the distortion of long-term grain prices resulted in farmers having a negative attitude towards grain production. This directly impacted the grain output in the region.

(2)

Rapid growth stage from 2008 to 2015. A significant acceleration occurred following the initial low growth, driven by policy support and technological advancements. Supported by a series of agricultural policies in China, such as the “No. 1 Central Document”, which has focused on agricultural development and rural issues annually since 2004, and increased subsidies for grain production, advancements in high-tech agriculture have been consistently observed. The widespread implementation of efficient water-saving and fertilization techniques has resulted in a substantial increase in land productivity. This transition has also facilitated the shift from extensive to sustainable green agriculture in the arid regions of Northwest China. During this period, grain production surged from 55.4 million tons to 80.2 million tons, with an average annual output of 68.5 million tons and an average annual growth rate of 3.1 million tons. This growth rate was dramatically higher, 14.23 times that of the initial stage (2000–2007) and 3.3 times that of the subsequent stage (2016–2020), marking this as the primary period of the production increase.

(3)

The period of fluctuating growth from 2016 to 2020. The rapid growth phase moderated into a slower, more variable growth period. The reduction in the area of high-yield grain cultivation as a result of agricultural structural adjustment measures, such as “grain-to-feed” and “grain-to-oil” policies, aimed at increasing the production of feed crops and oilseeds to meet the growing demand for meat and edible oils, coupled with frequent droughts, resulted in a declining trend in the grain production in the northwest drought-prone areas from 2015 to 2017. However, there was a gradual recovery after that. During this period, grain production decreased from 7.67 × 10⁷ to 7.51 × 10⁷ t, then increased to 8.21 × 10⁷ t. The average annual production was 7.88 × 10⁷ t, with an average annual growth rate of 1.1 × 10⁶ t. This growth rate, while positive, was significantly lower than the rapid growth stage, reflecting the challenges of balancing agricultural structural adjustments with food security goals.

3.1.2. Grain Output per Unit Area

The grain output and planting area per unit of land directly determines the level of grain production in a region. The impact of the grain yield per unit area on regional grain production is often more significant than that of the sown area. Examining the spatiotemporal variation in the grain yield per unit area in northwest arid regions can reflect differences in agricultural production technology and cultivated land productivity among regions and crops, thereby elucidating the production potential of each food crop to ensure regional food security. From 2000 to 2020, the yield per unit area of five grain crops in the northwestern arid region exhibited a fluctuating increase (Figure 2b). During the study period, the highest yield was observed in rice at 7.5 t/ha, followed by corn, wheat, potatoes, and beans with 5.8, 3.8, 3.1, and 1.6 t/ha yields, respectively. These crops’ average annual growth rates were recorded as 0.6%, 1.7%, 2.5%, 1.4% and 2.8%, respectively. In Xinjiang, the grain yield per unit area peaked at 6.1 t/ha, which is significantly higher than that of the northwest arid area by 1.5 times. The same unit yield level also surpassed that of Ningxia by an impressive margin of approximately 4.3 percent above the regional average. Comparatively speaking, Xinjiang and Ningxia exhibit a distinct advantage in grain yield per unit area over Shaanxi, Gansu, Inner Mongolia, and Qinghai. While other provinces (autonomous regions) reported lower grain yields per unit area than the regional average, it is evident that moderately increasing the grain planting areas in Xinjiang and Ningxia would be conducive to boosting the overall grain production and ensuring better regional food security. The increase in the grain yield per unit area was noted as follows: Inner Mongolia at 3.6%, Ningxia at 3.3%, Xinjiang at 1.4%, Gansu at 3.2%, and Shaanxi at 2.2%.

Several factors contribute to these regional differences in grain yield. Xinjiang’s high yield is attributable to several key factors. First, more extensive and efficient irrigation systems, including the widespread adoption of drip irrigation and sprinkler systems, optimize water use in this arid environment. Second, the region benefits from relatively fertile soils in certain areas, particularly the oases along the Tarim and Junggar basins. These favorable soil conditions and advanced irrigation contribute significantly to Xinjiang’s higher yields. Third, the substantial investment in agricultural technologies, including improved crop varieties and mechanization, has boosted productivity. Finally, the longer growing season and abundant sunlight hours in parts of Xinjiang contribute to higher yields. Ningxia also benefits from advanced irrigation techniques, particularly along the Yellow River, which has a long history of irrigated agriculture and government support for agricultural modernization. Furthermore, both Xinjiang and Ningxia have relatively better access to markets than some other regional provinces, facilitating the distribution and sale of their agricultural products. While facing challenges of water scarcity and soil degradation in some areas, Gansu has seen improvements in yield due to the promotion of drought-resistant crop varieties and water-saving practices.

The differences in productivity are also aligned with specific crops. For instance, Xinjiang’s high yields are partly driven by its specialization in high-yielding maize and wheat varieties, while Ningxia’s rice production contributes to its higher-than-average yields. Gansu’s focus on drought-resistant crops like potatoes and minor grains reflects its adaptation to the region’s challenging environmental conditions. These differences in the grain yield per unit area highlight the importance of tailored agricultural strategies that consider regional variations in climate, resources, and technological adoption. Further investments in region-specific agricultural research, development, and policies promoting sustainable water management and soil conservation will be crucial for enhancing grain productivity and ensuring food security in the arid northwest region.

3.2. Planting Structure of Grain

3.2.1. Cultivated Area

The fluctuating trends in grain-sown areas within the northwest arid region from 2000 to 2020 are depicted in Figure 3a. Throughout the research period, the average annual grain-sown area stood at 1.44 × 10⁷ ha, with a majority of 61.9% representing an increase in the grain-sown area. The changes observed in the grain-sown area within the northwest arid region primarily exhibited a pattern of an initial decrease followed by a subsequent increase, reaching its lowest point of 1.20 × 10⁷ ha in 2003. Since 2004, the Chinese government has consistently issued Document No.1 focusing on “Agriculture, rural areas and farmers,” thereby boosting farmers’ enthusiasm for cultivating grains through measures such as exempting agricultural taxes, establishing and enhancing an agricultural subsidy system represented by the “four subsidies,” implementing a minimum grain purchase price system, and enforcing policies for grain purchase and storage. These policies had a measurable impact on farmers’ incentives. For instance, the introduction of direct grain subsidies in 2004 and the minimum purchase price contributed to a recovery in the national grain-sown area, which increased by approximately 10.5% between 2003 and 2015. This national trend was also reflected in the northwest region, although the specific rate of increase varied among provinces. With the implementation of systems such as “basic farmland” and the “balance of occupation and compensation”, a gradual stabilization has been observed in the grain-sown area over time.

Among the provincial administrative regions, Inner Mongolia boasts the largest grain-sown area, averaging 5.64 × 10⁶ ha annually, with the highest increase rate of 1.11 × 10⁵ ha per year. The adjustment in the planting structure no longer leads to a reduction in grain planted area, as Inner Mongolia ensures that the grain-sown area remains stable at over 6.67 × 10⁶ ha, thereby continuing to make significant contributions to national food security. Conversely, Shaanxi, Gansu, Qinghai, and Ningxia all experienced a decreasing trend in their areas of grain, with Gansu showing the most significant decrease primarily due to an optimized crop planting structure, which led to a reduction in low-yielding grains, such as multi-grain crops. However, there was an increase in the sown area of three major grains: wheat, corn, and potato.

3.2.2. Crop Planting Structure

Figure 3b illustrates the distribution of grain crop planting areas in the northwest arid region. Maize and wheat production accounted for 55.9% and 23.9% of the total output in the northwest arid region from 2000 to 2020, while rice, beans, and potatoes contributed 3.9%, 4.0%, and 7.9%, respectively. The productivity of corn is notably higher than that of rice, with a ratio of 14.2:1. From 2000 to 2016, Inner Mongolia consistently contributed the most to the grain output in the northwest arid region, representing an average of 31.2% of the total output compared to other provinces which were below this threshold at less than 20%. This highlights Inner Mongolia’s pivotal role in ensuring food security within the northwest arid region.

The yield level of the same grain crop exhibits a comparative advantage across different regions. In the case of rice, Ningxia achieved the highest yield per unit area (8.3 t/ha) from 2000 to 2020, followed by Xinjiang, Gansu, Inner Mongolia, and Shaanxi. Notably, three provinces—Ningxia, Xinjiang, and Gansu—surpassed the average level of rice yield per unit area in the northwest arid region. The rice yield per unit area in these three provinces is 1.2 times, 1.2 times, and 1.1 times higher than the regional average. This suggests that these three provinces possess a comparative advantage with higher rice yields than other regions.

During the study period, Xinjiang exhibited a higher wheat yield level (5.4 t/ha) than other provinces, with Qinghai, Shaanxi, Ningxia, Inner Mongolia, and Gansu following suit. The wheat yield per unit area in the largest province was 85.7% higher than in the smallest province, indicating Xinjiang’s tremendous comparative advantage over Gansu. Furthermore, the average wheat yield per unit area in the northwest arid region was notably higher in Xinjiang and Qinghai by 51.6% and 7.8%, respectively, compared to the regional average. This suggests that both Xinjiang and Qinghai possess a relatively high comparative advantage for wheat production in this region; therefore, increasing their respective planting areas (especially in Xinjiang) is crucial for improving the grain yield in the northwest arid region.

From 2000 to 2020, the provinces with the highest yield per unit area of corn, the most critical component of grain, were Xinjiang, Qinghai, Ningxia, Inner Mongolia, Gansu, and Shaanxi. The maize yield per unit area in Xinjiang, Qinghai, and Ningxia exceeded that of the northwest arid region by 33.8%, 18.7%, and 15.6%, respectively.

However, expanding the planting area of these advantageous crops, while seemingly beneficial for increasing the overall grain production, presents several ecological and economic trade-offs that must be carefully considered. The northwest arid region is ecologically characterized by severe water scarcity (as mentioned in Section 2.1). Increasing maize, wheat, and rice cultivation, particularly in areas already experiencing water stress, could exacerbate water shortages. This could lead to the over-extraction of groundwater resources, potentially causing land subsidence and the degradation of the water quality. Furthermore, intensified agriculture often relies on an increased use of fertilizers and pesticides, which can contribute to soil degradation, water pollution (as discussed in Section 4.1 about the matching index), and biodiversity loss in fragile arid ecosystems. The expansion could also pressure existing natural habitats, such as grasslands and wetlands, if agricultural land encroaches upon these areas.

Economically, while an increased production might initially lead to higher incomes for farmers, it also exposes them to market risks. An over-reliance on a few key crops (maize, wheat, and rice) increases the vulnerability to price fluctuations and market volatility. Furthermore, expanding production may necessitate investments in irrigation infrastructure and other inputs, the costs of which may outweigh the benefits, especially for smallholder farmers. Changes in the planting structure could also impact the availability of other crops, potentially affecting local food diversity and dietary habits.

3.3. Grain Supply and Demand Balance Analysis

3.3.1. Food Demand in the Northwest Arid Region

For a populous country like China, the food self-sufficiency rate is commonly used as the national standard for assessing food security. The food self-sufficiency rate measures a country or region’s ability to meet its residents’ grain consumption needs through domestic production. Currently, there are two primary methods for calculating grain consumption: one based on quota statistics and the other using consumption statistics.

(1)

Demand based on quota statistics. The medium- and long-term planning outline of China’s food security (2008–2020) explicitly states that the per capita grain consumption is 400 kg/(person·year). The results indicated a significant linear increasing trend in grain consumption in the northwest arid region from 2000 to 2020 (p < 0.01), with an annual increase of 0.5% (Figure 4a). The M-K trend analysis reveals that the grain consumption in the northwest arid region was not significant from 2000 to 2003 but showed a significant increasing trend from 2004 to 2020 (p < 0.05). The average grain consumption during these two stages was 4.62 × 10⁷ t and 4.91 × 10⁷ t, respectively. From a statistical perspective, it can be inferred that with time, the grain consumption in the northwest arid region will increase by an additional annual amount of approximately 2.6 × 10⁵ t per year. The average annual grain consumption during the study period reached approximately 4.85 × 10⁷ t. By comparison, grain consumption in 2020 exceeded the annual average in 2000 by approximately +5.2% and +11.5%, respectively.

(2)

Demand based on consumption statistics. In addition to quantifying the regional grain consumption through quota statistics, consumption statistics is the most commonly utilized method. This method categorizes grain consumption into five parts: direct consumption, industrial grain consumption, feed grain consumption, loss during transportation and storage, and grain retention. Due to the relatively limited amount of industrial grain data available, the calculation of the grain consumption in the northwest arid region primarily encompasses four components: direct consumption, feed consumption, loss during transportation and storage, and seed retention.

From 2000 to 2020, there was a significant linear increase in grain consumption in the northwest arid region (p < 0.05), with an annual growth rate of 0.4% (Figure 4b). The M-K trend test analysis revealed that the rise in the grain consumption in this region was not statistically significant from 2000 to 2011. However, it exhibited a significant upward trend from 2012 to 2020 (p < 0.05). The average grain consumption during these two periods was recorded as 4.37 × 10⁷ t and 4.50 × 10⁷ t, respectively. Statistically, the annual grain consumption in the northwest arid region is projected to increase by an additional 1.1 × 10⁵ t annually over time. The average annual grain consumption throughout the study period reached approximately 4.42 × 10⁷ t, with a notable increase of 3.7% and 5.1% compared to the yearly average and levels observed in 2000, respectively.

3.3.2. Grain Consumption Structure

The population and grain consumption changes in regions have regional differences (Figure 5). Inner Mongolia and Xinjiang have the most significant grain consumption, with an annual average of 9.7 × 10⁶ t and 8.8 × 10⁶ t, respectively. In contrast, Ningxia has a small population and an annual average of 2.5 × 10⁶ t. The grain consumption in the northwest arid region significantly differed over time. The grain consumption in Gansu and Inner Mongolia increased slowly and then decreased, while the grain consumption in Shaanxi, Xinjiang, Qinghai, and Ningxia increased gradually. Xinjiang recorded the highest annual growth rate of 1.7 percent, followed by Ningxia, Qinghai, and Shaanxi. Under the quota statistics method, the grain demand is directly related to the population. The change in the family concept and the population control policy in the new century has achieved initial results. The population of Inner Mongolia and Gansu grew slowly after 2000 and entered a negative growth period in 2010, and the grain demand also showed a negative growth phenomenon. Ningxia saw the largest increase in grain consumption, at 30.1 percent, as its population grew steadily along with its economic growth, with a population increase of more than 900,000 in the past decade.

From a dietary perspective, the consumption structure of grains in the northwest arid region has gradually shifted from raw grain to animal products. The direct food consumption by the population decreased by 33.7% between 2000 and 2020, with an average annual decrease of 2.5%. Conversely, the indirect consumption of edible oil, pork, beef and mutton, poultry, aquatic products, eggs, and milk exhibited a consistent increasing trend year by year, with average annual growth rates of 1.3%, 1.5%, 6.3%, 6.7%, 4.1%, 2.6%, and 6.6%, respectively. Notably, poultry and pork experienced the most significant increase in their proportion within the dietary structure during this period; their consumption rose from 3.3% and 19.6% to 7.7% and 23.9%, respectively, between 2000 and 2020. The water footprint per unit of calories varies across different foods; for instance, the water footprint per kilogram of calories provided by grains is only 2.87 m³, while that for dairy products is 31.03 m³. This shift towards the increased consumption of animal products, which require more water and land per calorie than grains, exacerbates the existing resource constraints in the arid northwest region. The increasing demand for feed grains to support livestock production puts additional pressure on water resources and arable land, potentially leading to further ecological degradation and reduced overall food security. It can be inferred from the changing consumption trends observed during this study period that there will likely be further increases in the demand for these animal products in the future within this region. This increased production will require more food resources and pose new challenges to food security within the northwest arid region.

3.3.3. Food Supply and Demand Relationship Analysis

China aims to maintain a food self-sufficiency rate of over 95%. A self-sufficiency rate of 100% or higher indicates complete food self-sufficiency, while a range of 95% to 100% signifies basic food self-sufficiency. A rate between 90% and 95% is considered acceptable food security. In contrast, a rate below 90% risks food security or fails to meet the minimum standards for ensuring food security.

(1)

The grain supply and demand balance analysis using quota statistics. Figure 6 illustrates the self-sufficiency rate of grain based on this method. The dynamic trend in grain production in the northwest arid region indicates a transition from a risk stage to complete self-sufficiency. From 2000 to 2001, the food self-sufficiency rate was at a risk stage, with an average rate of 90.2%. However, it decreased to 89.6% in 2001, falling below the minimum food security standard. From 2002 to 2004, there was an improvement, with an average acceptable level of food security at 94.5%. Subsequently, from 2005 to 2007, it reached complete self-sufficiency in grain with an average rate of 103.2%. The period from 2008 to 2020 signifies complete grain self-sufficiency with a steady growth state and an average annual rate of 145.8%. The increasing self-sufficiency rate of grain in the arid northwest region holds significant importance in safeguarding China’s food security, aligning with the country’s overarching food security strategy. Rapid urbanization and industrialization have led to the extensive occupation of arable land in China’s coastal and southern regions, posing substantial challenges to food security regarding production and demand. This poses a severe threat to both regional and national food security levels. Historically dominated by agricultural production, the relatively underdeveloped industrial status of the northwest arid region has elevated its role in ensuring China’s food security. Therefore, it is imperative to ensure the sustainable development of food production in the northwest arid region for the benefit of the local area and the entire nation.

(2)

Grain supply and demand balance analysis by consumption statistics method. Based on the dynamic change trend of the grain production relative to demand estimated via consumption statistics, as depicted in Figure 6, it is observed that the grain self-sufficiency rate in this area has gradually transitioned from the bare self-sufficiency stage to the complete self-sufficiency stage. Specifically, from 2000 to 2003, the grain self-sufficiency rate in the northwest arid region was at a basic level with an average of 97.4%. Subsequently, from 2004 to 2020, there has been a shift towards complete self-sufficiency in grain production, with an average self-sufficiency rate of 149.9%. It is noteworthy that when compared with the normal statistics method, the consumption statistics method yields higher estimates for food self-sufficiency rates in the northwest arid region, all indicating a high level of food security. It is important to note that these self-sufficiency rates represent the overall situation for the entire northwest arid region. Variations may exist among the different provinces due to local production capabilities, population densities, and economic development. Maintaining this positive trend will require addressing potential challenges like population growth, evolving dietary preferences, and the continued effectiveness of agricultural policies and practices.

4. Discussion

This study evaluates the future food security landscape of the northwest arid region by projecting grain production based on a comprehensive soil and water resources matching index and forecasting the grain demand considering demographic shifts and dietary changes. The resulting food self-sufficiency rate assessment provides critical insights for policy formulation in this environmentally sensitive region.

4.1. Forecast for Food Production

(1)

Soil and water resources matching index. Our evaluation system incorporates gray water discharge in agricultural production, which is essential for comprehensively assessing regional food security. This approach fully reflects the region’s agricultural water and soil resource endowment, the water-saving irrigation level, and the environmental impact of agricultural activities (Figure 7). A direct and significant connection exists between the soil and water matching index and food security. An increase in the index indicates improved resource matching, which directly leads to an enhanced agricultural production efficiency and increased yields; conversely, a decrease in the index signifies a resource allocation imbalance, which reduces agricultural productivity and threatens food security. Specifically, regions with a matching index greater than one have relatively abundant water resources that can support more intensive agricultural production. Meanwhile, regions with an index of less than one face relative water scarcity and need to adopt water resource management strategies and cropping structure adjustments to maintain food security. This quantitative index provides a crucial tool for policymakers: areas with consistently low or negative indices require urgent interventions focused on improving water use efficiency (e.g., mandatory upgrades to water-saving irrigation) or potentially shifting towards less water-intensive agricultural activities to align the demand with a sustainable supply.

From 2000 to 2020, Qinghai Province’s index increased from 15.29 in 2000 to 31.71 in 2020. This increase is primarily attributed to the small sown area of crops, resulting in a relatively small generation of agricultural gray water during agricultural production. The higher matching index in Qinghai can be further explained by its unique geographical conditions, including higher altitude regions with abundant snowmelt water resources, a lower population density leading to less intensive agricultural activities, and the implementation of effective water conservation policies in recent years. Despite limited arable land resources, these favorable conditions create a strong foundation for sustainable agricultural development. Nevertheless, the limitation of arable land resources in Qinghai remains a major bottleneck for agricultural development, which needs to be addressed through optimized land use and the introduction of high-efficiency agricultural technologies.

In contrast, other provinces and regions had negative matching indices. They were facing varying degrees of environmental water shortage, indicating that the northwest arid region was under tremendous ecological and environmental pressure while shouldering the burden of grain production [28,29]. The prevalence of negative indices across most regions signals a systemic challenge where the agricultural water demand frequently exceeds environmentally sustainable levels. This implies that water resource management policies must move beyond simple supply augmentation (where feasible) and focus critically on demand management, stricter water allocation enforcement, and enhancing productivity per unit of water consumed. These regional disparities in matching indices have significant implications for differentiated policy interventions. For Qinghai, with its positive indices, policies should focus on strategically expanding agricultural production while preserving its water resource advantages through a continued investment in efficient irrigation technologies. More aggressive water conservation measures are needed for regions with negative indices, like Gansu, Ningxia, and Xinjiang, including technological adoption, cropping pattern adjustments toward water-efficient crops, and a potential economic diversification beyond water-intensive agriculture. Targeted water policies, informed by the matching index, could include setting specific water efficiency benchmarks for different agricultural zones or offering tiered subsidies favoring the adoption of advanced water-saving technologies in areas with the most severe water stress (lowest indices).

Gansu Province has increased from −0.67 in 2000 to −0.11 in 2020. Similarly to the exponential growth observed in Qinghai Province, the increase in Gansu Province can be attributed to a significant rise in available agricultural water resources and a relatively minor change in agricultural gray water, resulting in the elevation of this index. This improvement demonstrates that targeted resource management approaches can effectively enhance matching indices even in challenging environments. Regional cooperation frameworks, allowing resource sharing and coordinated planning based on each area’s matching index, could optimize agricultural production and environmental sustainability across the northwestern region. The declining trend observed in most provinces and regions from 2000 to 2020 is primarily due to the higher growth rate of agricultural ash water compared to available agricultural water resources. The results of this study are similar to those of Huang et al. [30], who also identified the increasing pressure from agricultural greywater on the water resource sustainability across various regions in China. Our findings specifically confirm this trend within the arid northwest, highlighting that the water quality degradation associated with agricultural runoff is compounding the challenges posed by absolute water scarcity, a critical point for integrated water resource management in the region.

(2)

Forecast of grain production in 2035. The GM(1,1) model was chosen for projecting the comprehensive soil and water resources matching index due to its effectiveness in handling systems with limited data points and inherent uncertainties, which is often the case when analyzing long-term agricultural and environmental trends. Its strength lies in identifying and extrapolating underlying exponential trends from relatively small datasets without requiring strict statistical distributions. However, it is important to acknowledge its limitations; the model primarily captures monotonous trends and may be less accurate if future conditions involve sharp, non-linear changes or external shocks not reflected in the historical data.

Based on the comprehensive soil and water resources matching index, this study utilized the GM(1,1) model to project the grain output of the research area in 2035. Furthermore, it employed the comprehensive soil and water resources matching index of each provincial administrative region from 2000 to 2016 to forecast the future comprehensive soil and water resources matching index using a gray prediction model. The accuracy of these predictions is confirmed by comparing them with the 2017–2020 comprehensive soil and water resources matching index. By 2035, it is projected that the comprehensive soil and water resources matching index for each provincial administrative region in the northwest arid region will be as follows: Inner Mongolia (−0.93), Shaanxi (−3.29), Gansu (−0.07), Qinghai (28.44), Ningxia (−1.86), and Xinjiang (−1.00). It should be noted that ecological factors are taken into consideration when calculating these values, with negative figures indicating environmental water scarcity. The persistence of negative indices in the forecast for most key grain-producing provinces reinforces the conclusion that water scarcity will remain a fundamental constraint, demanding proactive and sustained water resource management efforts to support even the modest projected increase in grain output.

Based on the comprehensive index of soil and water resources matching, a regression analysis was conducted to examine its relationship with grain output. The regression models for provincial administrative regions in the northwest arid region are presented in

Table 2. Regression model of integrated soil and water resources matching index and grain yield.

Regions	Regression Models	Correlation Coefficient
Inner Mongolia	Y = 1664.31 − 2154.69x	0.59
Shaanxi	Y = 793.60 − 157.63x	0.73
Gansu	Y = 1229.37 + 354.68x	0.53
Qinghai	Y = 88.71 + 0.52x	0.35
Ningxia	Y = 147.77 − 171.93x	0.89
Xinjiang	Y = 1029.21 − 818.89x	0.87

. The multiple correlation coefficients of the regression models are all above 0.25, indicating a correlation, with most provinces exceeding 0.50, reflecting a moderate to strong correlation. This statistical link provides concrete evidence for policymakers that investments in improving water and soil resource matching (e.g., through enhanced irrigation efficiency, soil moisture conservation practices, and optimized water allocation) can yield tangible benefits regarding grain production, directly contributing to food security goals. However, correlation does not imply causation; other factors (technology and policy) also influence production.

Combined with the forecast results of the comprehensive soil and water resources index in 2035 and the regression model function, the grain output in the northwest arid region in 2035 can be calculated as follows: Inner Mongolia (3.67 × 10⁷ t), Shaanxi (1.31 × 10⁷ t), Gansu (1.20 × 10⁷ t), Qinghai (0.10 × 10⁷ t), Ningxia (0.47 × 10⁷ t), and Xinjiang (1.85 × 10⁷ t). In 2035, the total grain output in the northwest arid region will be 8.61 × 10⁷ t, an increase of 4.8% over 2020. Liu’s research showed a grain output of 9.46 × 10⁷ t in the northwest by 2035 [31], resulting in a difference of approximately 9.87% compared to our projection. This discrepancy may stem from differences in the definition of the study area, but our forecast focuses on the core arid zone, fully accounting for water resource constraints on the production potential.

4.2. Grain Demand Forecasting

The projected increase in population, the advancement of urbanization, and the rise in the residents’ consumption of animal products will result in a significant surge in the food demand within the northwest arid region. This substantial uptick in grain consumption heightens the necessity for food security within this region and presents formidable challenges to ensuring China’s overall food security. These projections are based on future population estimates provided by the Population Division of the United Nations Department of Economic and Social Affairs and China’s National Population Development Plan (2020–2030), yielding insights into China’s anticipated population size, demographic structure, and level of urbanization by 2035. Based on the 2020 ratio between the northwest arid region and China, data on the population number, population aging level, and urbanization level for each year were obtained.

The per capita grain consumption in the future was predicted using a gray prediction model, taking into account the different dietary structures of urban and rural residents in the northwest arid region from 2000 to 2016. Accounting for these dietary shifts is crucial, as the increasing preference for animal-based products translates into a significantly higher per capita demand for feed grains (predominantly corn in this region) and places amplified pressure on already scarce water and land resources due to the higher resource intensity of livestock production compared to direct grain consumption. The forecast results demonstrate high reliability, effectively capturing regional consumption trend variations and providing a solid foundation for grain demand projections.

By multiplying the per capita grain consumption with the population in the northwest arid region in 2035, we can obtain the projected grain consumption for that year, as shown in

Table 3. A forecast of grain consumption in the study area in 2035.

Factors	Level
	Low	Medium	High
Population (106)	130.07	134.45	138.83
Total grain consumption (106 t)	68.61	70.51	72.39

. The grain consumption forecasts under low, medium, and high population growth scenarios show increases of 34.4%, 38.1%, and 41.8%, respectively, compared to 2020. This substantial projected increase starkly contrasts with the forecasted 4.8% rise in grain production over the same period, highlighting a widening gap and underscoring the urgency for strategic policy interventions. When compared to the projected grain production of 8.61 × 10⁷ t, even under the most optimistic scenario, the grain self-sufficiency rate will only be 84.1%, falling below the FAO’s minimum standard of 90% and China’s national benchmark of 95%. Addressing this significant anticipated shortfall requires immediate and comprehensive planning. Key strategies must focus on multiple fronts: maximizing sustainable yield improvements through technological adoption and resource management within the region’s ecological constraints (as discussed regarding water-saving agriculture and resource matching), exploring measures to moderate demand growth patterns, significantly reducing post-harvest losses and food waste throughout the supply chain, and strategically leveraging the inter-regional and international grain trade to ensure supply stability. These data directly support our conclusion that the region faces a significant risk to food security and underscores the need for strategic interventions.

4.3. Analysis of Food Security

Based on the integrated land and water resources matching index, the forecasted grain production in 2035 is projected to surpass that of 2020. Additionally, the anticipated grain consumption based on the population size, aging demographics, and urbanization level is expected to exceed that of 2020 significantly. While the increase in grain production contributes to regional food security, it remains insufficient compared to the rising demand for grain consumption. This suggests a less optimistic outlook for food security in the northwest arid region. The food self-sufficiency rate in the study area is projected to be between 79.6% and 84.1% under low, medium, and high grain consumption forecasts, respectively, by 2035. This range is significant, representing the uncertainty inherent in long-term demand projections based on population and consumption trends. The existence of this range indicates that policy formulation must be flexible and adaptable to address the potential grain deficit under varying demand scenarios.

These rates fall significantly below the minimum food security standard of 90% as defined by the FAO and China’s national benchmark of 95%, indicating a serious challenge to regional food security. The implications of this projected shortfall are profound. From a water resource management perspective, this projected gap directly translates into the reality that current water availability and management practices are insufficient to meet future food needs sustainably from domestic production alone. This underscores the urgency for policies that maximize the ‘crop per drop’—enhancing water productivity. It signifies a growing reliance on external grain sources from other regions within China or international markets to meet basic needs. This heightened import dependence exposes the region to a greater vulnerability from fluctuations in global food prices, potential trade disruptions, and logistical challenges, thereby increasing risks to stable food access for its population.

Furthermore, it could strain the region economically and potentially contribute to higher consumer food prices. A reduced contribution from the northwest, traditionally a key grain supplier, could intensify pressure on China’s overall food balance and internal supply chains. This scenario is not unique; it mirrors the global challenges faced by many water-scarce regions, such as those in the Middle East and North Africa (MENA), where a high population density and limited water resources necessitate substantial food imports to ensure food security. Like those regions, Northwest China’s trajectory highlights the critical tension between the growing food demand driven by population and dietary shifts and the environmental limits imposed by resource scarcity, particularly water. Therefore, robust water management policies are not merely an agricultural issue but a cornerstone of regional stability and food security, essential for mitigating the risks associated with import dependence.

It should be noted that future climate change and policy adjustments may influence the forecast results, which requires further exploration in subsequent studies. Future food security levels in the northwest arid region are not promising, particularly given the increased attention given by state authorities to green agriculture development efforts. For instance, the government has issued the Technical Guidelines for Green Agricultural Development (2018–2030), the Action Plan for Zero Growth of Fertilizer Use by 2020, the Action Plan for Zero Growth of Pesticide Use by 2020, and the overall plan for a new round of returning farmlands to forests and grasses.

While crucial for long-term environmental sustainability in this fragile ecosystem, these green policies introduce complex dynamics concerning immediate food production goals. If not accompanied by efficiency-enhancing practices, the regulation and reduction in fertilizer and pesticide applications could potentially lead to yield stagnation or even declines in the short term, exacerbating the projected food gap. Similarly, returning farmland to forests and grassland directly reduces the arable land base. Balancing these environmental objectives with the imperative of food security requires careful policy design and implementation. Specifically for water management, this means ensuring that water conservation measures required by green policies (e.g., associated with land retirement or reduced input intensity) are coupled with investments in efficiency for the remaining agricultural land, preventing unintended negative impacts on the overall water availability for food production. The regulation of fertilizer and pesticide application will inevitably impact the regional grain production, necessitating the prompt implementation of practical measures [32], such as providing subsidies to farmers for adopting certified water-saving irrigation systems (like drip or sprinkler irrigation) and promoting integrated nutrient management practices that enhance fertilizer use efficiency. Policy adjustments might be needed to better support farmers during this transition, potentially including stronger technical assistance programs focused on maintaining yields with reduced inputs (e.g., through integrated pest management and soil health improvements) and financial mechanisms that compensate for potential initial yield losses. Furthermore, land retirement programs should strategically target ecologically sensitive or highly degraded marginal lands, minimizing the impact on prime agricultural areas. Investing in research and development for crop varieties that are both high-yielding and resource-efficient (water and nutrients) is also critical to reconciling green development with production needs.

Furthermore, as urbanization progresses, agricultural water resources and cultivated land significantly diminish while the soil quality deteriorates, decreasing farmers’ enthusiasm for grain cultivation. Addressing this requires concerted efforts, potentially including financial incentives for maintaining high-quality farmland, investing in research and extension services focused on soil health improvement techniques (e.g., cover cropping and conservation tillage), and supporting the development and adoption of crop varieties bred explicitly for their resilience to arid conditions and lower input requirements. Relying solely on the regional food production potential becomes increasingly challenging in ensuring food security [33]. While international trade is necessary (as indicated by the projected SSR), effective domestic water resource management remains critical to maximizing the region’s production potential within ecological limits, thereby reducing the scale of import dependency and enhancing resilience. Therefore, actively engaging in international food trade is an effective strategy to mitigate these threats to regional food production while guaranteeing domestic food security. International food trade can leverage inter-country or inter-regional industrial advantages to effectively overcome limitations imposed by domestic or regional water and land resources on grain production. Specifically, strategic trade agreements and partnerships can allow Northwest China to import water-intensive bulk grains from regions with more extraordinary resource endowments, effectively importing ‘virtual water’ and reducing the pressure on its scarce water supplies and fragile ecosystems. This approach aligns with domestic environmental goals like ecological restoration.

Furthermore, diversifying import sources through multiple agreements enhances the supply chain resilience against disruptions in any exporting region. To ensure environmental sustainability beyond the region’s borders, these trade partnerships could ideally prioritize sourcing from countries employing sustainable agricultural practices or even incorporate mutually agreed upon sustainability criteria. However, the environmental footprint of transportation remains a factor to consider. To enhance food security, the following policy recommendations are proposed:

(1)

Promote technological innovation: Prioritize and heavily subsidize the widespread adoption of proven water-saving agricultural technologies, such as drip irrigation, sprinkler systems, and soil moisture sensors, particularly in provinces with negative water–soil matching indices, like Xinjiang, Ningxia, Shaanxi, and Inner Mongolia. Simultaneously, invest in breeding and promoting drought-resistant crop varieties specifically adapted to the region’s conditions to improve the water resource utilization efficiency directly at the plant level.

(2)

Enhance regional cooperation: Establish a formal inter-provincial mechanism for coordinated water resource management in the northwest. This should involve sharing water availability and use data, jointly planning water allocation strategies based on regional matching indices and downstream needs (especially for transboundary rivers), and collaborating on developing and disseminating best practices in water-saving agriculture and agricultural technology. The experiences of Xinjiang and Ningxia in achieving relatively high yields (Section 3.1.2) can provide valuable lessons for the entire region.

(3)

Increase international grain trade: supplement domestic production through international grain trade to ensure the stability of the grain supply, recognizing that domestic water resources are insufficient to achieve complete self-sufficiency sustainably.

(4)

Implement ecological protection measures: Strategically implement policies like returning marginal farmland (especially in areas with critically low water–soil matching indices or high ecological sensitivity) to forests and grasslands. Ensure these programs are coupled with support for intensifying sustainable production on remaining, more suitable, agricultural lands, integrating water conservation goals with ecological restoration efforts. Promote green agricultural development through incentives for reducing the water pollution from agricultural sources.

4.4. Broader Implications and Limitations

The findings of this study carry significant implications beyond the immediate geographical focus. Given Northwest China’s role as a key grain production area, the projected decline in its regional self-sufficiency below critical thresholds poses a considerable challenge to China’s overall national food security strategy. It implies an increased pressure on other major grain-producing regions within the country and likely necessitates a greater reliance on international markets, potentially influencing global grain trade dynamics and prices. Furthermore, the interplay between water scarcity, climate factors, population growth, dietary shifts, and green development policies mirrors the challenges numerous arid and semi-arid regions face worldwide. Therefore, the methodologies and results can offer valuable insights for policymakers and researchers grappling with similar food security and sustainable agriculture issues in comparable environments.

While this study provides valuable insights into the food security dynamics of Northwest China, several limitations should be acknowledged, which in turn highlight directions for future research.

(1)

Model Limitations: The reliance on the GM(1,1) model for forecasting, while suitable for handling limited data series, primarily captures monotonous exponential trends. It may be less accurate in predicting future scenarios involving significant non-linear changes, cyclical patterns, or abrupt shifts resulting from unforeseen policy interventions, market shocks, or extreme climate events not reflected in the historical data (2000–2020). Furthermore, the regression analysis linking the matching index to the grain output establishes a correlation but does not fully capture the complex, potentially non-linear interplay of all factors influencing yield and assumes historical relationships will hold.

(2)

Data Aggregation and Assumptions: The analysis predominantly uses provincial-level data. This aggregation can mask significant intra-provincial heterogeneity in resource endowments, agricultural practices, and food security status (e.g., differences between irrigated oasis agriculture and more marginal rain-fed areas). Additionally, calculations for the water–soil matching index rely on certain simplifying assumptions (e.g., a fixed coefficient (α = 40%) for available surface water, specific methods for effective precipitation and ET0 calculation, and potentially averaged Kc values). While necessary for a regional overview, these assumptions introduce uncertainties. Precisely defining and calculating AWA and AWR components within the matching index also warrants a careful consideration and sensitivity analysis.

(3)

Scope of Analysis: This study focuses primarily on the biophysical and demographic drivers of food security (water, land, yield, population, and diet). While crucial, it gives less explicit attention to the socio-economic factors influencing food production and access, such as farmers’ adoption of technologies, the effectiveness of specific subsidy policies, land tenure issues, market access, input costs, and the costs versus benefits of recommended interventions.

(4)

Climate Change Impacts: Although climate change is acknowledged as a critical factor amplifying vulnerability (Section 1 and Section 4.1), its impacts are not dynamically modeled within the grain production forecasts. The projections implicitly assume a continuation of past trends influenced by climate but do not explicitly quantify the potential effects of future changes in temperature, precipitation patterns, or the frequency and intensity of extreme weather events (droughts and heatwaves) on crop yields and water resource availability beyond what is captured in the historical trend extrapolation.

(5)

Policy and Trade Dynamics: This study recommends promoting water-saving technology and increasing international trade. However, it does not deeply evaluate the implementation challenges, cost-effectiveness, or potential unintended consequences of specific policies (e.g., distributional effects of subsidies and impacts of water pricing). Similarly, while trade is identified as necessary, the analysis does not delve into the complexities of an increased import reliance, such as price volatility risks, geopolitical considerations, or the sustainability implications of virtual water trade.

4.5. Future Research Directions

Addressing these limitations suggests several avenues for future research:

(1)

Employing advanced modeling techniques: utilize more sophisticated forecasting models (e.g., system dynamics models incorporating feedback loops, agent-based models simulating farmer behavior, machine learning algorithms trained on wider datasets, or econometric models explicitly including policy and climate variables) to better capture complex interactions and potential future discontinuities.

(2)

Utilizing finer-scale data: incorporate finer spatial resolution data (e.g., county-level or watershed-level) and use region-specific parameters and dynamic coefficients where possible to refine resource accounting, matching assessments, and production modeling, revealing intra-regional disparities.

(3)

Integrating socio-economic factors: conduct detailed socio-economic analyses, potentially through farmer surveys, cost–benefit analyses of technologies/policies, or integrated assessment models, to understand adoption barriers, policy effectiveness, and equity implications.

(4)

Explicitly modeling climate change impacts: integrate outputs from downscaled climate models (e.g., CMIP6) into crop simulation models (e.g., DSSAT and APSIM) and hydrological assessments tailored to the northwest region to provide more robust projections of climate change impacts on yields, water resources, and overall food security.

(5)

Conducting in-depth policy and trade analyses: perform a comparative policy analysis (e.g., evaluating different subsidy schemes or water pricing mechanisms) and detailed assessments of trade strategies, including a quantitative risk analysis (e.g., supply chain disruptions and price volatility) and an evaluation of the environmental and social sustainability of import sources (virtual water trade implications).

5. Conclusions

This study contributes significantly to understanding the long-term food security trajectory of arid Northwest China by developing and applying an integrated assessment framework. Its primary distinction lies in quantitatively linking future grain supply and demand projections (to 2035) directly to the carrying capacity of regional water and soil resources, evaluated through a comprehensive matching index. This resource-constrained approach yields crucial insights into the region’s future capacity: (1) The analysis of the 2000–2020 period revealed a consistent growth in grain production (3.5% CAGR) and a significant improvement in regional self-sufficiency, highlighting Xinjiang and Ningxia’s high per-unit yields as key contributors. Corn and wheat remain the dominant crops. (2) Despite historical self-sufficiency achievements, a marked dietary shift towards animal products is amplifying the feed grain demand, adding structural pressure to the supply–demand balance. (3) Most critically, the integrated forecast, constrained by the projected water–soil matching potential (using GM(1,1) and regression), projects a decline in the regional grain self-sufficiency rates to between 79.6% and 84.1% by 2035, quantifying a significant future challenge.

This projected decline to levels substantially below critical security benchmarks (FAO’s 90% and China’s 95%) represents a key outcome derived from integrating resource limitations into the long-term food security assessment. It highlights a stark contrast to past achievements and signals a serious impending challenge to regional food security, driven by the quantified disconnect between the modest future production potential (constrained by resources) and rapidly escalating consumption demands. This increases the reliance on external grain sources and associated vulnerabilities, with potential repercussions for national food security. Consequently, addressing the explicitly identified constraints of water scarcity and escalating demand necessitates urgent and targeted policy interventions. Strategic priorities derived directly from this research must include accelerating the adoption of water-saving technologies and drought-resilient crops; establishing robust inter-provincial water resource management frameworks for optimized allocation; leveraging a strategic and diversified international grain trade to supplement domestic production within sustainable boundaries; and integrating ecological conservation measures with policies promoting sustainable agricultural intensification on suitable land. Further research incorporating finer-scale data, dynamic climate impacts, and socio-economic factors is needed to refine these projections and policy recommendations.