Crop Redistribution Increases Regional Production While Reducing Water Deficit, Fertilizer Use, and Production Losses: Evidence from a Multi-Objective Optimization at the County Level in Northeast China

Abstract

Given the increasing crop yield losses, water scarcity, and fertilizer application in Northeast China, a systematic assessment is increasingly necessary to investigate the potential of crop redistribution to enhance grain production while alleviating environmental pressures. Here we quantify the potential of crop redistribution in Northeast China through a multi-objective optimization approach. First, we construct a dataset that contains four objectives including crop yield, yield losses, water deficit, and nitrogen fertilizer application based on their annual data in 273 counties over two decades (2001–2020). Second, we optimize the county-level distribution of rice, maize and soybean using the developed dataset and evaluate the benefits on each objective. Finally, we design a crop redistribution scheme and analyze its impact on the cropping structure in Northeast China based on the optimal solution. Results show significant potential of crop redistribution, with crop production increases by 1.70% (2.41 × 10⁶ tons), production losses decrease by 2.69% (1.84 × 10⁵ tons), water deficit decreases by 6.78% (3.88 × 10⁸ m³) and N fertilizer application decreases by 10.87% (5.41 × 10⁷ kg) when all the objectives are optimized simultaneously. The crop redistribution scheme is summarized as follows: compared with the baseline crop structure, rice area increases by 69.58%, maize reduces by 12.8%, and soybean reduces by 54.79% in Northeast China. Specifically, rice area increases in northwestern Heilongjiang, eastern Jilin, most counties in Liaoning, and reduces elsewhere. Maize area reduces in most of the counties, except for several counties in southwestern Heilongjiang, northern Jilin, and northern parts of the Four Eastern Leagues. Soybean area reduces in northern part of Heilongjiang and Four Eastern Leagues and increases in western Jilin and most counties in Liaoning. Although crop redistribution scheme was generated, the model remains limited in terms of crop types, spatial resolution, and the range of factors influencing crop distribution. Future work will address these limitations to enhance the reliability and applicability of the crop redistribution model.

1. Introduction

As the global population continues to expand, humanity requires increased food production to satisfy worldwide demand [1]. Extreme events, such as drought, flood and heatwaves, have increased globally in both intensity and frequency under climate change, introducing additional uncertainty into global food security [2,3]. Studies have shown that crop losses caused by disasters—including both weather extremes and pests or diseases—have increased significantly over the past century and are projected to rise even further in the future [4,5,6,7]. On the other hand, despite the substantial increase in food production worldwide in recent decades [8], intensified agricultural inputs such as chemical fertilizers and pesticides have resulted in a number of environmental issues including soil degradation and GHG emissions [9,10]. Meanwhile, global warming has increased crop water demand, rendering crops more vulnerable to drought and exacerbating the irrigation pressure [11]. In the face of these severe challenges, crop redistribution has increasingly emerged as an effective technique for enhancing the productivity and sustainability of agricultural systems [12,13].

Previous research shows the significant potential of crop redistribution despite differences in study area, agricultural aspects, and optimization approach. For instance, Davis et al. reshaped the spatial distribution of crops globally, resulting in a new distribution that would feed an additional 825 million people while reducing the consumptive use of water resources [14]. Focusing on the environmental impact, Yin et al. optimized the distribution of cropping systems in China, reducing total GHG emissions without compromising food security, water use, or farmers’ incomes [15]. Different from conventional research, Rising et al. examined how crop redistribution may support the adaptation to climate change and found that the relocated cropland would avoid half of the agricultural losses in the United States [16].

Although crop redistribution has demonstrated its effectiveness, obstacles still exist. The trade-offs between dimensions in agriculture, which have been well documented by considerable studies [17,18,19], lead to the conflicts between different objectives in crop redistribution model [20]. For instance, Xie et al. found that, when optimized individually, farmer incomes in China can be raised by as much as 90.5% but at the cost of other environmental outcomes [21]. Hu et al. reported clear trade-off between economic and environmental benefits when they optimized the crop planting structure in Heilongjiang Province [22]. Therefore, identifying a method that balances multiple agricultural objectives in crop redistribution model, thereby ensuring food security and promoting sustainability, becomes an urgent task.

The conventional multi-objective optimization methods such as weighted sum method and ε-constraint methods [23,24], attempt to transform the optimization problems from multiple objectives to single objective. In contrast, Genetic Algorithms can optimize all the objectives simultaneously without knowing their relative importance, making them well-suited for multi-objective optimization problems. Non-dominated Sorting Genetic Algorithm II (NSGA-II), proposed by Deb et al. in 2002, introduces elitism and crowding distance to enhance solution diversity and convergence in multi-objective problems, and is widely applied in agricultural optimization [25,26]. For instance, Huang et al. conducted crop redistribution in China’s Yellow River Basin based on NSGA-II and alleviated blue water scarcity by 20%, combined with a 5% increase in crop economic output [27]. Liu et al. used the NSGA-II to optimize crop distribution across Chinese provinces, reducing the blue water footprint by 16% while increasing caloric supply by 12% [28].

Northeast China is one of the most important grain-producing regions in China and a critical area for national food security. In 2022, Northeast China accounted for 30.01% of China’ food production [29]. Although grain production in Northeast China has increased significantly over the past few decades, this has been accompanied by a substantial increase in fertilizer use and irrigation [30]. Under the context of climate change, disaster-induced crop losses have increased significantly in the region. The crop losses in Northeast China arise from both unfavorable climatic stresses (e.g., drought, flooding, and cold damage) and biotic pressures, including major pests and diseases such as rice brown planthopper and blast, maize borer and northern leaf blight, and soybean moth and root rot [31]. Although the potential of crop redistribution has been demonstrated in numerous studies, research focused on Northeast China remains limited. Existing work has examined this potential to some extent, yet few studies have optimized crop distribution at the county level. Therefore, there is an urgent need to investigate county-level crop redistribution and to develop detailed optimization strategies for Northeast China.

Here, we optimize the crop distribution in Northeast China, considering four objectives including: crop production, production losses, water deficit, and nitrogen fertilizer application. Our study built a crop redistribution model based on NSGA-II, combining multi-source datasets including weather data, crop yield, phenology data, fertilizer application, then optimized the distribution of 3 major crops (rice, maize and soybean) across 273 counties in Northeast China. Finally, we analyze the impact of crop redistribution on the cropping structure and the locations of three crops’ geographical centroid in the region. This study is intended to serve as a reference for crop redistribution and cropping structure adjustment in Northeast China.

2. Materials and Methods

2.1. Overview of the Study Area

Northeast China comprises three provinces of Heilongjiang, Jilin and Liaoning, along with the Four Eastern Leagues of the Inner Mongolia Autonomous Region (Hulunbeier, Xing’an, Chifeng and Tongliao). The region features a basin-like topography with central plains surrounded by mountains on three sides. Agricultural production is primarily concentrated in the Plains, which consists of the Sanjiang Plain, Songnen Plain, and Liaohe Plain. Northeast China experiences a temperate monsoon climate characterized by hot summers and long, cold winters. The annual mean temperature ranges from −6 °C to 11 °C, with annual precipitation totaling 400–1000 mm. Most of the precipitation occurs between July and September, coinciding with the crop growing season. The frost-free period lasts between 90 and 180 days, resulting in a short growing season and supporting only a single-cropping system. Most agriculture in the region is rain-fed; however, areas cultivated with rice generally have well-developed irrigation infrastructure. The black soil region in Northeast China is one of the largest black soil regions around the world, which is rich in organic matter and highly suitable for crop cultivation. As one of the most important grain-producing areas in the country, Northeast China primarily grows rice, maize, and soybeans (Figure 1). In 2023, these three crops accounted for 16.04%, 51.38%, and 20.56%, respectively, of the region’s total cultivated area.

2.2. Data Sources

2.2.1. Meteorological Data

Meteorological variables including air temperature, pressure, humidity, wind speed, and solar radiation were obtained from the ERA5-Land dataset, produced by the European Centre for Medium-Range Weather Forecasts (ECMWF). ERA5-Land is a high-resolution reanalysis dataset driven by near-surface forcing from ERA5 and a land surface model [32]. The dataset provides hourly data at a spatial resolution of 0.1° × 0.1° (approximately 9 km × 9 km) since 1950. For this study, hourly data from 2001 to 2020 were resampled into daily data to support subsequent calculations.

2.2.2. Phenology Data and Crop Coefficients

The phenology data in this study were derived from observations at agrometeorological stations and were used to calculate crop water requirements and effective precipitation across different growth stages. The dataset covers the period from 1991 to 2012 and 3 major crops including rice, maize and soybean. Days of the year (DOY) for five key phenological stages are selected, and their multi-year averages are presented in Figure 2. Due to the substantial latitudinal variation across Northeast China, phenological stages typically differ by approximately 15 to 20 days between the northern and southern regions. We interpolated the site data to a 0.1° × 0.1° grid using the Kriging method. Crop coefficients used in the calculation of crop water requirement for different growth stages were sourced from FAO-56 guidelines (see

Table 1. Crop coefficients of rice, maize and soybean during different growth stages.

Crop	Kcini	Kcmid	Kcend
rice	1.05	1.20	1.00
maize	0.30	1.20	0.60
soybean	0.40	1.15	0.50

Note: The crop coefficient during the initial stage of crop growth is Kc_ini; the crop coefficient during the crop development stage increases linearly from Kc_ini to Kc_mid; the crop coefficient during the middle-season stage is Kc_mid; and the crop coefficient during the late-season stage decreases linearly from Kc_mid to Kc_end.

).

2.2.3. County-Level Yield and Planting Area

Crop-specific yield and planting area data at the county level were collected from statistical yearbooks of prefecture-level divisions in Northeast China, covering the yield and planting area of rice, maize, and soybean for 273 counties from 2001 to 2020. Somes counties were excluded from the study due to their poor data availability. Prior to analysis, the data were cleaned to remove abnormal records to ensure the accuracy of the results.

2.2.4. Nitrogen Fertilizer Data

Nitrogen fertilizer application data were derived from a dataset developed by Yu et al., which reconstructed nitrogen fertilizer use for 10 major crops in China from 1952 to 2018 at a 5 km × 5 km resolution [33]. We extracted the gridded nitrogen fertilizer application rate in Northeast China from the dataset. Then, the extracted data was spatially averaged over each county using ArcGIS (v10.8) zonal statistics to obtain mean values at the county level from 2001 to 2018.

2.3. Calculation of Average Yield Loss Rate

The average yield loss rate represents the long-term disaster-induced yield loss for a given county, encompassing both meteorological hazards and biotic factors such as pests and diseases. This study used a detrending method to estimate the average loss rate for each county [34], calculated as follows:

$$R={\displaystyle \frac{1}{n}}{\sum }_{i=1}^n\max \left[{\displaystyle \frac{Y_t-Y_i}{Y_t}},0\right]$$

(1)

where R is the average yield loss rate during the study period, Y_i is the actual yield in year i, Y_t is the yield trend in year i, and n = 20 (for years 2001–2020). A three-year moving average method was used to estimate yield trend, as it has been shown to be a simple and effective method for trend estimation.

2.4. Estimation of Crop Water Requirement and Water Deficit

Crop water requirements were estimated using the single crop coefficient approach recommended in FAO-56 [35]. The crop growth cycle was divided into four stages: initial stage, crop development stage, mid-season stage, and late season stage. Daily crop water requirement was calculated as:

$${ET}_{ci}=K_{ci}\times {ET}_{0i}$$

(2)

where ET_ci is the water requirement of crop c on day i. Kc_i is the crop coefficient of crop c on day i, and ET_0i is the reference evapotranspiration on day i. Reference evapotranspiration was calculated using the FAO-56 Penman–Monteith equation:

$${ET}_0={\displaystyle \frac{0.408∆\left(R_n-G\right)+\gamma \frac{900}{T+273}{U}_2\left(e_s-e_a\right)}{∆+\gamma \left(1+0.34U_2\right)}}$$

(3)

where Rn is net radiation (MJ·m⁻²·d⁻¹), G is soil heat flux density (MJ·m⁻²·d⁻¹), T is mean daily air temperature at 2 m height, e_s and e_a are saturation and actual vapor pressure respectively, Δ is the slope of the vapor pressure curve, γ is the psychrometric constant, and U₂ is wind speed at 2 m height. Wind speed and solar radiation from ERA5-Land were converted accordingly:

$$U_2=U_z{\displaystyle \frac{4.87}{\mathit{ln}\left(67.8z-5.42\right)}}$$

(4)

$$R_n=SSR+STR$$

(5)

where U_z is wind speed at height z (z = 10 m), SSR is surface net solar radiation, and STR is surface net thermal radiation. Other parameters were derived following FAO-56 recommendations. Daily crop water requirements were summed over the entire growing season to obtain total water requirement per year. Then its units were converted from mm to m³/ha:

$$ET=10\times {\sum }_{i=1}^n{ET}_{Ci}$$

(6)

where ET is the crop water requirement during its entire growing season, whose unit is m³/ha. Water deficit (WD) was then calculated as the difference between water requirement and effective precipitation:

$$WD={ET}_c-P_e$$

(7)

where ET_c and P_e are the 2001–2020 average crop water requirement and effective precipitation during the growing season, respectively. P_e is obtained by accumulating the daily precipitation within the crop growing season.

2.5. Crop Redistribution Model

2.5.1. Model Framework

This study developed a multi-objective crop redistribution model based on NSGA-II (Non-dominated Sorting Genetic Algorithm II) [36], which can optimize all the objectives simultaneously and generate a pareto front consisting of non-dominated solutions. The decision variables in the model are defined as the area proportions allocated to rice, maize, and soybean in each county, with a total of 819 such variables. The four objectives are listed as follows:

• Maximize total crop production;
• Minimize total production losses;
• Minimize total water deficit;
• Minimize total nitrogen fertilizer application.

Objective functions are defined as follows:

$$max\left(T_1\right)=max{\sum }_{i=1}^N{\sum }_{j=1}^n{A}_i{x}_{ij}{Y}_{ij}$$

(8)

$$min\left(T_2\right)=min{\sum }_{i=1}^N{\sum }_{j=1}^n{A}_i{x}_{ij}{Y}_{ij}{R}_{ij}$$

(9)

$$min\left(T_3\right)=min{\sum }_{i=1}^N{\sum }_{j=1}^n{A}_i{x}_{ij}{WD}_{ij}$$

(10)

$$min\left(T_4\right)=min{\sum }_{i=1}^N{\sum }_{j=1}^n{A}_i{x}_{ij}{N}_{ij}$$

(11)

where T₁, T₂, T₃ and T₄ are total crop production, total production losses, total water deficit, total nitrogen fertilizer application after crop redistribution, respectively, A_i is the total cropland area in county i, x_ij is the area proportion allocated to crop j in county i, and Y_ij, R_ij, WD_ij, and N_ij are the average yield, average yield loss rate, average water deficit, and average nitrogen fertilizer application rate of crop j in county i, respectively. The year 2015 was selected as the optimization baseline due to data completeness. Constraints were applied such that all four objectives after optimization must outperform their pre-optimization values. In addition, sum of the three decision variables within a county must lie within [0, 1]:

$$s.t.\left\{\begin{array}{c}{T}_1>T_{1before}\\ T_2<T_{2before}\\ T_3<T_{3before}\\ T_4<T_{4before}\\ 0\le {\sum }_{i=1}^N{x}_{ij}\le 1\end{array}\right.$$

(12)

where T_{1 before}, T_{2 before}, T_{3 before}, T_{4 before} are total crop production, total production losses, total water deficit, and total nitrogen fertilizer application in 2015, respectively.

2.5.2. Selection of the Optimal Solution from Pareto Front

The crop redistribution model generates a Pareto front containing 2000 non-dominated solutions of the problem. To select the optimal solution, we adopted the approach proposed by Xie et al., which prioritizes both individual objective performance and balance among objectives [21]. After normalizing the objective values, the optimal solution is selected using the following criterion, which optimizes all objectives such that the improvement in all objectives is as high as possible while their between-dimension differences are as low as possible:

$$G={\displaystyle \frac{Aver\left(N_d\right)}{Var\left(N_d\right)}}$$

(13)

$$N_d=\left|1-{\displaystyle \frac{T_d}{T_{dbefore}}}\right|\times 100\%$$

(14)

where a higher G indicates better overall performance and smaller disparity among objectives, N_d is the normalized value of objective d, T_d and T_{d before} are the objective values after and before crop redistribution of objective d. The solution with the highest G value is selected as the optimal solution.

2.6. Calculation of Geographical Centroid

To assess the impact of crop redistribution on crop planting structure, a geographical centroid model was applied in the study, following the approach of Fan et al. [37]. First, coordinates (longitude and latitude) of the geographical centroids in 273 counties were extracted using ArcMap 10.8 based on the shape and area of each county. Then, a weighted average of the counties’ centroid coordinates was calculated, using their crop planting areas as weights. We use the method to estimate the centroids for each crop and year in both historical and optimized crop distribution. The equations are listed as follows:

$$X_t={\displaystyle \frac{{\sum }_{i=1}^n{(A}_{i,t}\times X_i)}{{\sum }_{i=1}^n{A}_{i,t}}}$$

(15)

$$Y_t={\displaystyle \frac{{\sum }_{i=1}^n\left(A_{i,t}{\times Y}_i\right)}{{\sum }_{i=1}^n{A}_{i,t}}}$$

(16)

where X_t and Y_t are the longitude and latitude of the geographical centroid in year t, X_i and Y_i are the coordinates of the geographical centroids in county i, and A_i,t is the crop area in county i during year t (for t = 2001, 2005, 2010, 2015, 2020).

3. Results

3.1. Temporal Trends in Yield, Yield Loss Rate, Water Deficit, and Nitrogen Fertilizer Application Rate for Rice, Maize, and Soybean Across Counties in Northeast China

Figure 3 illustrates the temporal variations in county-level yield, yield loss rate, water deficit, and nitrogen fertilizer application rate for rice, maize, and soybean in Northeast China from 2001 to 2020. In terms of yield (Figure 3a–c), rice and maize showed relatively high average yields of approximately 6–8 t/ha and 5–7 t/ha, respectively, while soybean maintained a comparatively lower average yield of around 1–3 t/ha. Over the study period, yields of the three crops exhibited increasing trends at annual increases of 0.039 t/ha (p < 0.01), 0.036 t/ha (p < 0.01), and 0.013 t/ha (p < 0.1), respectively. Regarding yield loss rate (Figure 3d–f), rice suffered less yield loss than maize and soybean. From 2001 to 2020, the yield loss rates for the three crops declined at rates of 0.16%/a (p < 0.1), 0.28%/a (p < 0.05), and 0.24%/a (p < 0.1), respectively. As shown in Figure 3g–i, rice experienced significantly higher levels of water deficit compared to maize and soybean. During the study period, average water deficit decreased at rates of 2.53 mm/a (p > 0.1), 1.30 mm/a (p < 0.1), and 1.75 mm/a (p < 0.1) for rice, maize, and soybean, respectively. In terms of nitrogen fertilizer application rate (Figure 3j–l), maize required substantially more N fertilizer than rice and soybean, with soybean being the least fertilizer-intensive crop. Between 2001 and 2018, the N fertilizer intensity for rice and maize increased annually by 0.02 kg/ha (p < 0.01) and 0.10 kg/ha (p < 0.01), respectively, while for soybean it decreased by 0.004 kg/ha per year (p < 0.05).

This analysis highlights the distinct characteristics of the three crops across four optimization objectives: rice is characterized by high yield and relatively low N input but suffers from high water deficit; maize achieves high yield and low water deficit but requires substantial N input; and soybean has the lowest yield yet benefits from the lowest fertilizer requirements. These complementary characteristics of crops form a foundation for multi-objective crop redistribution.

3.2. Spatial Distribution Patterns of Yield, Yield Loss Rate, Water Deficit, and Nitrogen Fertilizer Application Rate in Northeast China

Figure 4 presents the spatial distribution of multi-year average values of yield, yield loss rate, water deficit, and nitrogen fertilizer application rate for rice, maize, and soybean across counties in Northeast China. The spatial patterns of yield (Figure 4a–c) are similar across all three crops. High-yield areas are mainly located in southern Heilongjiang, central-western Jilin, and northern Liaoning, whereas low-yield zones are primarily found in northern Heilongjiang, eastern Jilin, and the Four Eastern Leagues. Regarding average yield loss (Figure 4d–f), high-loss zones for rice are concentrated in southern parts of the Four Eastern Leagues and eastern Jilin; for maize, they are found in northwestern Heilongjiang, northern parts of Four Eastern Leagues, western Jilin, and western Liaoning; for soybean, they are concentrated in southwestern Heilongjiang, western Jilin, central parts of Four Eastern Leagues, and parts of western Liaoning. The spatial patterns of water deficit (Figure 4g–i) are similar among the three crops, with high-deficit regions primarily located in southwestern Heilongjiang, western Jilin, western Liaoning, and southern parts of Four Eastern Leagues. For rice, most counties—except for those in central Heilongjiang, eastern Jilin, and eastern Liaoning—experience insufficient precipitation, requiring supplemental irrigation. For maize and soybean, precipitation generally meets water demands in most counties, except in southwestern Heilongjiang, western Jilin, western Liaoning, and southern parts of Four Eastern Leagues. Regarding nitrogen fertilizer use (Figure 4j–l), maize and rice exhibit similar spatial patterns, with high application rates observed in western parts of Heilongjiang, Jilin, and Liaoning. Soybean fertilizer use is relatively low overall, with higher values mainly observed in counties within Heilongjiang.

3.3. Selection of Optimal Solution and Evaluation of Optimization Benefits

The crop redistribution model generates a pareto front, which consists of 2000 non-dominated solutions. Figure 5 and

Table 2. Optimization effects (absolute quantities) of crop redistribution when four objectives are optimized simultaneously or individually.

Solutions	Crop Production (106 t)	Production Losses (105 t)	Water Deficit (107 m3)	N Fertilizer (106 kg)
optimal solution	2.41	−1.84	−38.75	−54.09
max production	4.20	−0.10	−0.42	−50.15
min losses	0.16	−4.17	−30.98	−74.05
min water deficit	0.009	−2.65	−96.91	−67.58
min fertilizer use	0.008	−2.65	−7.62	−93.54

illustrate performance of all the solutions in the pareto front on the four objectives, as well as the maximum optimization range of each objective. We can observe obvious trade-off between total production and three other objectives, which demonstrates the conflicts between food security and sustainable development goals. When each objective is optimized individually under the constraint that others objectives do not deteriorate, the total crop production can be increased by up to 4.20 × 10⁶ t, production losses reduced by up to 4.17 × 10⁵ t, water deficit reduced by up to 9.69 × 10⁸ m³, and nitrogen fertilizer application reduced by up to 9.35 × 10⁷ kg (Table 2). When using a balanced strategy to select the optimal solution from pareto front, production in Northeast China can be increased by 2.41 × 10⁶ t, while production losses, water deficit and fertilizer application can be reduced by 1.84 × 10⁵ t, 3.88 × 10⁸ m³ and 5.41 × 10⁷ kg, respectively (Table 2). Although less competitive on single objective, the best solution that balances all the objectives outdid other solutions on co-benefits across all four dimensions.

Based on the optimal solution we selected, we analyzed the effectiveness of crop redistribution in Northeast China on three crops across four dimensions.

Table 3. Optimization effects of crop redistribution on four optimization objectives for rice, maize and soybean in Northeast China.

Crop	Crop Production		Production Losses		Water Deficit		N Fertilizer
	Percentage (%)	Absolute (106 t)	Percentage (%)	Absolute (105 t)	Percentage (%)	Absolute (107 m3)	Percentage (%)	Absolute (106 kg)
rice	63.29	17.49	77.17	7.26	3.73	9.65	45.72	15.77
maize	−11.3	−12.31	−13.33	−7.40	−18.84	−56.91	−13.69	−61.84
soybean	−51.08	−2.78	−50.46	−1.69	−85.08	8.51	−69.4	−8.03
total	1.7	2.41	−2.69	−1.84	−6.78	−38.75	−10.87	−54.09

presents the relative and absolute changes in the four objectives following crop redistribution. When considering the three crops as a whole, total crop production in Northeast China increased by 1.70%, while production losses, water deficit, and nitrogen fertilizer application decreased by 2.69%, 6.78%, and 10.87%, respectively. Then we analyzed the contribution of each crop to the improvement of the four objectives. In terms of crop production, crop redistribution led to an increase of 63.29% (1.75 × 10⁷ t) in rice production, whereas total maize and soybean production declined by 11.3% (1.23 × 10⁷ t) and 51.08% (2.78 × 10⁶ t), respectively. The overall increase in crop production was primarily driven by the rise in rice production. Regarding production losses, rice experienced an increase of 77.17% (7.26 × 10⁵ t), whereas losses in maize and soybean production decreased by 13.3% (7.40 × 10⁵ t) and 50.46% (1.69 × 10⁵ t), respectively. The reduction in overall production losses was mainly attributable to the decreased losses in maize and soybean.

For water deficit, total rice water deficit increased by 3.73% (9.65 × 10⁷ m³), maize water deficit decreased by 18.84% (5.69 × 10⁸ m³), and soybean water deficit rose by 85.08% (8.51 × 10⁷ m³). The reduction in total water deficit was primarily due to the decline in maize water deficit. In terms of nitrogen fertilizer application, rice required 45.72% (1.58 × 10⁷ kg) more nitrogen, while maize and soybean nitrogen use decreased by 13.69% (6.18 × 10⁷ kg) and 69.40% (8.03 × 10⁶ kg), respectively. The reduction in total nitrogen fertilizer application in Northeast China was mainly driven by the decrease in maize fertilization, followed by soybean.

3.4. Changes in Cropping Structure at Different Spatial Scales in Northeast China After Crop Redistribution

Figure 6 illustrates the crop area before and after crop redistribution in Northeast China and its four administrative subregions.

Table 4. Percentage change in crop area in Northeast China and its provinces after crop redistribution.

Crops	Northeast China	Heilongjiang	Jilin	Liaoning	Four Eastern Leagues
rice	69.58%	46.45%	62.74%	83.46%	611.45%
maize	−12.8%	2.0%	−19.36%	−33.69%	−18.78%
soybean	−54.79%	−71.28%	47.69%	323.22%	−82.73%

presents the percentage changes in the area of each crop following the optimization. For the whole region, the crop redistribution led to an increase of 69.58% in rice planting area in Northeast China, while the areas devoted to maize and soybean decreased by 12.8% and 54.79% compared with their respective areas before optimization (Table 4). Despite these adjustments, maize remained the dominant crop in the region before and after crop redistribution, with its planting area exceeding that of rice and soybean (Figure 6).

At the provincial level, all four subregions experienced an increase in rice area, with Heilongjiang, Jilin, Liaoning, and the Four Eastern Leagues showing respective increases of 46.45%, 62.74%, 83.46%, and 611.45%. Except for a slight increase of 2.0% in maize area in Heilongjiang, the other three regions saw reductions in maize cultivation, with Jilin, Liaoning, and the Four Eastern Leagues experiencing decreases of 19.36%, 33.69%, and 18.78%, respectively. As for soybean, the area increased by 47.69% in Liaoning and 323.22% in Jilin while declining by 71.28% and 82.73% in Heilongjiang and the Four Eastern Leagues, respectively (Table 4).

Figure 7 presents the spatial distribution of percentage changes in crop planting area proportions at the county level in Northeast China after crop redistribution. Rice area increased in 189 out of 273 counties. Reductions in rice area were mainly concentrated in southwestern Heilongjiang, western Jilin, and the southern parts of the Four Eastern Leagues. Maize area increased in 57 counties and decreased in 216 counties, with reductions widespread except in northwestern and central-western Heilongjiang and northern Jilin. Soybean area increased in 171 counties and decreased in 102 counties, with gains concentrated in central-western Jilin and most of Liaoning. Although the majority of the counties experienced an increase in soybean area, the total soybean planting area decreased compared with the area before crop redistribution.

3.5. Shifts in the Planting Area Centroids of Rice, Maize and Soybean After Crop Redistribution

Figure 8 illustrates the historical shifts (2001–2020) and post-optimization positions (based on 2015 data) of the geographic centroids of rice, maize, and soybean planting areas in Northeast China. Historically, the centroid of rice planting area gradually shifted from the southwest to the northeast, moving from northern Jilin in 2001 to south-central Heilongjiang in 2020. The maize area centroid moved southwest from 2001 to 2005, northeast from 2005 to 2015, and southeast from 2015 to 2020, ultimately shifting from southwestern to northwestern Jilin. Soybean’s centroid remained within Heilongjiang, shifting northeast during 2001–2010, northwest during 2010–2015, and northeast again during 2015–2020. After crop redistribution, spatial shifts in crop area centroids were more pronounced: the rice centroid moved northwest (from 127.23° E, 45.40° N to 126.73° E, 45.76° N), maize continued its northeastward shift (from 125.08° E, 45.01° N to 125.34° E, 45.36° N), and soybean’s centroid shifted markedly southwest (from 126.73° E, 47.59° N to 125.78° E, 44.90° N), compared with the positions of the centroids in 2015.

4. Discussion

4.1. Potential of Crop Redistribution in Northeast China

Our study demonstrates the potential of crop redistribution to increase crop production and reduce production losses in Northeast China, even within the constraints of existing cropland area and productivity levels. Under the scenario focused solely on maximizing crop production, total crop production in the region could increase by up to 2.96%. In the multi-objective optimization scenario, production gains could reach 1.70%. These figures are lower than those reported by Davis et al. on a global scale (10% increase in caloric supply and 19% in protein supply) [13] and by Wang et al. in China (34% increase in protein production) [38]. Although the relative increase is smaller, the absolute gains in grain output are still substantial. The primary driver of increased grain production under the optimized distribution is the enhanced production of rice. Moreover, crop redistribution significantly reduced total production losses (6.10% under single-objective, 2.69% under multi-objective optimization), supporting its role as a disaster mitigation strategy, consistent with previous research [16]. On the one hand, the decline in such losses indicates that crop redistribution mitigates the adverse impacts of meteorological hazards on crop production; on the other hand, it also contributes to a certain degree of reduction in pest and disease pressures.

The optimization also significantly reduced total water deficit in the region. When water deficit was the only objective, it decreased by up to 16.97%, and by 6.78% under the multi-objective scenario. This reduction could alleviate irrigation pressure and conserve water resources, which is of great significance in the context of increasing drought risk under future climate change. Notably, the crop redistribution led to a 3.73% increase in total water deficit for rice, yet rice production rose by 63.29%. This indicates that boosting rice production does not necessarily come at the expense of increased water stress. The main contributor to the overall water deficit reduction was the decrease in water stress for maize. Although maize area declined by 12.8%, its water deficit dropped by 18%, indicating both a reduction in maize cultivation and a shift to less water-stressed areas.

In addition to enhancing production and alleviating water stress, the optimization also lowered nitrogen fertilizer use. Total nitrogen fertilizer application was reduced by up to 18.8% in the single-objective optimization and by 10.87% under multi-objective optimization. These results align with those of Xie et al., who reported a nationwide nitrogen fertilizer reduction of 8.1–12.0% in China [21]. In Northeast China, the reduction is mainly attributed to decreased N fertilizer use in maize. Overall, lower fertilizer input could reduce greenhouse gas emissions and prevent soil degradation, contributing to sustainable agriculture.

4.2. Rationality of the Recommended Strategy for Crop Redistribution in Northeast China

Our results show that crop redistribution increased rice planting area by 69.58%, while reducing maize and soybean planting areas by 12.8% and 54.79%, respectively. The reduction in maize area is in line with China’s recent maize reduction policies [39]. Wu et al. suggest that, in scenarios prioritizing economic returns, expanding more profitable crops like rice is advisable [40]. Qi et al. argue that, under conditions of adequate rainfall and efficient irrigation, rice area in Northeast China could be increased by up to 13.7%, supporting our findings [41]. However, both studies emphasize that in water-scarce scenarios or when pursuing environmental benefits, rice area should be limited due to its higher environmental costs. For instance, methane emissions from rice cultivation are substantially higher than those from maize or soybean. Consequently, shifting cultivation from maize or soybean to rice would result in additional greenhouse gas emissions. On the other hand, although rice area expanded, in our study, its total water deficit only slightly increased by 3.73%, suggesting that area expansion is feasible under current water conditions. Although our results led to a reduction in soybean planting area, the importance of soybeans should not be overlooked. As China’s soybean supply has been relying on the importing from international market, the expansion of soybean area will improve China’s soybean self-sufficiency. Meanwhile, increasing soybean cultivation could also contribute to reducing nitrogen fertilizer use [40,42]. Given soybeans’ relatively lower yield compared to other major crops, government subsidies could be an effective way to encourage the production of soybean.

From a regional perspective, in Heilongjiang Province, rice and maize areas should increase by 46.45% and 2.0% respectively, while soybean area should decrease by 71.28%. This contrasts with Hu et al., who suggested reducing both maize and rice areas and increasing soybean area in Heilongjiang Province. However, their study mainly focuses on environmental benefits (carbon emissions and water footprint) and optimizing crop distribution at the expense of economic benefits [22]. On the other hand, the study by Li et al. suggests that rice area should be expanded while soybean area should be reduced in the Songnen Plain and the Sanjiang Plain, the major grain-producing areas of Northeast China, which is consistent with the conclusions of this study [43]. Overall, the proposed crop redistribution strategy in this study is considered rational.

4.3. Uncertainty and Future Development of the Crop Redistribution Model

This study identified clear trade-offs among the four optimization objectives. Specifically, increasing total crop production often came at the cost of increased production losses, water stress, or fertilizer application—similar to the findings of Hu et al. [22]. To address these trade-offs among the four optimization objectives, we adopted the approach proposed by Xie et al. in our study, selecting the solution with the highest overall synergy from the Pareto front as the final recommendation [21]. The choice of an optimal solution remains inherently flexible and subjective, depending on decision-makers’ preferences and practical needs.

This study conducted crop redistribution in Northeast China using counties as the smallest optimization units. The advantage of using the county level lies in its closer alignment with real-world agricultural situation, as there tends to be consistency within a county in terms of agricultural policies, management practices, and crop cultivars. However, it also limits spatial precision, as within-county variability in topography, hydrology, soil, and management practices is not captured. Future research should incorporate high-resolution gridded data to enable finer-scale optimization and generate more detailed crop redistribution scheme.

This study focused only on the three staple crops in Northeast China—rice, maize, and soybean. Other food crops, such as wheat, were not included in the crop redistribution model due to their relatively small share of the total cultivated area in Northeast China [44]. Similarly, minor or cash crops like sorghum, tobacco and vegetables were not included due to poor data availability. Otherwise, factors such as production costs, crop prices, farmers’ expertise, soil types, infrastructure, facilitation, and agricultural subsidies—which play critical role in shaping crop distribution—were not fully incorporated into this study. These limitations introduce uncertainty into the optimization outcomes and constrain the practical applicability of the crop redistribution model. Future efforts should focus on collecting relevant data on these crops and improving realism and comprehensiveness of the crop redistribution model.

Lastly, farmers may face some practical challenges in adopting the crop redistribution strategies. For instance, changes in crop types in a specific region may result in switching costs. These transformations include adjustments in seed use, machinery, and farming practices, which were not accounted for in our study. Although crop redistribution may achieve long-term benefits at a macro perspective of the whole region, it might pose short-term challenges for small-holders by increasing their production costs. These factors should be incorporated in future studies to enhance the practical applicability of the crop redistribution scheme.

5. Conclusions

This study developed a crop redistribution model for Northeast China using the multi-objective evolutionary algorithm NSGA-II. Leveraging meteorological data, crop yield statistics, and fertilizer application data, we jointly optimized four objectives: crop production, production losses, water deficit, and nitrogen fertilizer application. The model optimized the spatial distribution of rice, maize, and soybean across 273 counties in Northeast China, assessing optimization potential and generating optimal distribution under current agricultural conditions.

The results reveal significant potential for crop redistribution in the region. Under single-objective scenarios, total crop production could increase by up to 2.96%, production losses decrease by 6.10%, water deficit reduce by 16.97%, and nitrogen fertilizer application decline by 18.80%. Under multi-objective optimization, simultaneous improvements can be achieved: production increases by 1.70%, production losses drop by 2.69%, water deficit by 6.78%, and N fertilizer application by 10.87%.

The result also shows that: after crop redistribution, rice area increases by 69.58%, maize area reduces by 12.8%, and soybean area reduces by 54.79% in Northeast China compared with 2015 planting areas. The centroid of rice planting area moves northwest, maize northeast, and soybean southwest. Specifically, rice area reduces in southwestern Heilongjiang, western Jilin, and the southern parts of the Four Eastern Leagues, and it increases elsewhere. Maize area expands in northwestern and central-western Heilongjiang and northern Jilin, and it reduces elsewhere. Soybean area increases in central-western Jilin and most of Liaoning, and it decreases in other counties. It should be noted that, given the exclusion of relevant factors from the crop redistribution model, the outcomes should be interpreted with caution. Further decisions regarding crop redistribution need to integrate model results with additional considerations, including production costs, soil characteristics, infrastructure, and other factors.

Future work will improve the model’s spatial resolution, expand crop types, and account for crop switching costs, thereby enhancing its reliability and applicability. The findings of this study can serve as a valuable reference for crop redistribution in Northeast China.