Drought stress has become one of the main abiotic constraints restricting rice production. Applying drought - resistant functional fertilizer is a key way to ease this stress. In this pot experiment, “Nanjing 46” rice variety was chosen. Two basal fertilizers (conventional and drought - resistant functional, both at 40 kg/hm²) and three stress levels (mild, moderate, severe) were set, with a normally irrigated treatment as the control (CK), totaling eight treatments. Indicators including rice growth, yield, quality, and economic benefits were monitored and analyzed. Results showed that: ① Compared to the CK treatment, under various drought stresses, the drought-resistant functional fertilizer had a better alleviating effect. In different drought stress levels, the reductions in plant height, tiller number, dry matter accumulation, and yield were 1.98%~4.26%, 3.57%~16.75%, 3.03%~41.87%, and 0.50%~15.63% respectively. ② Under various drought stresses, the drought-resistant functional fertilizer treatment increased tiller number by 0.80%~5.48%, total dry matter accumulation by 6.57%~21.65%, and root, stem, and leaf dry weights by 6.56%~22.66%, 5.52%~24.15%, and 8.38%~20.08% respectively. The effective panicle number, grains per panicle, and 1 000-grain weight increased by 0.38%~0.84%, 0.40%~0.75%, and 0.87%~2.29%, leading to a 2.54%~5.29% yield increase. ③ The alleviation degree of the drought-resistant functional fertilizer varied under different drought stresses. Generally, as drought stress increased, the fertilizer's alleviating effect on rice growth and yield strengthened. In conclusion, the drought-resistant functional fertilizer can effectively mitigate the negative impacts of drought stress during the booting stage and also has a yield-increasing effect under normal irrigation conditions.

This study addresses water scarcity and low nitrogen use efficiency in winter wheat production across the Huang-Huai-Hai Plain. Based on climate, soil, and field management data from 1961 to 2014, the region was divided into six representative subregions and further subdivided into 613 simulation units. A regionalized DSSAT model was constructed and calibrated, and a genetic algorithm was employed to systematically optimize irrigation and nitrogen management strategies under different hydrological year types. The results show that under typical wet, normal, and dry years, the optimal total irrigation volumes were 945、 1 181 and 1 830 m3/hm², respectively, while optimal nitrogen rates clustered at 170~190 kg/hm². The corresponding mean yields reached 6 385、 6 384 and 6 073 kg/hm²; mean water use efficiencies were 1.94、 1.94 and 1.64 kg/m3; and mean nitrogen partial factor productivities were 37.5、 36.8 and 35.3 kg/kg, respectively. From these results, 18 optimal water-nitrogen management schemes were developed for different hydrological year types and subregions, enabling precise water and nitrogen application at key stages like green-up, jointing, and grain-filling. The proposed optimized strategies simultaneously balance high yields, resource efficiency, and water and nitrogen savings under varied hydrological scenarios, providing a scientific basis for decision-making in precision irrigation and fertilization in the region.

The progressive study of the spatial distribution of crop planting, the statistics of water demand and the impact of drought and flood disasters is of great significance for the modern agricultural regulation systems. In this study, a classification model based on multi-source remote sensing data was constructed to extract the high-precision planting area information for early rice and late rice in the Wuhua irrigation area. The study analyzed the temporal dynamic characteristics of crop water consumption and the crop water surplus and deficit index (CWSDI) during the growth periods of early and late rice in the five years from 2019 to 2023, and quantitatively evaluated the risk of drought and flood water stress at different growth stages of the double-cropping rice. The results showed that there were differences in the planting areas of early rice and late rice in the Wuhua irrigation area. The peak water requirement of early rice occurred during the tillering-jointing stage, while the peak water for late rice was at the booting-heading stage, and the daily crop water requirement (ET _c) of late rice was lower than that of early rice. During the five-year period, the year with the lowest ET _c for early rice was 2019, which was affected by frequent precipitation from typhoons. The highest ET _c for late rice occurred in 2022, which corresponded to the high temperatures and drought in autumn of that year. The daily average CWSDI from 2019 to 2023 for early and late rice were 0.80 and 0.04, respectively, indicating that extreme precipitation events in this region occurred in the yellow maturity stages of early and late rice. The proportion of severe drought and severe flood in the growth period of early rice was 64.6% and 86.4%, respectively, indicating that early rice was susceptible to the dual stress of drought and flood. The frequency of drought for late rice was 6% higher than that for early rice, and the proportion of severe drought reached 75.9%, indicating that late rice faced a more severe risk of water shortage during the growth period. While waterlogging events were concentrated over short periods, the overall duration is short. The CWSDI was positive overall; however drought occurred more frequently.

Soil salinization is a prevalent ecological issue in arid and semiarid irrigation districts, and an accurate understanding of the spatial distribution of soil salinity is crucial for agricultural production and ecological conservation. This study focused on the Hetao irrigation district and used Sentinel-1 and Sentinel-2 remote sensing data. Three variable optimization methods—Pearson correlation coefficient, ridge regression coefficient, and variable importance in projection (VIP)—were employed to construct four machine learning models: a back propagation neural network (BPNN), support vector machine (SVM), random forest (RF), and extreme gradient boosting (XGBoost). These models were utilized for the quantitative inversion of soil salinity. The Shapley additive explanations (SHAP) method was used to determine the impact of each input variable on the inversion result of the optimal model. Finally, the spatial distribution of soil salinity in the HID was revealed using the best inversion model. The results indicated that the RF model based on the VIP optimization method performed the best in soil salinity inversion, achieving an R ² of 0.88 and an RMSE of 2.82 g/kg. The salinity indices SI-T, NDSI, and Aster-SI and the polarization combination index PCI3 had significant impacts on the optimal model inversion. These variables had negative impacts on the inversion results when the soil salinity was low, but their impacts became positive when the soil salinity exceeded a certain threshold. The inversion results of the optimal model indicate that moderately saline soil was the most widespread in the Hetao irrigation district, followed by severely saline soil. Nonsaline and saline soils occupied relatively limited area.

This study focuses on diamond shaped drip irrigation emitters with diamond-shaped flow channels, employing a combined approach of numerical simulation and machine learning to investigate the relationship between flow channel structural parameters and the flow index, and to establish a corresponding regression prediction model. The research first utilized SolidWorks for 3D modeling of the flow channels and Fluent fluid simulation software to determine the reasonable ranges of values for each structural parameter, subsequently generating 125 sets of experimental sample data based on an orthogonal experimental design. For numerical simulation, the k-ε turbulence model was used to analyze flow characteristics, while for machine learning modeling, the Random Forest (RF) algorithm was employed and combined with Particle Swarm Optimization (PSO) for parameter optimization. The results demonstrate that with an 80% training set proportion, the PSO-RF model achieves an optimal balance between learning capability and generalization performance, attaining a test set coefficient of determination (R2) of 0.854 8 and significantly reduced error metrics, outperforming the traditional RF model in both prediction accuracy and stability. Parameter sensitivity analysis revealed that the descending order of influence of flow channel structural parameters on the flow index was: a (inlet width) > c (horizontal distance of the front inlet) > b (horizontal width of the baffle) > d (vertical distance from the top of the inlet to the uppermost point). These findings provide both a novel approach for applying machine learning to drip irrigation emitter optimization design, as well as a theoretical basis and technical support for optimizing emitter flow channel structural parameters.

In agricultural suspended sprinkler irrigation systems, there is a dynamic coupling between pressure and flow rate, which is influenced by various factors and exhibits nonlinear, time-varying, and time-delay characteristics. Traditional PID control struggles to effectively address these issues, resulting in large overshoots and slow regulation. Therefore, a pressure-flow adaptive PID decoupling control algorithm is proposed. The method uses a nonlinear filter to smooth the flow signal, designing an adaptive fuzzy PID controller to achieve preliminary control, and then introducing a fuzzy decoupling controller for decoupling compensation. Experimental results show that this method has a fast response, an overshoot of less than 10%, precise control, and stable decoupling performance.

This study investigates the role of a rainfall harvesting and infiltration irrigation system in utilizing light rainfall resources in the Hongmeixing (Prunus armeniaca) economic forest in Pengbao Town, Yuanzhou District, Guyuan City, Ningxia Hui Autonomous Region. The aim is to define the characteristics of light rainfall in the local Hongmeixing economic forest and provide a scientific basis for quantitative research on its utilization and for improving the efficiency of local rainfall resource use. With Hongmeixing trees as the research subject, the CASC2D model was employed to simulate light rainfall runoff processes and evaluate the collection efficiency of the rainfall harvesting and infiltration irrigation system. From April to July, a total of 30 rainfall events were observed at the Hongmeixing base in Pengbao, with a cumulative rainfall of 154.1 mm. The average rainfall per event was 5.14 mm. The collection efficiency of the infiltration irrigators ranged from 51.64% to 86.6%, with an average efficiency of 65.5%. The rainfall harvesting and infiltration irrigation system demonstrated high collection efficiency. There was a clear positive correlation between collected water volume and rainfall amount, as well as enhanced collection capacity under higher rainfall intensity. The collection efficiency increased continuously with increasing rainfall but at a diminishing rate, eventually stabilizing. Analysis of rainfall collection across different rainfall magnitudes revealed that the system exhibited a significant upward trend in collection efficiency as rainfall level increased.

Understanding the mechanisms of water and salt transport in saline soils under mildly saline and saline water irrigation is crucial for conserving freshwater resources, advancing the theoretical framework for the safe use of saline water, and fostering sustainable agricultural development. To analyze the effects of different types, concentrations and sources of salts on the soil water and salt transport process and water and salt distribution, three infiltration scenarios were set up, that is, NaCl/MgCl₂ solution + non-saline soil, distilled water + NaCl/MgCl₂ saline soil, and distilled water + non-saline soil. The concentrations of the salt solutions were 2、6 and 30 g/L, and the concentrations of the saline soils were 0.2% and 0.5%. A one-dimensional constant-head infiltration experiment was conducted to investigate the water infiltration characteristics and water and salt distribution under different scenarios. Results showed that: ① In the "salt solution + non-saline soil" scenario, low-concentration salt water infiltration could accelerate the movement rate of the wetting front, and the promoting effect of MgCl₂ solution on the movement rate of the wetting front was greater than that of NaCl solution. 2~30 g/L MgCl₂ solution could effectively reduce the amount of soil water infiltration, and increasing the concentration of MgCl₂ could enhance the blocking effect on soil water infiltration.The duration of infiltration was reduced by approximately 29%~38%. However, in the "distilled water + saline soil" scenario, at a certain concentration, NaCl and MgCl₂ saline soil could promote soil water infiltration, and the promoting effect of MgCl₂ saline soil on water and salt transport was stronger. The mean infiltration duration decreased by 72.45% compared with the NaCl saline soil. ② After infiltration, the soil profile moisture content in both infiltration scenarios showed a trend of the highest moisture content in the surface soil and a decrease with depth. ③ In the "salt solution + non-saline soil" scenario, the total soil salt content and salt ion concentration after infiltration were the highest in the surface soil and decreased with depth. However, in the "distilled water + saline soil" scenario, the total soil salt content and salt ion concentration were the lowest in the surface soil and increased with depth.

To explore the effects of Yellow River sediment combined with amendments on soil water-salt dynamics and winter wheat yield in coastal saline-alkali soil, and to provide a scientific basis for soil improvement in the Yellow River Delta. A field split-plot experiment with six treatments was conducted: T1 (control), T2 (bio-organic fertilizer), T3 (alkaline soil conditioner), ST1 (sediment), ST2 (sediment + bio-organic fertilizer), and ST3 (sediment + conditioner). Soil water-salt parameters, wheat growth, and yield were analyzed. Compared with T1, ST3 reduced electrical conductivity (EC) in the 0–20 cm layer by 27.93%, increased water content in the 20–40 cm layer by 5.30%, and decreased sodium adsorption ratio (SAR) and pH by 32.19% and 2.49%, respectively. ST3 achieved the highest yield (8.53 × 103 kg/hm²), showing a 25.07% increase over T1. Yield exhibited significant negative correlations with soil salinity, pH and SAR (p<0.05), and a highly significant positive correlation with spike number (r=0.948). The combined application of Yellow River sediment and alkaline soil conditioner (ST3) effectively alleviates saline-alkali stress by improving soil structure, enhancing salt leaching, and regulating ion balance, offering a practical solution for coastal saline-alkali soil improvement and wheat yield enhancement.

As a crucial region for the Shaanxi section of the South-to-North Water Diversion Project's Western Route, the Jinghe River Basin, located in an arid and semi-arid area, has long faced water shortages that constrain its high-quality development. The Jinghe River Basin is a traditional agricultural production area and an important grain base in China, where crops rely on irrigation because of the arid and semi-arid climate. Clarifying the spatiotemporal variation of irrigation water requirements for the main grain crops, winter wheat and summer maize, is of great significance for the future construction of high-standard farmland and the planning of related water conservancy facilities. Based on observation data from 22 meteorological stations in the Jinghe River Basin from 1990 to 2022, we calculated the irrigation water requirements of the main crops by using the Penman-Monteith formula, effective rainfall, and crop coefficients. The results show that the annual irrigation water requirements of both winter wheat and summer maize generally exhibited a fluctuating trend, with change rates of +11.99 mm/10 a and -1.0 mm/10 a, respectively, during the study period. The irrigation water requirement for winter wheat was the largest in May (approximately 99.25 mm), and that of summer maize was the largest in July (about 51.52 mm). The irrigation water requirement for winter wheat showed a gradually increasing trend from the southeast to the northwest, while that of summer maize exhibited a decreasing trend from the southeast and northwest regions towards the interior of the study area. Temperature and precipitation are the basic meteorological elements affecting the changes in irrigation water requirements for main grain crops in the Jinghe River Basin. Simultaneously, vapor pressure deficit and solar radiation are the key elements affecting the irrigation water requirement, with correlation coefficients generally around 0.80. During the study period, the total irrigation water requirement in the Jinghe River Basin showed a clear downward trend due to the reduction of cultivated land area, with a multi-year average value of approximately 7.025 billion m3. In the future, with the in-depth construction of high-standard farmland and the completion of the Shaanxi section of the South-to-North Water Diversion Project's Western Route, it is expected that the irrigation water demands for the main grain crops in the Jinghe River Basin will be fully met.

Climate change and intensifying human activities pose significant challenges to the sustainable management of water resources. Therefore, predicting the status of water resource utilization under the combined influence of climate change and human activities is crucial for sustainable regional water resource management. By coupling the water resource ecological footprint with a system dynamics model, we established a water resource ecological footprint system dynamics model for Gansu Province. This model considers the impacts of climate change and human activities and incorporates 16 development scenarios designed based on the specific conditions of the study area. It simulates the level and sustainability of water resource utilization in Gansu Province from 2024 to 2040. The results indicate that: During the prediction period, water resources are projected to be in deficit under all 16 scenarios, with this deficit gradually intensifying. The mean values of the water resource ecological pressure index are all projected exceed 10.000, signifying an unsafe level of utilization. Furthermore, the ecological-economic coordination of water resources remains poor. However, the water resource ecological footprint of 10⁴ RMB GDP shows a year-on-year decrease, indicating continuous improvement in water resource utilization efficiency. Analysis using the Tapio decoupling model reveals that the relationship between water resource ecological footprint consumption and socio-economic development primarily exhibits weak decoupling in most years, reflecting a state of sustainable and coordinated development. Comparative analysis identifies Scenario S₄ as having development indicators conducive to achieving sustainable water resource utilization. This scenario balances socio-economic development with water environmental protection, representing the most suitable future development pathway for Gansu Province. Building upon the reference Scenario S₄, achieving sustainable water resource development will require measures such as enhancing water conservation awareness, innovating wastewater treatment technologies, prioritizing the use of unconventional water resources, and establishing a unified wastewater discharge management system.

As a crucial energy base and ecological barrier in North China, Shanxi Province plays a pivotal role in national energy security and regional development. However, its resource profile characterized by "abundant coal but scarce water" and its water-intensive industrial structure have triggered severe water security challenges that constrain sustainable development. Given that systematic quantitative research on its water security remains scarce. Comprehensive assessment, identification of influencing factors, and trend prediction are essential prerequisites for addressing these water resource constraints. Therefore, this study constructs a water security evaluation index system for Shanxi Province based on the Driving Force-Pressure-State-Impact-Response (DPSIR) model. The Single indicator quantification-Multi-indicator integration-Poly-criterion integration (SMI-P) method was employed to quantify the dynamic characteristics of water security from 2011 to 2022. An obstacle degree model and a coupling coordination degree model were used to identify key constraints and synergistic mechanisms among subsystems. Finally, the ARIMA model was applied to predict water security trends from 2023 to 2027. The results indicate that Shanxi’s water security level exhibited a fluctuating upward trend with relatively mild volatility from 2011 to 2022, remaining "relatively unsafe" overall. Only in 2021 did it reach a "basically safe" status, attributed to a favorable combination of water resource endowment, economic growth, and population pressure. The major obstacle factors affecting water security included the comprehensive utilization rate of industrial solid waste, GDP growth rate, per capita water resources, and water yield modulus, forming a dual constraint of "supply-side limitations" and "demand-side pressures." The coupling coordination degree of water security showed high relevance but fragile equilibrium, progressing from "barely coordinated" to "primarily coordinated," peaking at "moderately coordinated" in 2021 before declining in 2022. The water security level from 2023 to 2027 is projected to first rise and then decline, remaining at a "basically safe" level from 2023 to 2026 before reverting to a "relatively unsafe" level in 2027.

Flooding stress is a pivotal environmental factor in ecosystems such as wetlands and paddy fields, significantly influencing soil physicochemical properties and microbial metabolic processes, and thereby regulating the production and emission of greenhouse gases (i.e., carbon dioxide and methane). Flooding induces a decline in soil redox potential (Eh), shifting microbial metabolism from aerobic respiration to anaerobic fermentation. It also alters key soil parameters, including pH and the decomposition rate of organic matter. These changes in the soil habitat further affect the production and emission of carbon dioxide and methane. This review describes the variations in soil habitats under flooding conditions, analyzes the impacts of flooding stress on carbon dioxide and methane fluxes, and focuses on summarizing the interaction mechanisms between soil habitat alterations and their production/emission of these two greenhouse gases under flooding stress. In the future, strategies such as developing and improving regulatory measures for soil habitats and carbon dioxide/methane emissions, analyzing multi-scale interaction mechanisms, and the exploration of interdisciplinary ecological models will improve the comprehensive understanding of carbon cycling processes within soil ecosystems. This understanding will provide a critical scientific basis for ecosystem conservation, addressing climate change adaptation and mitigation, and the sustainable development of agriculture, with the aim of promoting further research and practical applications in related fields.

To enhance the interpretability of machine learning (ML) models for predicting rice water requirements and to fully exploit data potential, this study proposes a prediction model based on feature engineering. The model constructs interaction features and incorporates decomposed factors from a modified Penman-Monteith (PM) formula as inputs for the ML model to predict rice water requirements. Features were selected as inputs using the Maximum Information Coefficient (MIC > 0.3), and the SHAP (SHapley Additive exPlanations) method was applied to analyze the interpretability of the input features based on the prediction results. The results demonstrate that after applying feature engineering, the coefficient of determination (R2) for the three ML models increased by 4.5% to 7.8%, the Root Mean Square Error (RMSE) decreased by 43% to 52%, and the Mean Absolute Error (MAE) decreased by 45% to 53%. In cross-model comparisons, the constructed features exhibited higher average SHAP values than the original features when used as inputs. The ML models using engineered features as input not only reduced prediction errors and improved prediction accuracy but also provided a basis for enhancing the interpretability of the ML model.

In order to clarify the multi-temporal scale characteristics and driving mechanisms of winter wheat evapotranspiration (ET) in the North China Plain, based on the eddy covariance system and meteorological observation data from 2017 to 2018, this study utilized data from an eddy covariance system and meteorological observations from 2017 to 2018 to systematically reveal the dynamic patterns and evolution of driving mechanisms of ET on multiple time scales, highlighting the scale-dependence effect. On an hourly scale, ET showed a significant single-peak diurnal variation, with the peak occurring between 13∶00 and 14∶00 and approaching zero at night. An inverted U-shaped pattern was maintained at different growth stages, with the daily average ET being the highest at the heading-jointing stage, significantly higher than that at the regreening-jointing stage. On a daily scale, ET gradually increased with the growth process, from 0.50 mm/d in the emergence-regreening stage to 5.10 mm/d in the milking-maturity stage, showing obvious growth stage dependence. On a monthly scale, cumulative ET showed significant seasonal differences, peaking in May (170.0 mm in 2017 and 135.0 mm in 2018) and reaching a minimum in January (7.1 mm in 2017 and 2.9 mm in 2018), reflecting the comprehensive regulation of climate and phenology during the year. The dominant environmental factors for ET changed dynamically with scale and growth stage. The daily scale was dominated by vapor pressure deficit (VPD) and net radiation (Rn) (r = 0.88), while the monthly scale primarily reflected the cumulative effect of Rn (r = 0.72). The driving mechanisms also evolved in stages during the growth period: from the seedling emergence to regreening stage, ET was synergistically driven by Rn, air temperature (Ta), and VPD; Rn and VPD were dominant during the jointing-heading stage; the contribution of VPD was greatest during the heading-grain filling stage; and after grain filling, the process returned to an Rn-dominant mode. Notably, Rn maintained a highly significant positive correlation throughout the whole growth period. This study reveals the scale dependence of the evapotranspiration process and the dynamic transformation patterns of its driving mechanisms, providing solid theoretical support for the development of accurate farmland water consumption estimation models and intelligent water-saving irrigation systems.

In the context of paddy field expansion, there is an urgent need for a high temporal and spatial resolution ET estimation method with high spatiotemporal resolution, based on remote sensing and energy balance models, to enable rapid and accurate monitoring of regional ET dynamics. This study utilized the SEBAL model, combining Landsat imagery and meteorological data from 2000 to 2020, to estimate actual ET during the paddy field growing season (May-August). The study first verified the applicability of the SEBAL model in the Jiansanjiang Reclamation Area and then analyzed the spatiotemporal impacts of paddy field expansion on actual ET and their driving mechanisms. The results showed that, temporally, the paddy field area increased from 1 866.62 km2 in 2000 to 6 362.97 km2 in 2020, and the dynamic degree of area change reached a maximum of 21.42% in 2010. Spatially, paddy fields were initially scattered but later covered nearly the entire area. Validation of the SEBAL model’s accuracy showed correlation coefficients of 0.76 (June 2004) and 0.72 (July 2020) with an all-weather ET dataset, confirming its applicability. Comparing the pre- and post-expansion periods, the average monthly actual ET increased by 47.56 mm (May), 45.88 mm (June), 51.41 mm (July), and 57.16 mm (August), respectively. During the growing season, a spatial shift in ET clustering occurred: high-high ET clusters moved northward and westward and increased in number, while low-low ET clusters contracted from the periphery toward the center and decreased substantially in number. Geodetector results indicated that the Normalized Difference Vegetation Index (NDVI), Land Surface Temperature (LST), and surface albedo were the primary drivers of spatial heterogeneity in actual ET, with the interaction effect between NDVI and albedo reaching 0.85. This study confirmed the feasibility of using the SEBAL model for rapid and accurate ET estimation in the Jiansanjiang Reclamation Area, providing methodological support and a data reference for the optimization of agricultural water resources in cold regions.

In order to develop a drought index better suited to the climatic and topographic conditions of Yunnan Province, this study introduces LAI and LST-T parameters into traditional Temperature-Vegetation-Precipitation Drought Index (TVPDI), thereby developing an improved drought index, ITVPDI. Using remote sensing and meteorological data from 2001 to 2020, we analyzed the spatiotemporal evolution and frequency of drought in the region. The results show that the correlation coefficients of ITVPDI with Solar-Induced Chlorophyll Fluorescence (SIF) and Moisture Index (MI) were 0.52 and 0.48, respectively, significantly outperforming the traditional TVPDI. Droughts in Yunnan were most severe in spring and winter, with multi-year average ITVPDI values of 0.409 2 and 0.336 4, respectively, and with drought affecting 85.18% and 93.02% of the province. Summer drought was the mildest, with an average ITVPDI of 0.453 5, where light drought accounted for 80.42% of the area. On an interannual scale, Yunnan Province was in a state of moderate drought (ITVPDI = 0.420 1), with drought conditions affecting more than 91% of the region. Spatially, the drought frequency showed an increasing trend from southwest to northeast, with Xishuangbanna having the highest frequency of mild drought (48.74%) and Diqing having the highest frequency of extreme drought (40.95%). Hurst index analysis indicates that future drought is likely to intensify in spring, autumn, and winter, while showing a tendency to moderate in summer and on an interannual scale. The ITVPDI is suitable for drought monitoring in the complex terrain of Yunnan and provides a data-driven and theoretical basis for regional drought zoning management and policy formulation.