Unraveling wheat yields dynamics under variable sowing dates and irrigation schedules: A multivariate and agronomic trait perspective

Subsurface drip irrigation (SDI) is an efficient irrigation method widely used in arid and semi-arid regions; however, its effectiveness largely depends on soil physical properties and the resulting soil water distribution. This study investigated soil wetting pattern dynamics under SDI and developed empirical models to predict wetted soil geometry as influenced by emitter discharge, installation depth, irrigation time, soil texture, and bulk density. Laboratory experiments were conducted using two representative soils from the Kurdistan Region of Iraq (Semel—silty clay and Zakho—clay loam). Emitters with discharge rates of 2 and 4 L h⁻¹ were installed at depths of 12.5, 25, and 37.5 cm, and equal water volumes (8 L) were applied to monitor the advancement of the wetting front. Nonlinear regression models were developed to estimate maximum horizontal wetted diameter (H) and average vertical wetted depth (V) as functions of emitter discharge (q), irrigation time (t), clay content (c), and bulk density (ρb). The models showed strong predictive performance, with coefficients of determination (R²) ranging from 0.87 to 0.98. Model validation indicated acceptable accuracy, with root mean square error (RMSE) values of 1.676–5.033 cm, mean absolute error (MAE) of 4.531–10.745 cm, and mean absolute percentage error (MAPE) of 5.468–10.613%. The index of agreement (d) ranged from 0.968 to 0.995, while mean bias error (MBE) values were close to zero, indicating minimal systematic error. The results demonstrated that wetting front expansion was primarily governed by irrigation time, whereas emitter discharge had a comparatively smaller influence, particularly at shallow depths. Increased clay content reduced both horizontal and vertical wetting dimensions, while the effect of bulk density varied with emitter depth. Lower discharge applied over longer durations enhanced lateral water movement, whereas higher discharge promoted vertical flow and deep percolation. The developed models provide practical tools for predicting soil water distribution and can support the optimization of emitter spacing, irrigation scheduling, and water-use efficiency in SDI systems under semi-arid conditions.

Relevance . When growing sweet peppers, the quality of the product is determined by strict adher­ence to agricultural regulations. Sweet peppers, which are highly sensitive to soil moisture, place particular demands on irrigation regimens, necessitating the development and strict adherence not only to irrigation parameters, a key factor in the successful cultivation of this crop, but also to the application of mineral fertilizers. Materials and methods . The article investigates impact of various irrigation schedules (70%, 70-80%, 80% MWCminimum water capacity) and mineral fertilizers norms (N60P120K60, N90P180K90, N120P240K120) on pepper growth, development, water consumption and yield under climatic conditions of Central Non-Black Soil Region. Results . Investigation results showed when before irrigation soil moisture content increases from 70% to 80% MWC, the total plants water consumption grows in 15-20%. At the same time, mineral fer­tilizers application didn’t significantly change this index value. Maximal yield (161 c/ha) was achieved at irrigation schedule of 80% MWC and fertilizer norms of N90P180K90, that is in 59-60 c/ha higher than the control one. At the same time, water consumption index went down up to 213,7 m 3 //t, that shows higher water consumption efficiency. It is interesting that further increase of fertilizer norms up to N120P240K120 led to insignificant yield decrease that shows the optimal level of mineral consumption. Phenological studies showed that phases of plants development under all variants went on synchro­nously but biometric parameters (plants height, number of leaves and fruits) had maximal value at combination of high irrigation schedule (80% MWC) and mean fertilizer norms. Under these condi­tions stable marketability of fruits remained high (95,4-98,1%) independently of the testing variants. The obtained data have important practical meaning for growers working in the central part of Non­Black Soils zone of Russia; these data enable optimization of pepper water consumption and miner­al fertilizing. Investigation revealed that optimal results in pepper growing can be obtained at irriga­tion schedule of 70-80% MWC is practiced with application of mineral fertilizers (nitrogen N90, phosphorus P180, potassium K90). Such combination enables to obtain high yields at simultaneous resources consumption decrease.

This study presents and evaluates a real-time decision support system (DSS) for site-specific irrigation scheduling based on soil moisture forecasting with SWIM 2 (Sensor Wielded Inverse Modeling of a Soil Water Irrigation Model). The SWIM 2 framework integrates a soil water balance model with in situ sensor data and soil moisture samples through Bayesian inverse modeling to generate probabilistic 10-day soil moisture forecasts. We assess the performance of the soil moisture forecasts and the irrigation DSS by integrating the model parameter ensemble with either deterministic or ensemble-based probabilistic weather forecasts, providing insights into their benefits and trade-offs in real-time irrigation management. Both approaches resulted in high detection rate and accuracy in predicting water stress triggering the irrigation threshold. The full ensemble yielded slightly better reliability at longer lead times whereas the probability distribution of the soil moisture predictions at short lead times was dominated by the SWIM 2 parameter uncertainty. Simulation of different irrigation treatments using the calibrated SWIM 2 -based model illustrated and confirmed its potential for evaluating water use efficiency and crop response. Overall, this work illustrates the application and practical advantages of a probabilistic, ensemble-based modeling framework in supporting site-specific, data-informed irrigation strategies. • SWIM 2 enables real-time, site-specific irrigation advice using probabilistic soil moisture forecasts. • Soil moisture uncertainty shifts from model parameter to weather uncertainty dominance with longer lead times. • Deterministic and ensemble weather forecasts were compared in water stress predictions for irrigation support. • Simulated irrigation strategies reveal impacts on water use efficiency, irrigation efficiency, and productivity.

An IoT-based predictive irrigation scheduling framework for precision soil moisture control in greenhouses

At the experimental research farm of the Department of Agronomy, Experimental Research Farm in Palampur, Himachal Pradesh, during the rabi seasons of 2021–2022 and 2022–2023, a field experiment titled “Agro meteorological indices influenced by different sowing windows and irrigation scheduling using weather model and spatial data of wheat in Western Himalayas” was carried out. The experiment comprised a split plot design with 3 replications and included 5 irrigation treatments: rainfed conditions (I1), 2 irrigations (I2), 3 irrigations (I3), irrigation scheduling based on Penman Monteith modified (I4) and irrigation scheduling based on spatial reference ET of grid (I5). The crop sown on 25th October had the highest values of growing degree days (GDD), helio-thermal units (HTU), heat use efficiency (HUE), photo-thermal units (PTU) and photo-thermal index (PTI) compared to the crop sown on 10th December in 2021–2022 and 2022–2023. This helped to increase the grain yield by 47.4 and 35.8 %, respectively, over the crop sown on 10th December. Due to the highest values of GDD, HTU, HUE, PTU and PTI observed under the highest irrigation treatments, the largest increase in grain yield was recorded with 3 irrigations (33.1 and 19.4 %) and Penman Monteith modified (33.7 and 13.5 %) over rainfed conditions. The findings recommend timely sowing around late October and the use of weather-based irrigation scheduling-especially the Penman-Monteith method as sustainable strategies to improve wheat productivity and climate resilience in the Western Himalayan region. These practices optimise thermal indices and water use efficiency, supporting adaptive agriculture under variable climatic conditions.

HighlightsWhat are the main findings?An IoT-based precision irrigation system integrating soil moisture and water-level sensors with scheduled sensor-based automation effectively controlled AWD irrigation without manual intervention.The IRRI35 treatment maintained high grain yield (7.76 t ha−1) while reducing water use by 28%, energy consumption by 37%, and significantly improving water use efficiency compared to continuous flooding.What are the implications of the main findings?Automated irrigation can overcome key adoption barriers of AWD by minimizing labor requirements and improving water governance.Large-scale implementation could reduce irrigation energy demand, and enhance sustainability in rice production systems.Despite its significant water-saving potential, the adoption of alternate wetting and drying (AWD) irrigation remains limited due to infrastructure constraints and intensive manual monitoring requirements. An automated precision irrigation system was developed and tested at the Bangladesh Rice Research Institute research farm in Gazipur, Bangladesh. The system combined ultrasonic water-level sensors, capacitive soil moisture sensors, an Arduino-based microcontroller, a GSM communication module, and solar-powered automatic control gates. Field performance was evaluated following a Randomized Complete Block Design (RCBD) under four irrigation treatments: IRRISAT, IRRI35, IRRI25, and continuous flooding (CF). The first three irrigation treatments were operated using scheduled daily decision windows, in which irrigation actions were automatically triggered based on predefined schedules and sensor threshold values. In IRRISAT, irrigation started when soil moisture dropped slightly below saturation and stopped at a ponding depth of 5 cm, while IRRI35 and IRRI25 were triggered at volumetric soil water contents of 35% and 25%, respectively, with the same upper cutoff of 5 cm ponding depth; CF served as the control. The IRRI35 treatment achieved a high grain yield (7.76 t ha−1) while reducing water use by 28% and energy consumption by 37% compared to CF. Water use efficiency was considerably higher under IRRI35 (9.4 kg ha−1 mm−1) than under CF (6.7 kg ha−1 mm−1). The automated system proved to be reliable and precise in scheduled irrigation control, significantly reducing water use and labor requirements. The findings suggest that large-scale adoption of the system under real-world cultivation conditions could reduce irrigation energy needs and contribute to sustainable water governance in rice production.

Water scarcity is the primary constraint on the development of the potato industry in Northwest China. Improving water use efficiency (WUE) under limited water supply is, therefore, an urgent priority to promote the green and sustainable development of potato production in this region. This research was conducted from 2023 to 2024 in the rain shelter of the Agricultural Science Research Institute in Dingxi City, Gansu Province, using the potato cultivar ‘Gan Yin No. 9’ as the experimental material. Throughout the growing season, the control treatment (CK) was maintained at 75–85% of the field water capacity (FWC). Based on CK, three deficit-irrigation treatments were established: W75 (75% of the CK irrigation amount), W50 (50% of CK irrigation amount), and W25 (25% of CK irrigation amount), with three replicates per treatment. We evaluated the effects of different irrigation regimes on plant growth characteristics, physiological characteristics, tuber yield, and WUE. The results showed that the W75 treatment significantly (p < 0.05) promoted the growth of plant height and stem diameter, and significantly increased them by 8.70–10.20% and 13.03–18.70%, respectively, compared with CK. The total dry matter accumulation under W75 was significantly higher than CK (by 10.90–11.40%) and markedly higher than W50 and W25 (by 24.10–45.50%). No significant differences were observed in tuber yield, large tuber rate, and medium tuber rate between W75 and CK. Notably, W75 significantly improved WUE by 36.43–38.51% compared with CK. Overall, under the conditions of this study, W75 treatment was identified to be the optimal irrigation regime for potato cultivation, as it promoted plant growth, maintained tuber yield, and enhanced water use efficiency. This study aims to establish a definitive irrigation threshold for potato production in Northwest China. The findings provide a precise basis for formulating irrigation schedules, which can contribute to the development of water-efficient agriculture and support the sustainable development of the potato industry in the region.

To address peak edge operation and excessive valve switching in hydraulically coupled multi-zone campus irrigation, this study proposes a collaborative scheduling framework that combines short-term evapotranspiration (ET) forecasting with safety-constrained reinforcement learning. Temperature, relative humidity, and light intensity are used to construct vapor pressure deficit and radiation proxy features, and a lightweight predictor provides two-hour-ahead ET statistics as forward-looking disturbance information. A safety layer composed of Top-2 gating and total flow projection is then used to map policy outputs into a feasible action space under parallel irrigation and total flow constraints. Using seven consecutive days of field data from October 2025, the proposed method reduced total water consumption to 131.04 m3, corresponding to reductions of 9.13% and 6.12% relative to fixed-schedule and hysteresis threshold rotational irrigation, respectively. It also reduced the maximum total flow from 2.00 to 1.60 L/s, lowered valve switching cycles to 12, and reduced the border ratios at 0.90 and 0.95 to 0. Additional ablation, sensing noise/packet loss, and Top-K extension experiments further showed that ET forecasting improves anticipatory scheduling, whereas safety projection is essential for zero-violation operation. These results demonstrate that the proposed framework provides a practical and deployable solution for safe and water-efficient multi-zone irrigation scheduling under shared pump constraints.

ABSTRACT Quantifying crop evapotranspiration (ET c ) dynamics and their partitioning into soil evaporation (E) and plant transpiration (T) is crucial for improving water productivity and developing precise irrigation strategies. However, the spatiotemporal distribution of soil moisture and its influence on ET c partitioning in drip‐irrigated maize fields are poorly understood. This study investigated soil water dynamics, leaf area index (LAI), ET c and its components under varying irrigation lower limits (W1: 50%–60% FC, W2: 65%–75% FC, W3: 80%–90% FC, where FC was the soil field capacity) during different maize growth stages in 2023 and 2024. The results indicated that as the irrigation amount increased, the vertical advance depth of the soil wetting front increased more significantly than the surface wetting radius Maize growth was suppressed at the tasselling stage under W1. T/ET c showed a quadratic relationship with ET c ( R 2 = 0.50) when reference evapotranspiration (ET 0 ) exceeded 1.9 mm d −1 , with an inflexion point at ET c = 6.1 mm d −1 . T/ET c correlated strongly with LAI through a logarithmic function ( R 2 = 0.95), especially at early growth stages. Soil water content (SWC) demand peaked at the tasselling stage. An irrigation strategy that maintains higher soil moisture (e.g., W3: 80%–90% FC) during the water‐sensitive tasselling and seedling stages, while applying mild deficit irrigation (e.g., W2: 65%–75% FC) during the jointing and filling‐to‐maturity stages, can optimize the T/ET c ratio and achieve synergistic improvements in grain yield (GY) and water productivity (WP). This study enhanced the understanding of soil water's role in ET c partitioning in drip‐irrigated maize, providing a quantitative basis for optimizing irrigation scheduling to enhance both GY and WP.

ABSTRACT Multi‑scenario flowchart showing CMIP6 climate models → bias correction → CROPWAT model → wheat irrigation projections for Bangladesh under SSP1‑2.6 to SSP5‑8.5, with temperature as dominant driver. Climate change poses a significant threat to wheat production in Bangladesh, a staple food. This study projects future irrigation requirements for wheat in Mymensingh by integrating an ensemble of five CMIP6 climate models with the FAO CROPWAT model, employing rigorous quantile-mapping bias correction and out-of-sample split-sample validation. Four emission scenarios (SSP1-2.6, SSP2-4.5, SSP3-7.0, SSP5-8.5) were analyzed across near (2026–2050), mid (2051–2075), and far (2076–2100) futures. Projections indicate progressive warming, with January maximum temperatures increasing by up to 21.71% under SSP5-8.5 by late century. Consequently, potential crop water requirements rise from 1.38–3.87% near-term to 15.06% under SSP5-8.5 in the far future. While some precipitation increases were noted, their benefits are offset by rising evapotranspiration. Irrigation scheduling will be compressed, with the third irrigation advancing by one week under higher emission scenarios. Sensitivity analysis identified temperature as the dominant driver, increasing irrigation demand by up to 17.19% when isolated; this represents an upper-bound estimate as CROPWAT excludes CO2-mediated stomatal effects. A 20% reduction in soil moisture significantly altered irrigation efficiency. Substantial inter-model variability underscores projection uncertainty, reinforcing the necessity for adaptive, climate-resilient water management strategies to safeguard wheat productivity and national food security in Bangladesh's evolving agricultural landscape.

Abstract India's agricultural economy is burdened by a fragmented supply chain in which smallholder farmers—constituting more than 86% of total farm holdings—are consistently denied fair market prices due to multi-tiered intermediary networks that collectively absorb 30–50% of the final consumer price. This paper presents KisanVyapar, a full-stack, web-based farm-to-market digital platform designed to eliminate intermediary dependency. The system provides a role-differentiated dual-user environment for farmers and merchants, incorporating a dynamic product marketplace with category filtering and image upload, a lifecycle-managed order processing system, a real-time peer-to-peer messaging module, and KisanMitra—an AI-powered agricultural advisory chatbot powered by the Claude AI API (Anthropic). KisanMitra detects and responds in English, Hindi (हिन्दी), and Marathi (मराठी) automatically, providing expert guidance on crop diseases, fertilizer selection, mandi price estimation, irrigation scheduling, and government scheme navigation. The entire platform UI is fully internationalized across the same three languages via a server-side PHP translation layer. Implemented on a LAMP stack using PHP 8.x and MySQL 8.0, the platform is deployable on standard shared hosting infrastructure. Evaluation through structured functional testing and expert chatbot assessment confirmed: average module response times of 1.2–2.1 seconds, chatbot domain accuracy of 91.2% (English), 88.8% (Hindi), and 87.6% (Marathi), and zero critical failures across all 42 user-role workflow test cases. Keywords — agricultural e-commerce, farm-to-market platform, AI chatbot, multilingual NLP, rural digitization, PHP web application, smallholder farmers, KisanMitra, Claude AI API, supply chain disintermediation

A field experiment was conducted over two consecutive rabi seasons (2017-18 and2018-19) at three locations in Punjab (Ludhiana, Faridkot, and Ballowal Saukhri). The experiment followed a Factorial Split Plot Design with three dates of sowing (D1 – 25th October, D2 – 15th November, D3 – 5th December) and three wheat cultivars (V1- WH1105, V2 - UNNAT PBW 550, V3 - PBW 590) in the main plots, and two irrigationtreatments (I1: Recommended, I2: Recommended ± weather forecast based) in the subplots. Across both seasons, the crop sown on 15th November consistently exhibited the highest tiller count and dry matter at harvest. Grain yield was significantly higher for D2 compared to D3, and was comparable to D1 across all three locations. Among the cultivars, UNNAT PBW550 yielded significantly more than PBW 590. Among the locations Ballowal Saukhri had lowest grain yield. Analysis indicated non-significant differences in yield under different irrigation schedules across the three sowing dates

Data Driven Irrigation Scheduling for Water Saving

Effect of Irrigation Scheduling on Forage Yield and Quality of Rhodes Grass and Common Vetch under Water Stress Conditions

This study evaluates the separate and combined effects of sodium polyacrylate (hydrogel) and irrigation schedules on the growth and flowering of Petunia plants. Water stress is a major abiotic factor that negatively affects flower morphology and the production of high-quality blooms by disrupting plant growth, physiological processes, and metabolic functions. An experimental study was conducted using different sodium polyacrylate treatments (SP0: pure soil; SP1: 90% soil, 10% sodium polyacrylate; SP2: 70% soil, 30% sodium polyacrylate; SP3: 50% soil, 50% sodium polyacrylate) combined with varying irrigation intervals (W0: twice daily; W1: daily; W2: every other day; W3: every third day) to assess their effects on vegetative and reproductive attributes of Petunia. The results indicated that SP2 (70% soil + 30% sodium polyacrylate) significantly reduced the time to flowering and increased both the quantity and size of flowers, while SP1 (90% soil + 10% sodium polyacrylate) enhanced vegetative growth, including the number of leaves, branches, plant height, root length, and root volume. Irrigation every three days promoted earlier flowering and improved floral traits, whereas irrigation at two-day intervals favored vegetative growth. Interactive effects of sodium polyacrylate and irrigation schedules significantly influenced both morphological and floral characteristics. Based on these findings, SP2 with a three-day irrigation interval is recommended for reproductive performance, and SP1 with a two-day interval is recommended for vegetative growth of Petunia.

As water resources are becoming increasingly scarce in the North China Plain, irrigation strategies that simultaneously improve grain yield and reduce irrigation water input are needed for winter wheat (Triticum aestivum L.) production. Current irrigation decision rules are based either on fixed soil moisture thresholds or on evapotranspiration (ET)-based ratios applied uniformly across the growing season, limiting their flexibility for growth stage-specific irrigation management. In this study, a multi-objective simulation optimization framework was developed to jointly optimize soil moisture lower control limits (irrigation trigger thresholds) and evapotranspiration-based irrigation replenishment ratios across key winter wheat growth stages. The framework integrated the AquaCrop-OSPy crop model with the PyFAO56 soil moisture balance, irrigation scheduling model and the NSGA-II evolutionary optimization algorithm. A field experiment was conducted during the 2024–2025 growing season in Laoling City, Shandong Province, China, employing a four-dense–one-sparse strip cropping pattern with two irrigation treatments: T1 (subsurface sprinkler irrigation) and T2 (shallow subsurface drip irrigation). The AquaCrop-OSPy model was calibrated and validated using measured canopy cover, aboveground biomass, grain yield, and soil moisture content in the 0–60 cm soil layer. Simulated canopy cover and grain yield showed good agreement with observations, with the coefficient of determination (R2) ranging from 0.87 to 0.94. For grain yield, the normalized root mean square error (NRMSE) ranged from 2.24% to 3.75%, and the root mean square error (RMSE) ranged from 0.29 to 0.54 t·ha−1. For aboveground biomass, R2 was 0.99, while RMSE ranged from 1.02 to 1.11 t·ha−1, and NRMSE ranged from 14.25% to 15.49%. The PyFAO56 irrigation strategy model simulated average root-zone soil-moisture dynamics with satisfactory accuracy, with an R2 of 0.86 and an RMSE of 5%. Multi-objective optimization (maximizing yield while minimizing irrigation volume) generated 23 Pareto-optimal irrigation strategies, with irrigation volumes ranging from 51 to 128 mm, corresponding yields ranging from 9.8 to 10.8 t·ha−1, and irrigation water use efficiency (IWUE) ranging from 0.08 to 0.19 t·ha−1·mm−1. Correlation analysis within the Pareto set indicated that soil-moisture control lower limits during the regreening–jointing stage and higher soil-moisture control lower limits during the flowering–maturity stage were key controlling factors for achieving high yields and irrigation water use efficiency. The Entropy-Weighted Ranked Minimum Distance method identified an optimal irrigation scheme involving two irrigations (one at the end of the jointing stage and another at the beginning of the grain filling stage) involving an irrigation depth of 75 mm, achieving a simulated yield of 10.4 t·ha−1 and an IWUE of 0.16 t·ha−1·mm−1. The proposed AquaCrop-PyFAO56-NSGA-II framework provides a flexible, process-based workflow for jointly optimizing irrigation control thresholds and evapotranspiration-based irrigation replenishment ratios across different winter wheat growth stages. Under the monitored conditions of the 2024–2025 wet season, the framework identified a two-irrigation strategy that balanced grain yield and irrigation input. This study should, therefore, be regarded as a proof-of-concept evaluation conducted in a well-instrumented single-site field setting rather than as a universally transferable recommendation. Because model calibration, within-season validation, and optimization were all based on one wet growing season at one site, the derived stage-specific thresholds, Pareto front, and S5 recommendation are most applicable to hydro-climatic conditions similar to the study year and should be further tested across contrasting year-types and locations before broader extrapolation.

ABSTRACT Accurate determination of transient water flow density ( J w ) within the vadose zone is essential for elucidating hydrological processes in the Earth's Critical Zone (CZ), with key applications in irrigation scheduling, soil erosion mitigation, and sustainable land management. This study evaluates the thermo‐time domain reflectometry (thermo‐TDR) method for real‐time J w monitoring in unsaturated soils, addressing limitations of traditional heat pulse techniques and advancing instrumental development for Critical Zone Observatories (CZOs). A theoretical error analysis was conducted to quantify the sensitivity of J w estimates to inaccuracies in the input parameters of the heat‐pulse method. The results revealed high sensitivity at low flow rates (< 10 mm h −1 ), where probe‐spacing errors alone can produce relative errors up to 160%. An in situ probe spacing calibration method was developed to enhance accuracy without requiring complex sensor designs. A total of 33 soil column experiments across five sub‐tropical soils (one sandy soil, two silt loams and two silty clay loams) under varying unsaturated conditions were conducted to evaluate the accuracy of the improved thermo‐TDR method. The results demonstrated that thermo‐TDR measurements closely matched independent J w values in sandy soils but showed greater variability in finer‐textured soils, primarily due to time lags in heat and water transport and flow heterogeneity. Statistical analysis indicated reasonable correlations ( R 2 ≈ 0.62) between thermo‐TDR and independent J w values, with RMSE values of 10.6 and 6.24 mm h −1 for the upper sensor‐derived infiltration rates and lower sensor‐derived drainage rates, respectively. Despite challenges at low J w , the thermo‐TDR method shows promise for in situ J w monitoring, offering potential for broader application in CZOs to quantify anthropogenic impacts on water resources and enhance interdisciplinary CZ research.

Global agricultural water scarcity necessitates intelligent irrigation management systems capable of optimizing water allocation while preserving crop productivity. This paper proposes a reinforcement learning (RL) framework for adaptive irrigation scheduling under seasonal water budget constraints, formulated as a constrained Markov decision process (MDP) and solved using a deep Q-network (DQN) agent. The DQN agent processes multi-layer soil water content, crop developmental stage, accumulated evapotranspiration (ET) deficit, and short-term precipitation forecasts as state inputs, selecting from five discrete daily irrigation volumes. A calibrated AquaCrop crop simulation model serves as the training environment, providing physically realistic state transitions across soil compartments. A dual-component reward function jointly penalizes water application cost and crop water stress. Evaluated across ten held-out growing seasons, the DQN agent reduced total seasonal irrigation by 18.4% relative to a conventional threshold-based scheduler, while maintaining mean grain yield at 97.2% of the benchmark. These results confirm that RL-driven adaptive policies can effectively navigate the yield-water tradeoff under resource-constrained agricultural conditions.

Optimizing deficit irrigation and fertilization at growth stages to improve citrus yield, fruit quality, and water-fertilizer productivity in semi-arid Southwest China