Coupled weather and crop simulation modeling for smart irrigation planning: a review Open Access

Natural calamities such as floods, droughts, rainfall variability, and other manmade disasters have become more frequent and intense globally as a result of climate change (Dai et al. 2018; Kim et al. 2023). It is anticipated that rising water temperatures and variations in extreme weather events will have an impact on water quality and worsen a variety of water pollution issues, which will have crucial effects on agriculture. Food availability, stability, access, and consumption are expected to be impacted by changes in water quantity and quality caused by climate change. This is predicted to result in less food security and greater susceptibility for vulnerable rural farmers, particularly in the mega-deltas of Asia and Africa as well as in dry and semi-arid tropical regions. Two fundamental elements are crucial, namely, the agricultural sector utilizes freshwater by far the most (Molden et al. 2007) and water usage in agriculture often yields lower net returns than other competitive uses (Scheierling et al. 2014). It is predicted that by the 2050s, the area of land experiencing water stress will be aggravated as a result of climate change. India has some of the world's most fragile and uncertain water supplies, and the nation suffers from severe water stress. Sathyamoorthy et al. (2019) studied the effects of surface as well as subsurface drip irrigation levels and allowable deficit irrigation (DI) on various crops such as rice (Sathyamoorthy et al. 2022), tomato, cotton-maize, sunflower (Kaviya et al. 2018), and red gram by climatological approaches which results in higher water-use efficiency (WUE), water productivity (WP), and optimal water use under water stress conditions (Sathyamoorthy et al. 2022). Another study calculated the water requirements of sorghum crop using daily rainfall, potential evapotranspiration (ET), and crop coefficient data for 30 years. The results found that the seasonal water requirement was 428.4 mm, which is 73% of the total potential ET of 110 days duration (Kokilavani et al. 2018). The irrigation sector dominates India's current and future water-use scenarios, accounting for almost 78% of them and therefore it is very essential to increase the WUE in agriculture to optimize the usage of available water. The famous slogan ‘More Crop per Drop’ (Molden 1997) was relabeled by the Indian prime minister under the scheme PMKSY (Pradhan Mantri Krishi Sinchayee Yojana) as ‘Per Drop More Crop’ to provide support to farmers for the adoption of modern technologies requiring a large amount of capital. These methods (micro-irrigation) will always result in water savings during on-farm irrigation, but they might occasionally increase crop yields. The scientific community and planners have always faced challenges in increasing the productivity of water use in agriculture to obtain maximum production or value from each unit of water used or applied. These challenges arise from a growing physical shortage of water as well as a scarcity of economically accessible water due to rising production and resource costs (Kijne et al. 2003; Kumar et al. 2009). Irrigation currently provides additional water to approximately 18% of the world's croplands (Reid et al. 2005). According to study of Chand et al. (2020), the most effective way to increase agricultural productivity is through expanding access to irrigation and technological advancement. Enhancing agricultural techniques, such as converting from gravity surface irrigation systems to pressurized drip or sprinkler systems, can help make irrigation scheduling much easier (Abioye et al. 2020), but we lack the ability to manage risks, such as using more or less water in terms of normal irrigation to increase farmers' knowledge, which will lead to reduced yield, increased use of water, and may risk sustainability as a cause of improper planning and management (Van Dam et al. 2006). To precisely estimate plant growth, assess the need for water, forecast yield, avoid disease, and comprehend the effects of climate change on agricultural output, crop simulation models have been created (Gouvêa et al. 2009). In the modern era, a computer simulation is an effective tool for examining how irrigation techniques affect the balance between soil and water (Lashari et al. 2010). Further models are increasingly being used to visualize processes associated with irrigation (Assouline 2002; Gärdenäs et al. 2005). Fully calibrated process-based models have become valuable research tools for predicting complicated and interactive water flow and solute transport processes in and below the root zone. They can quickly evaluate different irrigation management strategies for varying climate and soil conditions, as well as different crops, without requiring labor-intensive fieldwork (Chen et al. 2023). When combined with the right available data sources, irrigation system simulation models have a tremendous ability to advance irrigation research into contemporary technology. In regard to irrigation planning, crop simulation models are essential for ensuring maximum crop yield and water efficiency. According to study of Bouisse et al. (2011), models help in the collection and analysis of data on a variety of factors, including the need for irrigation, soil type suitability, and the selection of crop type, temperature, and moisture content, and some models allow farmers and policymakers to assess the potential impacts of different irrigation strategies and technologies on crop yields, WUE, and overall farm profitability. However, simulation models for irrigation planning have some potential weaknesses (Li et al. 2010). These weaknesses include their complexity, uncertainty about the physiological process, and the requirement of a large number of parameters to simulate, except for some models that use a limited number of explicit parameters (e.g., the Food and Agriculture Organization (FAO's) AquaCrop) and assess the yield response to water. Forecast-based irrigation planning, optimization tools, and remote sensing are crucial aspects of modern agriculture, especially in the face of climate change and increasing water scarcity. Many studies have used optimization techniques for irrigation planning, but they have difficulties such as attaining global optimal solutions for other algorithms. Recently, the combined use of optimization and simulation models has been preferred for exploring the unique advantages of this technique. The main objectives of this article are to review the papers and reports that provide reliable information as follows:

• 1. This article examines well-acclaimed crop simulation models such as AquaCrop, WOrld FOod Studies (WOFOST), Decision Support System for Agrotechnology Transfer (DSSAT), and Agricultural Production Systems sIMulator (APSIM) and their capabilities for effective irrigation planning.

• 2. The integration of advanced techniques such as forecast data, optimization tools, geographic information systems (GISs), and human‒machine learning tools into crop simulation models for real-time irrigation scheduling (RISs) highlights the importance of weather forecasting and coupled simulation-optimization tools.

This review revealed that while there is some research on coupling weather and crop simulation models for smart irrigation planning, there is a lack of studies that fully integrate emerging technologies such as remote sensing, machine learning or algorithms, and the Internet of Things (IoT) into models. While some studies may have developed advanced models for smarter irrigation planning, there may be a gap in translating these models into user-friendly decision support systems that can be easily adopted by farmers.

The Scopus database was used to download bibliographic data related to forecasting, simulation models, and irrigation research over 40 years from 1985 to 2023. To assess the productivity and influence of research outputs, institutions, and individuals, scientometric studies frequently employ the most recent information about scientific publications that are maintained by Scopus (Baas et al. 2020).

• I. The Scopus search keywords for irrigation and crop simulation model research were as follows:

TITLE-ABS-KEY ((irrigation OR evapotranspiration OR ‘Crop water’ OR ‘Soil moisture’ OR ‘Water productivity’) AND (‘Crop simulation’ OR ‘hydrological model’)) AND PUBYEAR >1985 AND PUBYEAR <2023 AND (LIMIT-TO (DOCTYPE, ‘ar’)) AND (LIMIT-TO (LANGUAGE, ‘English’)).

• II. Scopus search keywords for irrigation and crop simulation models for specific applications:

• • (a) TITLE-ABS-KEY ((irrigation scheduling) AND (‘Crop simulation model’ OR ‘hydrological model’ OR ‘optimization model’)) AND PUBYEAR >1985 AND PUBYEAR < 2023 AND (LIMIT-TO (DOCTYPE, ‘ar’)) AND (LIMIT-TO (LANGUAGE, ‘English’)).

  • (b) TITLE-ABS-KEY ((‘Weather forecasting’) AND irrigation AND (‘Crop simulation model’ OR ‘Optimization model’)) AND PUBYEAR >1985AND PUBYEAR <2023 AND (LIMIT-TO (DOCTYPE, ‘ar’)) AND (LIMIT-TO (LANGUAGE, ‘English’)).

  • (c) TITLE-ABS-KEY ((‘remote sensing’ OR ‘machine learning’ OR ‘GIS’ OR ‘Algorithms’) AND irrigation AND (‘Crop simulation model’ OR ‘Hydrological model’)) AND PUBYEAR >1985 AND PUBYEAR <2023 AND (LIMIT-TO (DOCTYPE, ‘ar’)) AND (LIMIT-TO (LANGUAGE, ‘English’)).

Based on the assumptions made by the scheduling, four types of irrigation scheduling approaches can be identified: (1) ET and soil water balance (WB); (2) soil moisture (Δ) status; (3) plant water status; and (4) simulation model output. Among these different approaches, the ET-WB method, a widely used technique, estimates crop evapotranspiration (ET_c) based on FAO Irrigation and Drainage Paper 56 (Allen et al. 1998). Despite its benefits, ET-based irrigation sometimes yields lower crop productivity than traditional experience-based planning (Hunsaker et al. 2015). Different irrigation software and control systems based on ET, such as the ET-based irrigation scheduling tool by Danny H. Rogers (2012) and SIMDualKc Software (Rosa et al. 2012), smart irrigation apps for implementing ET-WB-based irrigation scheduling (Migliaccio et al. 2013) have been used. However, all these techniques have several limitations, such as

• The ability to integrate models with real-time WB models,

• The ability to apply only to specific crops near weather stations and the ability to integrate them with GIS platforms.

These challenges can be effectively addressed by crop simulation models, and further integration of these models with different cutting-edge technologies such as remote sensing, irrigation platforms, and hydrological models has proven to be successful. Model-based irrigation scheduling is limited by determining the timing and quantity of water for the target field rather than for several fields (Gu et al. 2020). Nevertheless, there is agreement that the model is a useful tool for irrigation management when properly calibrated and validated. Among the different types of models, process-based and regression-based models such as the agro-hydrological soil water atmosphere and plant (SWAP), AquaCrop, CropWat, EPIC (Environmental Policy Integrated Climate), WOFOST, APSIM, and DSSAT models are reviewed because they commonly simulate irrigation scheduling, and certain crop simulation models have the ability to generate input data that may be applied to other models by themselves and come with built-in modules that include generic soils or soil profiles that are conducive to specific crops. Models such as APSIM, EPIC, AquaCrop, and DSSAT have been utilized in numerous studies to assist farmers globally in making management decisions on the timing and frequency of irrigation, plant population density, and sowing time under a variety of circumstances (Kephe et al. 2021). The majority of simulation models have built-in modules. Some of these models facilitate the assessment of DI strategies that otherwise require great effort and expense. Furthermore, simulation modeling enables farmers to make proactive decisions regarding irrigation, ensuring that water resources are utilized optimally and that waste is limited (Singels & Smith 2006).

The DSSAT model can be a valuable tool for irrigation scheduling by simulating crop growth and water use under different scenarios. In the fields of the La Violada Irrigation District, Spain's farmers' irrigation practices lack the management skills necessary to properly manage their irrigation systems compared with the DSSAT calibration and evaluation, and the model demonstrated good performance for simulating maize, wheat, barley, and sunflower crops in intensive cropping systems under Mediterranean conditions. Optimal irrigation management significantly increased the amount of irrigation water utilized by adjusting the amount of irrigation water applied based on the actual ET needs and the soil retaining capacity rather than the farmers' current irrigation calendar, according to the results of the DSSAT simulation of various water management scenarios (Malik & Dechmi 2019). The DSSAT-CERES (The Crop Environment Resource Synthesis)-Wheat model (V-4.6) was calibrated and validated against field experiment data (2008–2014) collected from phenological data in Varanasi, and the DSSAT-CERES-Wheat model (V 4.6) was calibrated and validated for wheat cultivars in Varanasi, suggesting that when irrigation water was in its optimum state, the model performed well; however, when irrigation water was stressed, the model performed poorly (Patel et al. 2017). As a result, the model needs to be improved for wheat crops to function well under stress. Compared with full irrigation, the combined DSSAT-CERES-Wheat model accurately simulated phenology, yield, and ET, which led to an increase in yield and WUE under DI during the booting stage. This model was used to predict winter wheat WUE under various irrigation regimes in the Texas High Plains (Attia et al. 2016). In a long-term DSSAT simulation (1991–2020) of DI-based ET in multiple locations (10) in the middle delta of Egypt, Tanta, who simulated irrigation based on 90% ET_c, increased grain yield and WP by 1.7 and 63%, respectively, as a result of varying automatic irrigation with percentages of 50, 60, 70, 90, and 100%, respectively, from ET_c; this was one of the simulation treatments compared with farmers' practices at all sites. Another study using the DSSAT-maize model revealed that early-to-mid-April planting dates and irrigation during the jointing and tassel phases are optimal. The optimal irrigation amount ranges from 1,000 m³/ha in wet years to 4,800 m³/ha in dry years (Kheir et al. 2021). Nearly half of the irrigation water could be saved under the simulated irrigation schedules, contributing to sustainable water resource management in the region (Jiang et al. 2016). The coupled framework of the PRECIS Regional Climate Model and DSSAT model was used to assess the impact of climate change on agriculture in Tamil Nadu. Increased CO₂ levels increased rice and groundnut yields, but there was no definite trend between the predicted temperature changes and crop yields in the selected districts of Tanjore and Thiruvannamalai (Ramaraja et al. 2009). This shows that future water requirements under climate change can also be identified by the above approaches.

WOFOST is a simulation model for the quantitative analysis of the growth and production of annual field crops. It also calculates attainable crop production, biomass, water use, etc. From our point of view, this model is also a useful tool for irrigation scheduling. Various environmental factors, such as temperature, humidity, and precipitation, must be considered to accurately determine the optimal irrigation schedule for crops (Math & Dharwadkar 2020). Incorporating the WOFOST model for irrigation scheduling can greatly benefit farmers by maximizing crop productivity while minimizing water usage. In an effort to increase crop WP, this model was used in conjunction with SWAP in arid and semi-arid regions of Northwest China. The results indicated that this model provides a reference for properly modifying autumn irrigation modes (flooded irrigation after harvest to maintain soil moisture and reduce salts present in fields), which will help that region achieve proper yield growth with available water and use the same SWAP-WOFOST combination in the study area mentioned above; however, the authors attempted to simulate yields and WP under sprinkler and surface irrigation between 2000 and 2010. The results showed that after zoning in the sprinkler irrigation scenario compared with surface irrigation, the annual average yields of spring wheat, spring maize, and sunflower improved by 16.9, 8.0, and 11.4%, respectively, and the annual average WP increased by 7.9, 5.0, and 14.1%, respectively (Xue et al. 2020). In addition to the WOFOST model, incorporating IoT technology and real-time weather forecasting can further improve the accuracy and effectiveness of irrigation scheduling (Goap et al. 2018).

APSIM is a well-established modeling platform that can be used for a variety of agricultural simulations, including irrigation scheduling. AQUAMAN, a brand-new web-based decision support tool, was created to help Australian peanut farmers plan irrigation. The irrigation scheduling guidelines from the FAO-56 are combined with the modeling framework of the APSIM, which models the timing and depth of future irrigation. Since being launched in 2004–2005, the tool for planning irrigation schedules on commercial peanut farms has been well received by local peanut growers and has the potential to greatly increase productivity. Based on a limited comparison with farmers' practices of matching the pan evaporation requirements during the rain-free seasons of 2006–2007 and 2008–2009, it was possible to realize enhanced water and irrigation usage efficiencies and to save up to 50% of irrigation water with AQUAMAN (Chauhan et al. 2013).

Currently, the most widely used crop models are DSSAT (Ma et al. 2020), EPIC (Qiao et al. 2018), and WOFOST (De Wit et al. 2019). Accurate simulations of crop growth status and water usage at the field scale are possible. However, these models are more complicated than the AquaCrop model (Han et al. 2020; Zhang et al. 2021). With the FAO's simplified crop model, AquaCrop can schedule irrigation events by either manually specifying the time and depth of each application or by having the model automatically develop a schedule (Raes et al. 2009; Steduto et al. 2009). The AquaCrop model's requirements can be satisfied with a small number of input parameters. It takes into account various factors, such as crop type, soil characteristics, climate conditions, and irrigation system efficiency, to provide optimal irrigation schedules (Math & Dharwadkar 2020), making it simpler, more logical, and easier to obtain. Because it works with a wide range of crops, the AquaCrop model is becoming increasingly popular and applicable worldwide. AquaCrop was developed to simulate the attainable crop biomass and harvestable yield in response to available water. In the second scenario, irrigation schedules are set either by a predetermined percentage of allowable root zone depletion or at a preset time interval and depth criterion. Geerts et al. (2010) improved irrigation frequency throughout sensitive crop growth stages based on suggestive crop development attributes at the beginning of the stage using the AquaCrop model and an extended set of historical climatic data. For farmers, a more easily accessible irrigation chart was then developed. Nevertheless, this basic irrigation chart does not provide information on irrigation amount or adjustments for specific years' atmospheric circumstances; it merely indicates the frequency of irrigation throughout the important crop growth stage. Several studies and tests on crop production prediction in the context of climate change and irrigation management optimization have made use of the AquaCrop model. Its use has been widely distributed. By utilizing the AquaCrop model, farmers can ensure that their crops receive the right amount of water at the right time, resulting in improved crop yield and WUE. In addition, the AquaCrop model allows for better water management, as it helps to avoid over-irrigation or under-irrigation, which can lead to crop stress or water waste, but the effects of severe stress on crops such as maize need further assessment and development in addition to mild stress because of faults and errors in the model (Heng et al. 2009). In general, the vegetative stage of maize is strongly affected by water stress, which leads to a reduced LAI (leaf area index) due to reduced leaf size (Cakir 2004), and water stress also affects the HI (harvest index) either positively or negatively depending upon the timing and severity of the stress (Raes et al. 2009). When properly calibrated, this model should prove to be a powerful tool in the analysis of maize WUE (Hsiao et al. 2009), and the impact of different irrigation scheduling options on yields should be examined to identify viable strategies to enhance WUE for maize (Wang & Cai 2009; Greaves & Wang 2017). In addition, the integration of other technologies, such as remote sensing and data analytics, with the AquaCrop model can provide real-time monitoring and analysis of soil moisture levels and crop cover. This model is used to indicate irrigation timing based on the readily available water (RAW) and total available water (TAW) thresholds, below which a crop begins to experience water stress. DI strategies established using RAW thresholds can remarkably improve WP (Ket et al. 2018) and can also be used to simulate the net irrigation requirements (NIRs) of a crop, and these results can be used to make decisions about irrigation management. For example, Nunes et al. (2021) used a model to investigate the optimal planting dates for cowpea based on simulations of the NIR region. Similarly, Paredes & Torres (2017) obtained the seasonal NIR for a long-term data series in AquaCrop and classified the demand for crop water based on an empirical frequency distribution. Compared with those of established models, the simplified calibration and minimal input needs of the AquaCrop model, together with its simplicity and lower input requirements compared with those of other crop models, make it perfect for simulation (Vanuytrecht et al. 2014; Kephe et al. 2021). AquaCrop, a successor of CropWat, separates ET, considers crop biomass and harvest indices for yield, and uses daily time steps for drought stress response. This approach potentially improves WP modeling but lacks batch processing capability. The results support its suitability over CropWat (Kloss et al. 2012).

• Optimized water use: These models help optimize water use by simulating crop water requirements based on factors such as soil type, weather conditions, crop type, and growth stage. By providing accurate estimates of water needs, they enable farmers to apply irrigation more efficiently, avoiding both over- and under-irrigation.

• Reduced water waste: By tailoring irrigation schedules to match crop water demand, simulation models can significantly reduce water waste. This is particularly important in regions facing water scarcity or where water resources are limited.

• Increased crop yield: Proper irrigation scheduling based on crop simulation models can lead to increased crop yields. By ensuring that crops receive the right amount of water at the right time, these models help maintain optimal growing conditions, which can result in improved productivity and quality of the harvest.

• Cost savings: Efficient irrigation scheduling can lead to cost savings for farmers by reducing water usage, energy consumption for pumping water, and labor costs associated with irrigation management. In addition, by preventing water stress or waterlogging, these models can help minimize crop losses, further contributing to cost savings.

• • Complexity: Crop growth simulation models have been successfully applied in several studies in many developed countries. Nevertheless, its wide application has been limited, primarily due to its complexity, high nonlinearity, and input requirements, which often make it difficult for researchers in developing countries to run these models.

• • The accuracy of crop model calibration and validation: Irrigation planning simulated by a model representing variations in terms of accuracy and robustness.

• • (i) The accuracy is influenced by various factors, such as the calibration unit, sample number, and model input data.

  • (ii) According to the first law of geography, using a large calibration unit may lead to errors in simulation results due to the lack of homogeneity within the same region. Therefore, using a smaller calibration unit that ensures greater similarity in the geographical environment within the same unit can improve the simulation accuracy of the crop model.

• • Specificity: Crop models are typically packaged software and require a specific file format to simulate. Although APSIM is a promising tool for optimization, more experiments are needed to obtain crop-specific parameters for model calibration despite its potential (Kloss et al. 2012) because it requires a large amount of input data, which limits its application.

• • The following criteria must be met for crop growth models: They should be checked for eligibility for simulation-based optimization models targeting WP improvement within a stochastic framework (Kloss et al. 2012). The main criteria are as follows:

• • (i) Accurate representation of plant physiology processes.

  • (ii) Realistic response to water stress.

  • (iii) The ability to account for the spatial distribution of water in the soil for modern irrigation systems.

  • (iv) Daily temporal resolution for better consideration of plant response to water stress.

  • (v) Robust crop parameters for wide range transferability and application.

• • The fact that modules from one set of models cannot be used with another is a challenging aspect of the crop models that are currently in use.

• • For the GA, the larger the population is, the greater the genetic algebra, and the more accurate the optimization results are; thus, the calculation costs become high.

Weather forecasting has significantly improved over the past few decades at most temporal and spatial scales. This achievement is the result of a combination of enhanced numerical weather prediction (NWP) forecast models that take advantage of today's greater computing power to more accurately simulate the atmosphere and more precise and high-resolution satellite and earth observational data from the AMS (American Meteorological Society) portal. Accurate forecasting plays a key role in irrigation scheduling by providing essential information on upcoming weather patterns, such as rainfall and temperature fluctuations (Gu et al. 2020). This allows farmers to make informed decisions on when and how much to irrigate their crops, taking into account the current soil moisture levels and the expected weather conditions to avoid over-irrigation or under-irrigation. By incorporating weather forecasting information into irrigation scheduling, farmers can better manage their water resources and reduce waste. This can lead to saving water and increasing net profits. Among all the forecasts used for irrigation scheduling, probabilistic forecasts exceed other benefits in decision-making (Rogers & Elliott 1989). Daily irrigation amounts are measured in real time based on probabilistic weather forecasts (Cai et al. 2011). Different probabilistic forecasts are provided by the National Oceanic and Atmospheric Administration (NOAA) (Wang & Cai 2009) and Weather Research and Forecasting (WRF) generated probabilistic forecasts can help to handle irrigation optimization problems (Wu et al. 2022). Compared with conventional irrigation scheduling based on soil water content, the optimal use of probabilistic forecasts substantially increases economic value (Wilks & Wolfe 1998). The usefulness of 3-day probability-based weather forecasts for irrigation scheduling is limited by the low uncertainty of forecasts with shorter periods of 3 days (Jamal et al. 2023); likewise, consecutive 3-day total rainfall forecasts rather than daily forecasts performed better by saving approximately 0–100 mm of water and reducing drainage (approximately 0–60 mm), and no yield loss occurred (Cao et al. 2019). Short-term forecasts are more accurate for use in irrigation scheduling (An-Vo et al. 2019) than seasonal forecasts, which provide better decisions about crop patterns and crop types (Sangha et al. 2023). A assessment was made in potential gain in annual income of marginal farmers over Karnataka (Hobli) by Nair (2021) from the implementation of WRF forecast-based irrigation scheduling of crop land. They revealed that the percentage potential gain from forecast-based irrigation scheduling for 7 days is much greater than that from 10 days, which is a 10–15% increase in yield (Nair et al. 2021). Currently, in a risk analysis framework for irrigation scheduling, ensemble short-term forecasts such as NWP forecasts, which are considered the most important weather inputs for optimized irrigation scheduling, are generated as critical inputs (Guo et al. 2023). Another rule-based approach determined by soil moisture depletion and short-term forecasting was used to trigger irrigation to avoid both stress and excess water conditions in Virginia to quantify the impact on yield and WUE, which shows that the accuracy of soil moisture storage is highly influenced by forecast accuracy (Sangha et al. 2023). Evaluation of the effectiveness of the India Meteorological Department (IMD) short-term weather forecast with different horizons (1, 3, and 5 days) in four paddy fields in south India for the two monsoon seasons 2018–2019. Compared with other scenarios, with 69% water savings and 23% greater yields, irrigation scheduling with a 5-day forecast horizon performed slightly inferior to the hypothetical perfect forecast, indicating that the IMD forecast is a valuable tool for proper irrigation scheduling (Anupoju et al. 2021). By applying only the suggested simulation-optimization modeling framework with current crop ET and soil moisture data without any forecasts, there is significant potential to enhance the practices of existing farmers. This outcome highlights the benefits of employing the well-established simulation-optimization modeling framework as opposed to a basic rule-based strategy, and another investigation showed that optimum irrigation decisions, as opposed to traditional rule-driven irrigation selections, can yield significant economic benefits.

With different forecast horizons, and 7-day forecasts integrated into crop simulation models, farmers can make more informed decisions about when and how much to irrigate their crops. This helps them optimize water usage and improve crop yield (Wang & Cai 2009). By anticipating weather conditions for the next 7 days, farmers can mitigate risks associated with both water scarcity and excessive irrigation, leading to more sustainable agricultural practices.

Forecasts with improved NWP models and enhanced observational data provide more accurate predictions. Probabilistic forecasts are more beneficial than deterministic forecasts for decision-making. Short-term forecasts are useful for crop selection and planning, while seasonal forecasts are better for crop selection. Forecast-based irrigation scheduling can lead to potential yield increases and water savings. Ensemble NWP forecasts are essential for risk analysis, and online weather forecast information is valuable for both short-term and long-term scheduling. Future research should focus on uncertainty-based frameworks for evaluating the risks associated with different irrigation options.

Computational power has risen dramatically in the last few years. Estimating agricultural productivity as a function of weather, soil, and crop management is one of the primary objectives of crop simulation models (Hoogenboom 2000). Crop simulation models are useful tools that have recently emerged to increase agricultural output (Surendran et al. 2010). The integration of forecast data into models provides a solution for various effects and risk assessments in agriculture. WRF and DSSAT models were coupled to study the spatiotemporal effects of crop-specific water stress and the yield loss that resulted in Punjab and Haryana, India, in 2009 and 2014. The results suggest that the DSSAT model, when combined with the WRF regional climate model output at 10 km, can satisfactorily capture the evolution of drought and its impact on crop yield (Rajasivaranjan et al. 2022). Conversely, projections made with shorter lead times tend to be more accurate. Uptake is similarly reduced when seasonal forecasts are communicated in probabilistic language, particularly when forecasts have longer lead times (Nyamekye et al. 2021). A study validated extended range forecast service (ERFS) data for maize yield prediction in Erode, Tamil Nadu, using the DSSAT model. The forecasted monthly rainfall correlated well (r = 0.97) with the observed data. However, the yield simulations using ERFS data deviated significantly (−29.7%) from the actual yields, while simulations with observed weather data showed a lower deviation (−14.7%). The results suggest that downscaled and accurate weather forecasts are necessary for reliable crop yield prediction using the DSSAT model. Further improvements in both the model and ERFS forecasts are required for higher resolution yield predictions and yield responses to water (Harinarayanan & Dheebakaran 2022).

Encouraging and demonstrating the benefits of integrating crop impact models with seasonal predictions will aid in the production of timely, location-specific, and useful information, which will increase farmers' chances of adopting such models (Mkuhlani et al. 2022). Benefits were realized by introducing weather forecasts, i.e., crop growth models, into the decision-making process because of subsequent decreases in both the frequency and amount of irrigation water applied (Rogers & Elliott 1989). A framework was developed based on ensemble short-term weather forecasts incorporated into the APSIM model for analyzing the risk of uncertainties arising from rainfall forecasts in irrigation scenarios, which in practice will benefit significantly from such a framework, which will more realistically represent the uncertainties associated with irrigation decision-making and allow for more efficient comparison and optimization of irrigation decisions (Guo et al. 2023). This demonstrates how crop simulation modeling may be used to evaluate the impact of various forecast horizons, create uncertainty frameworks, and optimize irrigation. To schedule irrigation, the simulation model was adjusted to benefit from the precipitation probability forecast by reducing the frequency of irrigation, which will save irrigation water (Hashemi & Decker 1969). This integration helps prevent over-irrigation, which can lead to water waste and nutrient leaching, as well as under-irrigation, which can result in crop stress and reduced yields. By incorporating weather forecasts into a crop simulation model for irrigation scheduling, farmers can effectively plan their irrigation practices in response to forecasted weather conditions. Furthermore, this integration allows farmers to anticipate potential future weather patterns and adjust their irrigation plans accordingly, thus increasing the efficiency and sustainability of their irrigation practices. In addition, by incorporating weather forecasting into a crop simulation model for irrigation scheduling, farmers can also mitigate the impacts of extreme weather events such as droughts or heavy rainfall (Stone & Meinke 2006). They can prepare for these events by adjusting their irrigation plans and implementing water-saving strategies in advance. Weather forecasts may occasionally be utilized when coupled with modern technologies such as machine learning and data assimilation (DA) to produce useful outputs for scheduling irrigation. According to study of Jamal et al. (2023), this approach may enable more advancements in targeted demand. The SWAP model simulates yields by utilizing weather forecasts from the growing seasons between 2002 and 2006. The profits gained from an irrigation schedule that the observed farmers generally followed in 2002 are compared with those from modeling analysis based on modeled soil moisture and weather forecasts (Cai et al. 2011). The modified SWAP model, which incorporates 7 days of forecast data, additionally increased the net profit and reduced the water use compared with the rule based on soil moisture alone, as shown in Table 1. It illustrates that the integration of weather forecasts and soil moisture data into irrigation scheduling models can significantly enhance water savings and profitability in agriculture. Longer forecast horizons and higher accuracy yield greater benefits. Studies across regions show consistent improvements, ranging from a few percentages to 48% water savings and up to 20% profit gains (Wang & Cai 2009; Cai et al. 2011; Gedam et al. 2023; Jamal et al. 2023).

Later, a coupled simulation-optimization framework is introduced in which the SWAP model simulates soil moisture, ET_c, and crop yield as farmers observe each day during the irrigation season, and the optimization model determines the irrigation dates and depths on the basis of the initial soil moisture and accumulated ET_c, along with the deterministic (today) and probabilistic weather forecasts within a single frame. Every day, decisions are made using a moving-window method. For example, the amount of irrigation to be performed today is determined by optimizing past conditions (today) and future conditions (future) for a maximum of one week of prediction (Wang & Cai 2009).

Even short-term forecasts provide substantial water savings over farmer observations. Coupling advanced weather and irrigation models can further amplify the potential benefits as forecast accuracy and lead times improve.

The integration of weather forecasts into crop simulation models for irrigation scheduling represents a promising approach for optimizing agricultural water management, increasing productivity, and enhancing sustainability. By addressing the challenges and leveraging advancements in computational power, weather forecasting, and interdisciplinary collaboration, this integration can contribute to achieving food security and mitigating the impacts of climate change on agricultural systems.

Advancements in irrigation scheduling through simulation technologies have revolutionized the way water is managed in agriculture. Although the soil WB approach is frequently used to allocate agricultural water, little attention has been given to the simultaneous use of this approach and the optimization of agricultural water allocation. To attempt to bridge this knowledge gap, an optimization and connected irrigation simulation model (ISM) is created. Managing irrigation in the agricultural sector is one use of optimization, a straightforward approach that uses linear programming (LP), nonlinear programming (NLP), dynamic programming, and GA to solve and analyze problems (Singh 2012; Li et al. 2014). Researchers worldwide have used these optimization techniques not only to determine the best way to allocate agricultural water but also to realistically optimize irrigation scheduling (Yakowitz 1982; Huang et al. 2012; Singh 2014; Nikoo et al. 2022). The application of optimization methods to irrigation scheduling usually aims to maximize crop yield while saving water. For instance, in Egypt, the LP model was used to calculate the financial and economic gains from irrigation (Bowen & Young 1985). Optimization approaches have been widely applied to various irrigation scenario challenges (Wardlaw & Bhaktikul 2004). Among all optimization models, GAs have recently been proposed to address the inability of conventional LP and NLP models to handle nonlinear nonconvex problems and are used for the optimization of irrigation planning and multi-reservoir systems (Moghaddasi et al. 2010). LP and simulation models are combined for planning irrigation (Kuo & Liu 2003). Considering that simulation-based optimization approaches have the benefit of properly estimating irrigation time, they were employed to address this challenge to provide optimum irrigation scheduling, while pure optimization approaches can overestimate the effects of irrigation on ET_o, yield, and field WB (Wen et al. 2017). Some studies have identified demands in response to irrigation and hydrometeorological conditions using soil moisture accounting models such as the SWAP (Jamal et al. 2023) and EPIC (Li et al. 2023) models. The application of multicriteria optimization approaches by integrating WOFOST and AquaCrop, as shown in Table 2, with the non-dominated sorting genetic algorithm (NSGA-II) has shown higher yields and reduced water consumption in sugar beets and other crops than real farmer strategies (Linker 2020; Gasanov et al. 2021; Liu & Yang 2021). Similarly, the use of AquaCrop with the NSGA-II and the exponential efficacy coefficient (EEC) increased the crop yield and WUE by 1.1–9.7% (Guo et al. 2023). However, there are certain drawbacks to LP-based irrigation planning, such as its inability to solve nonlinear issues, its ability to provide global optimal solutions for other algorithms, and its requirement for the use of NLP in irrigation management (Rohmat 2019). Despite the fact that NLP is superior to LP, Singh's (2014) analysis of several programming models led us to the conclusion that NLP did not significantly outperform LP. Finally, Wardlaw & Bhaktikul (2004) solved the irrigation scheduling problem using a GA and the time block approach. However, Haq et al. (2008) criticize this methodology. A GA was used by Nixon et al. (2001) to optimize the timing of off-farm irrigation.

Fontanet et al. (2020) combined simulation and optimization framework studies; Fontanet et al. (2022) developed frameworks to optimize irrigation parameters by considering factors such as water movement in the root zone, soil‒water‒crop productivity, and irrigation expenses, with the primary goal of maximizing the crop net margin. A study by Li et al. (2020) successfully optimized irrigation scheduling for maize in an arid oasis using a simulation-optimization model, demonstrating significant water-saving potential and maximizing crop yield. According to study of Endale & Fipps (2001), simulation-based irrigation scheduling through the IRDDESS (Irrigation District Decision Support System) model can serve as a vital tool in developing countries to maximize crop yields or minimize water use in large irrigation schemes. Wardlaw & Bhaktikul (2004) employed GA to optimize the utilization of water resources in irrigation systems operating on a rotational basis, demonstrating the use of advanced algorithms in irrigation scheduling. These studies illustrate the range of methods applied in simulation-optimization frameworks for irrigation scheduling, from complex simulation models paired with optimization algorithms to the implementation of machine learning and GA. The overarching goal remains to optimize irrigation schedules to achieve maximum efficiency in water usage and enhance crop yields, ultimately contributing to increased net profit gains for agricultural operations while conserving water resources.

The numerical simulation model can be externally linked to the optimization model using a GA-based optimization approach (Singh et al. 2016), as shown in Figure 6. It represents that the use of coupling optimization tools and crop simulation models in irrigation planning based on inputs from field trials, weather forecast data, and remote sensing offers a comprehensive and effective approach to sustainable agriculture (Acutis et al. 2009; Li et al. 2020). Recently, simulation and optimization models have been employed in combination to examine the special benefits of this method (Singh 2014).

Currently, simulation-optimization has been technologically advanced by adopting weather forecasts, field observations, DA techniques, and human‒machine interactions in a single framework called the RTIST (real-time irrigation scheduling tool) to make connections between farmers and computer-based tools to adopt recommended water applications. By applying field observation data and DA, the optimization and simulation of two crop fields were verified, demonstrating the accuracy of the current estimation and the future prediction of soil moisture and leaf area indices. The applicability of the RTIST is evaluated through a virtual irrigation test with a group of farmers for a corn field in Eastern Nebraska. Direct farmer participation in RTIST has greater productivity than traditional methods, and farmers' feedback indicates a desire to use the tool for real-world irrigation scheduling, confirming the research interest in fine-tuning the existing models and tools and discovering new models to achieve weather smart irrigation planning, which will reduce the usage of water and promote higher yields. Similarly, human‒machine interactive method-based RIS (real-time irrigation scheduling) studies are being conducted in the Hetao Irrigation District in northwestern China, which has an arid/semi-arid climate and shallow saline groundwater, and the results are the same as those obtained for eastern Nebraska, but the only difference between them was the model. Here, they used the EPIC model (Gao et al. 2017; Liu et al. 2020), and GA has been used to find the optimum irrigation applied (Wardlaw & Bhaktikul 2004; Haq & Anwar 2010). As a result, they found that the RIS model reduced irrigation inputs and increased profits. The irrigation advisory service (IRRISAT) in Campania, southern Italy, has implemented a 5-day crop ET_o forecasting system using Sentinel-2 EO imagery and a high-resolution NWP model. The system is usable and integrates advanced Web 2.0 technologies (services for delivering crop potential ET forecast maps) that have proven to be effective for monitoring canopy growth and for predicting irrigation water requirements during the mid-season stage of the crop when the canopy is fully developed. Future studies will evaluate service performance from the end-user's perspective (Vuolo et al. 2015). With the help of weather data and high-resolution data from Earth observation satellites, crop water requirements are assessed, and the service's goal is to give farmers and water managers access to this information in real time. A specialized WebGIS that is available on PCs, tablets, and smartphones provides information to users (farmers, the Water User Association, and water authorities) in almost real time. In contrast, when soil evaporation is significant relative to total ET throughout the early and development stages of a crop, the combination of Sentinel-2 imaging and a crop growth model can help to enhance estimates of irrigation water requirements (Dalla Marta et al. 2019). This same methodology has also been used in different studies to determine how well AquaCrop is integrated with remote sensing for wheat irrigation management. The Sentinel-2 crop-derived data were integrated with AquaCrop version 6.1, replacing the canopy cover. This method reduces biases caused by model simplifications and environmental conditions (Abi Saab et al. 2021). This demonstrates how the AquaCrop simulation may be better integrated with different satellite data for yield estimation and well-planned irrigation methods. Rajavel et al. (2022) focused on the CDZ (Cauvery delta zone) of Tamil Nadu. Using data from the Intergovernmental Panel on Climate Change (IPCC) AR5 scenarios representative concentration pathways (RCP) 4.5 and RCP8.5, they analyzed current and projected climate variability and trends in the region. The projections indicated potential increases in maximum temperatures of 1.1–3.6 °C, minimum temperatures of 2.8–4.5 °C, and a 15–16% increase in rainfall by the end of the century. These future climate data will help to determine future climate variability and trends, and their integration with models will reveal future changes in NIRs under climate change, as carried out by Busschaert et al. (2022). Another study revealed the effects of different irrigation schedules on maize yield, actual crop evapotranspiration (ET_a), and WUE in Heilongjiang Province, China, under future climate scenarios. The AquaCrop model was used along with future climate data from four global climate models under the RCP4.5 and RCP8.5 scenarios. Three irrigation schedules were tested: regulated deficit irrigation (RDI) at the jointing stage (W1) or filling stage (W2) of maize and full irrigation (W3). Key findings of this approach:

• Under RCP4.5, the yields under W1, W2, and W3 were 2.8, 2.9, and 2.5% lower, respectively, than those under RCP8.5.

• Under RCP8.5, the yields under W3 were 1.9 and 1.4% greater than those under W1 and W2, respectively.

The ET_a increased over time under RCP8.5, which is a suite of RCP scenarios that describe several potential future pathways, resulting in a higher ET_a than that under RCP4.5 (which is described by the IPCC as a moderate scenario in which emissions peak at approximately 2040 and then decline). The ET_a was lowest in W1. WUE was highest in W3, followed by W1 and then W2 under both scenarios. By optimizing WUE without reducing yield, they recommend regulated DI at the maize jointing stage (W1) for future irrigation scheduling. The AquaCrop model was used to evaluate different DI schedules for maize under future climate projections, and it was found that optimized WUE could be achieved while maintaining yields through RDI at the jointing growth stage (Nie et al. 2022). Another study used the AquaCrop model to assess the impact of climate change on irrigation water requirements for crop productivity. The model was run at a resolution of 0.5° over Europe, assuming a general C3-type crop. This project was forced by climate data from the ISIMIP3 (Inter-Sectoral Impact Model Intercomparison phase three) project. For cross-sectorally consistent climate impact modeling, ISIMIP creates and makes available socio-economic and climate-forcing data. It also develops relevant model output data. In addition to the input data, ISIMIP offers modeling techniques that are consistent across sectors and scales, integrating the effects of climate change into a multi-impact model framework. On this basis, the ISIMIP contributes to the understanding of climate change hazards, enabling better regional and global risk management (Busschaert et al. 2022).

The surface soil moisture of the AquaCrop model was evaluated using satellite-based estimates. The model showed good agreement with the satellite data. The model was then used to quantify future irrigation requirements under different climate scenarios. Under a high-emission scenario, the future irrigation requirement is expected to increase by 30% in the far future, with the most significant impacts occurring in central and southern Europe. However, there is a significant level of uncertainty in the projected irrigation requirements of different climate models (Busschaert et al. 2022). Shahid et al. (2018) and Wang & Cai (2009) investigated the decrease in yearly runoff in Pakistan's Soan River watershed between 1983 and 2012. A turning point was found using trend analysis in approximately 1997. According to the Budyko framework and abcd model, 68% of the runoff decrease was attributed to climate change (reduced precipitation, increased ET), and 32% was attributed to land use change (increased agricultural area, mini-dams). The two methods consistently partitioned climate and land use impacts driving basin runoff reduction. However, the Budyko framework has possible uncertainties because ET data derived from different equations and spatial distributions are not considered.

Overall, this study emphasizes the promising advancements in irrigation planning achieved by coupling crop simulation models with optimization techniques, weather forecasts, remote sensing, and human inputs within integrated modeling frameworks. However, this study has several limitations and areas for further improvement, as described below.

• Incorporating uncertainty: This combination of simulation and climate models emphasizes the great degree of uncertainty around climate projections while also highlighting the significance of mitigating climate change to maintain future irrigation levels at acceptable levels (Busschaert et al. 2022).

• The use of current weather forecasts and climate predictions for irrigation scheduling raises two concerns: the accuracy of the forecasts and the availability of this information to farmers. When combined with models, daily 7-day weather forecasts are not only more dependable than seasonal estimates but also easier for farmers to obtain.

• Forecast complexity: Crop simulation models need to be calibrated and validated with local weather and crop data to accurately represent the specific conditions of a particular region or farm. Integrating forecast data adds complexity to this process, as forecast accuracy can vary by location and time.

• Real-time scheduling of irrigation requires weather forecasts rather than historical records. One important issue is the lead time of the forecast, i.e., the forecast horizon, due to greater uncertainties and unreliability (Wang & Cai 2009).

• Crop sensitivity and reaction: Climate factors, including temperature, humidity, and precipitation, have varying effects on different crops. A thorough grasp of crop physiology and how weather affects crop growth and development is necessary to incorporate weather forecasts into crop models.

• The RTIST strategy discussed above, which combines field observations, human‒machine interactions, DA techniques, and weather forecasts, is useful but not practical for real-world applications. Farmers' input during the VIEs' evaluation phase offers insightful recommendations for enhancing the tool for practical use. As some farmers in the VIEs expected, the final goal is to deliver a smartphone app based on RTIST that is accessible to farmers during the irrigation season.

• Longer prediction periods and more unpredictable conditions demonstrate the significance of the stochastic technique, and future research could incorporate forecasts with heading times of up to 2 weeks (Zhang & Cai 2011; Hejazi et al. 2014).

We discussed advancements in irrigation planning through the integration of crop simulation models with optimization tools, weather forecasts, and other emerging technologies. This highlights the use of optimization techniques such as LP, NLP, and GAs coupled with crop models (e.g., AquaCrop, WOFOST) to optimize irrigation scheduling for maximizing yields and WUE across various crops and regions. This review then explores RIS tools that combine simulation-optimization frameworks with field DA, weather forecasts, and human‒machine interactions. Case studies demonstrate how these integrated tools can reduce irrigation inputs while increasing profits. The integration of crop models with remote sensing data (e.g., Sentinel-2) and web-based decision support systems is also discussed as a means to estimate irrigation requirements and provide accessible information to farmers. While noting the promising advancements, the discussion also acknowledges limitations and areas for further improvement, such as handling uncertainties, forecast complexity, and the need for user-friendly decision support tools. Overall, this article emphasizes the potential of integrating crop simulation models with optimization, forecasts, remote sensing, and farmer inputs within advanced modeling frameworks to transform irrigation planning and management for improved agricultural WP.

Crop simulation models such as AquaCrop, WOFOST, DSSAT, and APSIM have proven to be valuable tools for optimizing irrigation scheduling. When properly calibrated, they can help farmers determine the optimal timing and amount of irrigation for maximizing yields and WUE. Incorporating weather forecast data into these crop models can further improve irrigation planning by accounting for upcoming temperature and rainfall fluctuations. Short-term forecasts of up to 7 days tend to provide the most value for irrigation scheduling, but irrigation is a seasonal decision that requires a longer forecast period to improve further irrigation planning. Optimization techniques integrated with crop simulation models have shown a potential to significantly improve WP and farm profits compared with traditional farming practices. Multi-objective and GAs allow for the balancing of yields, water use, and economic returns. Emerging real-time irrigation scheduling tools that couple simulation-optimization with field DA, weather forecasts, and farmer interactions are demonstrating greater productivity with less water use. These methods provide promising platforms for promoting climate-smart precision irrigation. Satellite remote sensing data combined with crop models can be effective for monitoring crop growth and predicting irrigation requirements in real time. Web-based decision support systems are making this kind of information more accessible to farmers. In conclusion, ongoing innovation in coupling crop simulation models with optimization, weather forecasts, remote sensing, and farmer input shows great promise for transforming irrigation planning and management to improve agricultural WP. However, the limits of all the aforementioned technologies utilized for real-time irrigation scheduling will be further reduced by combining projected data with machine learning and IoT sensors for intelligent, real-time scheduling.

I would like to extend my utmost appreciation to Dr V.S. Manivasagam and Dr K. Bhuvaneswari for their contributions in conceptualizing the review.

S.M.N. and N.K.S. conceptualized and conceived the review. S.M.N. wrote the manuscript; Ga.D., S.P., and N.V. reviewed and edited the manuscript.

All relevant data are included in the paper.

The authors declare there is no conflict.